Wikipedia talk:No original research/Archive 48

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 45

Archive 46

Archive 47

→

Primary, secondary and tertiary – item order reversal

I have been bold and dared reverse the order in which the items were listed (tertiary, secondary, primary) mainly for 2 reasons:

the introduction listed them in the current order (primary, secondary, tertiary), so I believe that the list should respect it;
but more importantly, the old order contained forward references, which is, IMHO, bad. (E. g., a definition in a paragraph used a term defined only in some following paragraph; such as "tertiary sources" using the term "secondary sources", which was defined only in the next paragraph; by reversing the order, I believe that I have removed the problem.) --84.47.59.184 (talk) 19:16, 9 November 2009 (UTC)

I'd support this change as it makes sense to me to have them in primary/secondary/tertiary order. Not sure why it was in the reverse order before. Laurent (talk) 19:22, 9 November 2009 (UTC)

We used to list it in Primary, Secondary, Tertiary order ... and it was changed to the Tertiary, Secondary, Primary order about a year ago (with this edit). I do remember a some of discussion about it at the time... I will track it down and link. Blueboar (talk) 19:36, 9 November 2009 (UTC)

OK... found it... here is the discussion on why it was reversed in the first place. Personally, I don't think it really matters one way or the other, and I see valid reasons for doing it either way. I am just making sure that people understand that there were reasons why the order was reversed in the first place... and letting them know what those reasons were. Blueboar (talk) 19:46, 9 November 2009 (UTC)

(I am sorry about the delay, I had some work to do.) Actually, I have read the reasoning twice, but I am still a little confused about it. I really do not understand how it can possibly be easier to explain according to that proposal. First, one starts describing the (presumably) unknown term "tertiary sources", which is defined with a reference to the (still presumably unknown) term "secondary sources", which is in turn defined with a reference to the term "primary sources", which is then (finally) explained, thus requiring the reader to push 2 items onto their working stack, if the section is to be read from the top. I, on the other hand, prefer a "narrative" style, which absolves the user from the need of such a working stack; therefore, I consider it simpler, and therefore better. Honestly, I cannot understand how that approach could possibly be clearer, if it starts with describing a presumably unknown and abstract term with yet another presumably uknown and abstract term, etc., instead of starting from possibly well-known and obvious terms (photographs, original documents,...), establishing the term "primary sources", then moving on to "secondary sources", explaining that they are, among others, interpretations, analyses, and possibly syntheses of the primary sources, and then, knowing all of this, we may happily define "tertiary sources", that they are the sources that, among others, sum up secondary sources. That is how things are generally taught. Besides, the same order is in the following proposal Wikipedia:Primary_Secondary_and_Tertiary_Sources#Definitions_of_primary.2C_secondary_and_tertiary and also in Wikipedia:Reliable sources (medicine-related articles)#Definitions. I admit that I have possibly missed the original author's point for reversing it. (Maybe the author was a LISP evangelist, LOL.) But thank you very much for showing me the discussion, I had not even attempted to search for one. -- 84.47.59.184 (talk) 23:18, 9 November 2009 (UTC)

I think primary, secondary then tertiary works better. It is what people will expect. LK (talk) 10:16, 10 November 2009 (UTC)

Unclear wording under Verifiability

I see a problem with the wording of this short section which mainly serves to point to the larger article on Verifiability:

The threshold for inclusion in Wikipedia is verifiability, not truth. This policy and the verifiability policy reinforce each other by requiring that only assertions, theories, opinions, and arguments that have already been published in a reliable source may be used in Wikipedia.

"This policy and the verifiability policy" - which one is this policy? The threshold wording? Is that distinct from Verifiability itself? On first reading this sentence seems to list the same thing twice, as in "the verifiability policy and the verifiability policy reinforce each other." I think this section could use reworking, but I won't presume to do so myself as I'm unclear on what distinction is being made and I'm far from a policy Jedi. Birdbrainscan (talk) 13:31, 21 November 2009 (UTC)

This policy refers to the one you are reading... ie WP:No original research. Easy enough to fix. Blueboar (talk) 13:56, 21 November 2009 (UTC)

The reasoning?

I think this page focuses too much in telling people not to do this, not to do that, and doesn't explain right away why OR is wrong for an encyclopedia. I added two links in the "See also" section that I think are related to this question: Observer effect and Wikiality. I think it would be great if we could make people understand right in the intro (even better if in the nutshell) that OR in Wikipedia is wrong because our popularity could change the facts that we are supposed to collect neutrally. Please share your thoughts on this. --Waldir ^talk 22:11, 21 November 2009 (UTC)

Are translations primary or secondary sources?

In an RfC on Original Research an editor is claiming that published translations of primary sources are secondary sources, and so do not come in conflict with this page's restrictions on the use of primary sources. As I read this page, translations of primary sources are still primary sources. Would someone please address this issue, either here or in the RfC. --SteveMcCluskey (talk) 19:01, 25 November 2009 (UTC) minor edit 19:08, 25 November 2009 (UTC)

A pure translation of a primary source is still primary, assuming there is no added analysis or review of the material in a broader context. Crum375 (talk) 19:05, 25 November 2009 (UTC)

RfC on RS

Wikipedia_talk:Reliable_sources#RfC_on_page_move. Should Wikipedia:Reliable sources be moved to Wikipedia:Verifiability/reliable sources to become a subpage of the sourcing policy, WP:V? There would be no change in either page's status: the policy would remain policy, and RS would retain its status as a guideline. 23:41, 27 November 2009 (UTC)

Proposed Amendment to no original research

There are a hundreds of situations in which original research is needed. Some subjects are not important enough that some is published about it and there fore original research is NEEDED. If original research is not used there might be no article or a stub. Please say if you are in favor or not

Opposed "If no reliable third-party sources can be found on an article topic, Wikipedia should not have an article about it." Look at the mess the Christ Myth Theory article is in largely because the reliable source material available provides confused and conflicting information on what the definition of "Christ Myth Theory" even is. Now multiple that by hundred or worse thousand fold and you can quickly see what doing original research with no references what so ever can result in.--BruceGrubb (talk) 19:44, 10 December 2009 (UTC)

Oppose: I also refer you to WP:N. If there are no good sources for an article it is a good indication that there probably shouldn't be an article on that subject in the first instance. We arguably already have far too many articles about things that don't deserve them, without opening the floodgates for millions more on intrinsically non-notable topics. CrispMuncher (talk) 19:50, 10 December 2009 (UTC)

...that serves to advance a position.

I think there may be some clarification required on what is meant by

Wikipedia does not publish original research or original thought. This includes unpublished facts, arguments, speculation, and ideas; and any unpublished analysis or synthesis of published material that serves to advance a position.

It could be argued that most articles are an unpublished synthesis... but in most cases this is fine because the article is not advancing a position. i.e. it has NPOV.

Wikipedia editors obviously need some flexibility in how to present information. For example, can you imagine trying to write a disambiguation page if you had to find a single source which stated that these were the different meanings of the term being disambiguated?

I guess part of it is drawing conclusions which are not supported by any reliable source... then again you could still argue that a disambiguation page is presenting conclusions about what a term means.

I am using a disambiguation page as an example of a wider issue. I am not arguing that disambiguation pages should be some kind of special case.

Perhaps the section of routine calculations indicates some kind of way forward. If everyone can agree on something then it doesn't have to be published in a reliable source. This is just a thought... obviously it is a lot easier for everyone to agree on a routine calculation than on some things.

Yaris678 (talk) 16:23, 28 November 2009 (UTC)

"It could be argued that most articles are an unpublished synthesis"... no, most Wikipedia articles are (or should be) summaries of published material. The key to understanding the "that serves to advance a position" clause is that you should not cherry pick information to support your own novel argument. If someone else (ie a reliable source) has done so, you can report on that fact. Blueboar (talk) 17:16, 28 November 2009 (UTC)

And in response to Yaris's other question, if everyone (i.e. all the editors on the page) agrees that the sky is blue, we don't need to supply a source for it. We only need to supply a source if we suspect the material may be contentious or non-obvious, or there is an editor asking for the source in good faith (i.e. not a disruptive editor asking for every word on a page to be sourced to make a WP:POINT). Crum375 (talk) 17:24, 28 November 2009 (UTC)

There does seem to be a difference between the reality of many Wikipedia articles and the sentence "If no reliable, third-party sources can be found for an article topic, Wikipedia should not have an article on it" in WP:Verifiability. Examples of problem areas are articles about:

Geographical entities (such as villages, hamlets, etc.) which are too small to feature in published sources such as gazetteers, and which are typically sourced in Wikipedia (if sourced at all) by web sites strongly associated with the entity (e.g. a village's own web site).

Sports teams which are very local and haven't featured in sports writing outside of strongly associated sources. For example, articles on school-based sports teams are often sourced by information supplied in school newsletters or web sites, or at best from local newspapers, which typically reproduce supplied material with little or no checking. (The notability of some of these is another issue.)

Recently formed organizations, which, although notable in that references to them appear all over the web, have not yet been the subject of more 'scholarly' writing, so don't appear in published articles, books, etc.

When reading such articles, like Crum375, I incline to the view that they are ok if the material is not contentious, but this doesn't fit a literal interpretation of the policy. If there is a consensus that Crum375's view is ok, it would be useful to have this written down somewhere. On the other hand, as soon as you go beyond the "sky is blue" example, I get a bit worried. The overwhelming consensus of all the editors who have worked on an article may well be that something is so obviously true that it doesn't need referencing, but this may merely reflect the interests and implicit biases of that group of editors. Anything which dilutes the requirement for verifiability could be dangerous. Peter coxhead (talk) 09:48, 29 November 2009 (UTC)

What I described above is not "my view", but the current WP:V policy. It specifically says that all material on WP article pages must be verifiable. The emphasis is on the ability of every item to be verified if needed, not on providing a source for every statement we make. The current nutshell summary of WP:V states: Material challenged or likely to be challenged, and all quotations, must be attributed to a reliable, published source. So in your examples of little hamlets, if they are considered notable (which is another distinct criterion which must be met for WP inclusion), and the information is considered non-contentious, then we don't require high quality sources (although it would be good to have the best quality sources whenever possible). If any person feels that a particular bit of material in a WP article is contentious or non-obvious and unsupported by appropriate sources, they may add or request a (better) source. Please read WP:V carefully, e.g. WP:BURDEN. Crum375 (talk) 13:48, 29 November 2009 (UTC)

Crum375: actually it's WP:BURDEN that is the source of the concern I was trying to express. Discussing minute details of the wording of WP policies is, I'm sure, not useful; the overall meaning is mostly very clear. But the first and second paragraphs of WP:BURDEN seem to me to conflict in a non-trivial way. The first paragraph is what you are focussing on above, I think. But the second paragraph doesn't say "If there are challenges and no reliable, third-party sources can be found for an article topic, Wikipedia should not have an article on it." It just says, flat, that if there are no reliable third party sources there shouldn't be an article. This simply doesn't correspond to reality. Just try "Random article" for 5 minutes or so, and you'll soon see many articles on non-controversial topics with no discernible reliable third party sources, such as those in the categories I originally listed. Do you think that these should all be marked 'refimprove' or 'unreferenced'? Peter coxhead (talk) 10:05, 30 November 2009 (UTC)

Peter, you said "it just says, flat, that if there are no reliable third party sources there shouldn't be an article." But that's a misquote — that's not what the current policy says. It says the following (and I have bolded the difference): "If no reliable, third-party sources can be found for an article topic, Wikipedia should not have an article on it." The words "can be found" are key, and are synonymous in our case to "verifiable". In your random article search, you were focusing on unsourced material, and obviously there is plenty. But what the policy actually says is that sources should be "findable" for any statement which is or may be challenged. So if you find material which you believe is not obvious, and no reliable sources can possibly be found for it, then yes, it should be flagged or removed pending proper sourcing. Crum375 (talk) 16:00, 30 November 2009 (UTC)

Ok, that's a very useful explanation. The importance of "can be found" certainly wasn't obvious to me. I think it would be helpful to new editors if something along the lines of what you've written above appeared in WP:BURDEN. Peter coxhead (talk) 19:31, 30 November 2009 (UTC)

~~OK. This started as a question about part of one sentence but it appears we now have two other issues:~~

Blueboar is insisting that an article is a summary of the information in reliable sources and this is not the same as a synthesis. I think this needs to be clarified somehow. i.e. what it the difference between a synthesis and a summary? WP:SYNTH talks about drawing conclusions but I don't think this can work as the sole distinguishing feature. This is illustrated by the disambiguation example I gave above. i.e. the conclusion of a disambiuation page is that these are the meanings of a term.

There appears to be inconsistency between WP:NOR and WP:V.

~~Yaris678 (talk) 22:40, 29 November 2009 (UTC)~~

Blueboar is correct — every WP article is (or ought to be) a summary of what reliable sources have to say about a subject. This can often be a synthesis of what those sources say, and there is no problem with that as long as the synthesis does not advance a new position (or point of view, or a novel idea) which has not already been published by a reliable source. There is no problem with a disambig page as long as it neutrally presents links to articles using neutral language. It is up to the editors to decide and agree collectively of how to avoid any bias in doing so. Regarding your point about an inconsistency between V and NOR, can you be more specific? Thanks, Crum375 (talk) 22:50, 29 November 2009 (UTC)

OK. I think I misunderstood some of the previous posts. It looks like a summary and a synthesis can be the same thing, but the important point is the not advancing a new point of view. So actually my original post on this topic was actually nearer the mark than the one I just made.

I think there is a connection between what counts as a new point of view and WP:BURDEN. Perhaps this is the area that needs clarifying.

Yaris678 (talk) 23:06, 29 November 2009 (UTC)

Yes, a new point of view is effectively "material which can be challenged", and thus requires a reliable source which has already published that particular view. What specific area do you see as needing clarification, and where? Crum375 (talk) 23:12, 29 November 2009 (UTC)

Well, the start of WP:SYNTH says

Do not combine material from multiple sources to reach or imply a conclusion not explicitly stated by any of the sources.

But this isn't strictly true, as we have discussed above. Perhaps it needs to say something like

Do not combine material from multiple sources to reach or imply a conclusion that has been challenged or is likely to be challenged, unless the conclusion is explicitly stated by at least one of the sources.

Yaris678 (talk) 23:30, 29 November 2009 (UTC)

I think the way most people understand the wording "imply a conclusion not explicitly stated" is that if you do so without proper sourcing, it is likely to be challenged, and therefore it requires a source which makes that conclusion for us. Can you think of an example where there is a new conclusion, not explicitly stated by any source, which is not "likely to be challenged"? Crum375 (talk) 23:50, 29 November 2009 (UTC)

Note that the SYN section, both in its title and in the first paragraph, says that we may not synthesize published material to advance a new position. This caveat, coupled with the caution against reaching or implying conclusions not explicitly stated by sources, is equivalent to saying that it may be challenged. Crum375 (talk) 00:40, 30 November 2009 (UTC)

The sort of thing I am talking about is in the Eurofighter Typhoon article. The second paragraph states

The series production of the Eurofighter Typhoon is underway, and the aircraft is being procured under three separate contracts (referred to as "tranches"), each for aircraft with generally improved capabilities.

You probably couldn't find a source that states this. You could probably find a source somewhere that states that Eurofighter is being procured in three tranches, but not three contracts. You might find a procurement manual somewhere that says a tranche is a type of contract. The editors of this article have taken the decision that the word contract is more easily understood by the non expert but that the technical term should be provided in brackets.

I agree that the title of WP:SYNTH uses the phrase "advance a position". The problem is that although the section goes on to explain (with examples etc.) what is meant by a synthesis, it does not explain what is meant by advancing a position. Yaris678 (talk) 13:12, 30 November 2009 (UTC)

"Advancing a position" is analogous to "likely to be challenged", because it means that other editors may take issue with the material or its presentation and not be willing to accept it on face value. In your example, it would be up to editors on the page to decide whether using the word "contract" is novel, i.e. unsourced for this context, requiring substantiation by a reliable or source, or not. The rule is simple: if someone in good faith believes it needs sourcing, we supply it. So if you personally feel "contract" needs justification for use in that specific context, either find a source for it, or ask someone to do it. The bottom line is that if material is challenged, it needs specific sourcing, and "material" can be any statement or synthesis made, including one implied by juxtaposition or usage of a word in a new context. Crum375 (talk) 13:28, 30 November 2009 (UTC)

I agree entirely. The problem is this isn’t explained in WP:SYNTH and so people who aren’t as fully versed as yourself might not get this. This could lead to people not using good editing judgement where it is appropriate. It also makes things a bit easy for people to disrupt Wikipedia by being a wikilawer. Say that someone wanted to make things difficult for the editors of Eurofighter Typhoon. They can demand that people provide a source for the three contracts statement. If someone provides sources for three tranches and the meaning of tranche, the wikilawyer will say this is just synthesis. The correct response in this situation is to ask why a synthesis is invalid, given that no controversial statement has been made. The wikilawyer will then have to come up with a reason, beyond synthesis, why someone might object to the three contracts statement. This correct response might not happen if none of the editors is as clear about the application of WP:SYNTH as yourself. After all, the wikilawyer will provide a link to WP:SYNTH but not WP:BURDEN. Yaris678 (talk) 16:10, 30 November 2009 (UTC)

If someone points to WP:SYN as a reason not to use the word "contract", then you can quote "Carefully summarizing or rephrasing source material without changing its meaning is not synthesis—it is good editing." (Emphasis added) So it boils down to: is using the word "contract" changing the meaning? If so, then it needs to be sourced in this context. Otherwise, it doesn't. There is not much room for wikilawyering if all editors are acting in good faith and just want to improve the article. If someone insists on sourcing for a synthesis, they should be able to explain how that synthesis changes the meaning (or advances a position), in a way not supported by the provided sources. Crum375 (talk) 17:31, 30 November 2009 (UTC)

Hmmm... I can imagine saying that and the wikilaywer saying something like "I realise you think that tranche means contract in this context, but we have no evidence that this is the case. Using reliable sources is as the heart of Wikipedia and making assumptions like this is original research." Yaris678 (talk) 12:56, 1 December 2009 (UTC)

Someone whose intention is to disrupt by making a WP:POINT, can always argue about any policy, regardless of how well it's phrased. But if we assume that editors are here to improve articles, the question in your example should be: "do I really believe the word 'contract' changes the meaning in any significant way?" If an editor seriously feels that the word 'contract' is changing the meaning and thus introducing new material, he should be able to explain why and how. Just saying "it's a different word" is insufficient for an objection, since a valid objection has to show how using that word is "advancing a position" and is not part of the normal summarizing process which is allowed (and encouraged) for WP editing. I quote again from WP:SYN: "Carefully summarizing or rephrasing source material without changing its meaning is not synthesis—it is good editing" Crum375 (talk) 13:13, 1 December 2009 (UTC)

OK. I can see I'm not persuading anyone here so I am going to stop. Your responses have been very educational. Next time I come across a wikilaywer I will be better equipped. Thank you. Yaris678 (talk) 13:26, 1 December 2009 (UTC)

Refering back to original post

[replying to Yaris678 (talk) 16:23, 28 November 20, top of thread] I see several classes of articles that need editing or watching because of these two sentences:

Articles that probably shouldn't be here in the first place or which are only notable because they are not widely accepted, such as fringe theories. They usually die in AFD or are rewritten to make it clear the theory is fringe.
Articles that have WP:UNDUE weight given to one point of view, usually due to the editors who pay attention to those articles having different viewpoints than respected publications. Neutrality tags and outside editors who are familiar with the topic can usually steer the article to giving "due weight" to minority opinions without giving WP:UNDUE weight to them.
Highly-controversial topics, such as those that tend to wind up at ARBCOM, in which there is abundant material to support multiple points of view and it is nearly impossible to determine what the "off-wiki" "neutral" point of view is unless you've studied the topic at a "meta-level" for months or years. Political topics e.g. Palestine- or China-related ones come to mind.
Highly-controversial topics where the "proportional/due-weight" amount of coverage for a minority point of view is, say, 5%, but a faction of editors wants the article to have much more or much less than 5% of the article covering those points of view. These topics also generally suffer from the problem that accurately determining the "outside-world acceptance" of the minority point of view is difficult or impossible for your average editor. Recreational-drug-legalization, racism/KKK/Nazi-related articles, and a few other topics come to mind.

I'm not sure how changing the wording would help apply these sentences to these articles.

davidwr/(talk)/(contribs)/(e-mail) 00:43, 30 November 2009 (UTC)

Regarding contentious articles, the best way to avoid trouble is by sticking to the highest quality sources. That means using mainstream academic scholarly sources who have focused on the subject in highly regarded books, and using the most reputable mainstream news media (such as NYT). Conversely, this means staying away from lower quality sources, such as blogs, or advocacy or extremist websites. If we rely on the top level sources, there should be much less problem in finding the neutral POV. Crum375 (talk) 01:10, 30 November 2009 (UTC)

That helps a lot, until a Ph.D. or three supporting the minority opinion gets something published. This is where it takes an expert familiar with the academic reputations of those holding minority views to step into the "how much weight is appropriate" decisions. With political discussions, you will find academics in country X who support country X's point of view are probably well-regarded in country X and should have no problem getting their works published in country X, with the mirror situation in country Y. If X has academic presses in the field and country Y does not, it can give an appearance that the researchers in country Y have less respect than those in country X. Even academics writing in support of Neo-Nazi viewpoints can probably get published in some journal somewhere, if their research isn't flawed. davidwr/(talk)/(contribs)/(e-mail) 01:18, 30 November 2009 (UTC)

We have to be more discerning and demanding in contentious cases. The best sources are scholars who have devoted their career to the topic, have published well-reviewed academic books, and are internationally recognized as experts in their field. If we focus on these criteria and such experts, as well as the most reputable news media, and steer clear of advocacy websites and blogs, it will be much easier to find the neutral balance. Crum375 (talk) 02:03, 30 November 2009 (UTC)

Why secondary sources

I think I had more-or-less assumed that the reason for secondary sources is that they establish that an idea is part of the canon. E.g. Bernhard Riemann's famous 8-page paper that first put forth what later became known as the Riemann hypothesis is a primary source, and though it is regarded as brilliant and important, it does not establish that the Riemann hypothesis is widely viewed in the way it is, among present-day professionals in the field. A secondary source, on the other hand, could do that.

Is that a proper understanding of the policy?

Supposing it to be universally agreed that a certain philosopher is notable, and various secondary sources have been adduced to support that, is it proper to cite a primary source, i.e. written by that philosopher himself, to support an assertion in the article about him, that that philosopher held thus-and-so? Michael Hardy (talk) 05:19, 11 December 2009 (UTC)

My own personal answer to that question is "Yes", in that case. Primary sources are frequently useful for establishing the "existence" of a thing, as opposed to any other "quality" of the thing, such as its notability, etc. I always run back to the Hamlet example—most of the time we wan't secondary sources for all the things we'd like to establish such as literary criticism and notability of the work, but in cases like a quote from the play, writing the synopsis, and certain intrinsic facts about the work, I think citing the primary source is more appropriate than citing the "Cliff Notes" version (the secondary source). —Aladdin Sane (talk) 07:00, 11 December 2009 (UTC)

Introductory surveys of previous work in primary peer-reviewed publications as secondary sources

Does anyone dispute that often a paragraph or two in a short peer-reviewed paper that specifically describes previous work, especially in the sciences, can serve as secondary peer-reviewed source material? If you agree, do you think WP:SECONDARY should say so? Why or why not? 99.56.137.239 (talk) 04:18, 1 December 2009 (UTC)

No, I don't think it should say so... because it doesn't matter... we are allowed to discuss and cite both primary and secondary sources. Even if such paragraphs are considered primary, it is absolutely OK to discuss and cite them, as long as we don't introduce OR. I can not stress this enough... the policy does not bar the use of primary sources... in fact, it explicitly says that we may use primary sources... all this policy says is that we have to be extra cautious when doing so (because it is so easy to introduce OR when doing so). Blueboar (talk) 14:02, 1 December 2009 (UTC)

Let me address the OP's implication that we add 'literature review when included within academic papers' as an example for secondary source. Although I would agree that such reviews are effectively secondary sources, I don't think the example should be added, because all reviews of prior events or publications, regardless of where they appear, can be potentially classified as secondary sources. For example, the prior art review inside patents is a secondary source, and a newspaper report about a traffic accident at an intersection would be primary for that event if the reporter is reporting from the scene, but secondary if she reviews the prior accidents which took place at that intersection over the past decade. So the point is that a secondary source has to have some distance from the event or information being described, and may well be embedded inside a publication which is primary for another (typically more current) item. Crum375 (talk) 14:32, 1 December 2009 (UTC)

Right. Is the policy clear enough that we consider the peer-reviewed secondary sources most reliable, more than any of those other secondary sources? Is it clear enough that there are sometimes short sections of such secondary sources in primary peer-reviewed sources? I feel strongly that important information would likely lead to more rapid resolution of a variety of content disputes. The reason I feel so strongly about this is because it vastly increases the amount of secondary peer-reviewed information available to editors in situations where reviews and meta-analyses are less available, which is usually the case for new subjects and neglected topics. 99.27.202.252 (talk) 22:01, 1 December 2009 (UTC)

I believe it would be a mistake to add this level of detail to policy, because actual situations are so variable and involve many more parameters. For example, you can have a 'peer reviewed' article published in some Alternative Medicine field, which touts a procedure which most mainstream scientists would consider pseudoscience. Is that article more reliable than a non peer-reviewed article written by a world renowned mainstream scientist? And what about a non-peer reviewed article in the New York Times — is that less reliable than a peer reviewed article in some obscure scientific journal? And of course the issue of reliability is separate from "secondary" vs. "primary" sources, as both type sources can be highly reliable (or completely unreliable), but we prefer to use the former whenever possible for overview and analysis, while the latter can be used to enhance the former with details. So in summary, peer reviewed articles are good, in principle, but being peer reviewed is just one parameter of many in the bottom line equation of reliability and fitness for Wikipedia use in specific situations. Crum375 (talk) 22:42, 1 December 2009 (UTC)

The question highlights why I have long felt that we need to re-examine the PSTS section... neither the question, nor the issue it centers on has anything to do with the concept of No Original Research. The question highlights how the current language distracts editors from the centeral point of this policy (What is OR and why we should not add it to articles) and diverts them to tangential issues (In this case, comparative reliability.)

Yes, some sources are more reliable than others. Yes, there are sometimes short sections of secondary material within primary sources (and vice-versa)... the reason why the policy does not spell this out is that neither fact has anything to do with the the concept of Original Research. Neither fact matters to this policy... as long as you "stick to the sources" and do not misuse the source, this policy is complied with.

Sadly, this is a debate that we have had before... and I know there is a consensus to keep PSTS as is... yet I still think the current language harms the NOR policy by confusing the issue. Blueboar (talk) 15:01, 2 December 2009 (UTC)

I agree. I think PSTS should be moved to WP:SOURCES, with perhaps summary style description in WP:NOR, for perspective about acceptable sources. Crum375 (talk) 15:50, 2 December 2009 (UTC)

Sounds good to me, too. — Carl (CBM · talk) 16:39, 2 December 2009 (UTC)

Sounds good to me. It would make sense if it pointed out that the use of all three is allowed but that overliance on primary sources can lead to original research... and that would be a link to this article. Yaris678 (talk) 19:54, 2 December 2009 (UTC)

Yes, primary sources must be used very carefully. Misuse of primary sources can not only lead to original research, but also WP:NPOV violation, since it's easy to cherry-pick material which supports one's favorite POV. This is true for any source, but more so for a primary one because it's raw material without review by a third party with a broader perspective. Crum375 (talk) 20:30, 2 December 2009 (UTC)

Hmmm... thinking about it... isn't WP:PSTS more like a set of guidelines on applying policies, rather than a set of policies? Perhaps it should have its own page, rather than appearing on any particular policy page. Ofcourse, it would refer to all the relevent policies, with links. The page could also provide other pertinent information. For example, I once got in to a conversation with someone who refused to recognise a distinction between third-party sources and tertiary sources. Yaris678 (talk) 18:08, 6 December 2009 (UTC)

How is limiting the tertiary source to preferring secondary sources over primary (original) sources not related to excluding original research? 76.254.66.122 (talk) 17:39, 10 December 2009 (UTC)

Seriously, is there anything on this project page which would in any way help a beginning editor understand that the most common secondary sources are literature reviews? If there is, why can't I find it? Isn't that fact fairly important for most of the citations to WP:SECONDARY? Would there be any harm to saying literature surveys are secondary? Would there be any harm to saying surveys of past work in a peer-reviewed paper's introduction are secondary? 99.55.162.157 (talk) 18:32, 23 December 2009 (UTC)

My question is: what is the benefit of pointing this out. Is someone saying that a literature survey is not acceptable?

I have long agreed with Yaris's suggestion that we move PSTS to its own page. As is, it is a distraction from the issue being presented on this page. The problem is Original Research, not the type of source you are using. All three source types can be used appropriately, and all three can be misused in ways that constitute OR. We spend an inordinate amount of space on the policy page simply discribing what these these different source types are and only briefly discuss how they relate to the concept of OR (and for each we essentially say... "Don't use this type of source in ways that constitutes OR" ... Having the PSTS section here creates more confusion than clarification. Blueboar (talk) 19:44, 23 December 2009 (UTC)

The benefit of saying literature reviews are secondary sources would help beginning editors understand the meaning of the secondary sources we say we prefer. That would make Wikipedia a warmer, more welcoming place. It would make the life of every editor who has ever cited WP:SECONDARY just that much easier -- they wouldn't have to go looking for something that specifically says a peer-reviewed literature review is a secondary source. We should also link to WP:RS so that people can figure out ~~how~~ which kind of secondary sources are preferred.

The benefit of saying introductions in peer-reviewed sources can contain a paragraph or more of secondary material makes editors aware that there is vastly more secondary material available to them than merely literature reviews, and in that sense presents an alternative point of view in accordance with the NPOV policy. It also makes the life of any editor involved in a scientific content dispute just that much easier. Again, that in turn supports a more welcoming and friendly editing atmosphere. 99.34.78.67 (talk) 15:35, 24 December 2009 (UTC)

Let me rephrase... what is the benefit of saying this here... in the No original research policy? How does listing it clarify our policy on original research? Blueboar (talk) 15:42, 24 December 2009 (UTC)

Because this is where WP:SECONDARY lands and because preferring secondary to primary sources excludes original research because primary sources are more original than secondary. I think we should also say that meta-analyses are secondary, and give an example with the first seven checkboxes under "Type of Article" in a PubMed advanced search. An actual example would help people figure out what this policy means, and therefore also would make Wikipedia a friendlier, warmer, more welcoming place to edit. 99.34.78.67 (talk) 16:56, 24 December 2009 (UTC)

I'm a bit puzzled by the question opening this section, for it seems to take as a premise the idea that peer reviewed papers are primary sources. Historians and people in the humanities consider peer reviewed papers to be secondary sources, since they provide "analytic, synthetic, interpretive, explanatory, or evaluative" interpretations of the primary sources (original texts). As someone who reads in the sciences, scientific papers provide similar interpretations (either based on experimental data or working from theoretical principles). From this perspective most peer reviewed papers can be cited as a source. The only exceptions I can think of right now might be peer reviewed editions or translations of primary texts or a paper providing raw empirical data without any interpretation, which provide a version of the "insider's view". --SteveMcCluskey (talk) 17:11, 24 December 2009 (UTC)

I agree the policy should state the fact that peer-reviewed work is often considered primary in the sciences, and secondary in the humanities. I think all three of those changes and the PubMed example should be made by someone more experienced with editing this policy. 99.34.78.67 (talk) 17:38, 24 December 2009 (UTC)

I found a nicely sourced discussion of disciplinary perspectives on primary and secondary sources at Wikipedia talk:No original research/Archive 42#PSTS revised. Personally, however, I think the current article has it about right. --SteveMcCluskey (talk) 18:16, 24 December 2009 (UTC)

Renewing the call to move PSTS

I thought so... our IP editor has a common misunderstanding about what we mean by the term Original Research... He/she wants to spell out that something is a secondary source "...because preferring secondary to primary sources excludes original research because primary sources are more original than secondary." This is not what we mean by Original research. We don't care if the source says something original... what we care about is our editors saying something original. Citing a Primary source (or any other type of source) that says something original is not Original Research as we define it... what makes something OR is when Wikipedia is the first place of publication for the statement or idea. As long as what we say is solidly supported by a source... as long as we do not go beyond what that source says, it isn't OR. The source can contain original ideas.

This isn't the first time that someone has mistakenly thought that Primary = OR and Secondary = not OR... but his/her error does highlight (again) why we need to remove the PSTS section from this policy. I am not saying that we should get rid of PSTS entirely... we do need to have a policy/guideline statement on using different source types in Wikipedia... but seeing source typing confused with the concept of NOR so often tells me that it does not belong in the NOR policy. Blueboar (talk) 17:52, 24 December 2009 (UTC)

As I understand it, the rationale for the PSTS distinction is that Primary sources raise problems of interpretation that can easily lead to Original Research. For that reason alone, the section should remain a part of this policy statement, and not become part of a more easily changed essay or guideline. --SteveMcCluskey (talk) 18:17, 24 December 2009 (UTC)

I'm starting to see your point about moving PSTS from here. I haven't quite bought it yet, but I do have a concern that WP:V summarizes PSTS inadequately, on occasion, for the editor. Similar to the misunderstanding you point out, I catch editors stating "Wikipedia only allows secondary sources per WP:V", a patently false statement, and wrong source to cite, as, as far as I know, PSTS is the definitive statement on the subject.

My question is, where do you propose to move PSTS to? Does it deserve a stand-alone document?

I'm currently pretty happy with the way PSTS is written, I imagine this sort of "bell curve" with secondary sources at the top. Adding to it, in my mind, definitely seems WP:CREEPy at this point. —Aladdin Sane (talk) 18:25, 24 December 2009 (UTC)

I would agree with moving it anywhere WP:SECONDARY would explain more to beginning editors about what exactly a secondary source is, with examples. 99.34.78.67 (talk) 18:32, 24 December 2009 (UTC)

Yes, I would suggest that it become a stand alone guideline. Blueboar (talk) 19:46, 24 December 2009 (UTC)

Why not a stand alone policy? Is there a reason to demote it other than that it is too terse to be clear? 99.34.78.67 (talk) 20:06, 24 December 2009 (UTC)

No particular reason... it just struck me as being more of a guideline topic than one for a policy. If the consensus is that it should be policy, that is fine with me. Blueboar (talk) 21:44, 24 December 2009 (UTC)

It might help matters if primary source, secondary source, and tertiary source were developed into good or featured articles. I think that the articles on major concepts underlying major policies/pillars really ought to be FA! Шизомби (talk) 18:47, 24 December 2009 (UTC)

I completely agree[1]. 99.34.78.67 (talk) 19:28, 24 December 2009 (UTC)

I have created a draft for a separate page on PSTS at User:Yaris678/PSTS. Please have a look. I have based it on the current text in WP:NOR so that differences can be seen using the page history. I have made some changes, which I think address some of the issues brought up here. We can use User talk:Yaris678/PSTS to discuss whether it should be a separate page and what it should contain. Yaris678 (talk) 21:25, 26 December 2009 (UTC)

Sorry to be a late comer here. I don't watch these pages regularly. I don't think that separating the section describing what are primary, secondary, or tertiary sources from this document is a good idea. That information is tightly related to the concept(s) that this document describes, and having it here serves to further understanding for those who actually read the whole document. On the other hand, I understand and can agree with the thinking behind the criticism which is driving the idea to separate the content into a separate document. Personally, I simply come to a different solution. Rather then separating the section, why not simply rename this document to something more descriptive of the overall concept? I've always thought that the "No Original Research" title was... I'm not sure exactly, but I've never found it completely satisfying.I'll try to think of an alternative, I guess.
— V = I * R (talk to Ω) 22:06, 27 December 2009 (UTC)

Merry Holiday, all. My first thought on this is that the issue has to stay as part of the policy. The draft under discussion raises some interesting points and it may be possible to treat it in this policy in summary style, with reference to the expanded guideline, but that's something I want to consider in more depth. . . dave souza, talk 00:07, 28 December 2009 (UTC)

Just so people understand my concept here... I am not saying we shouldn't discuss the issue at all... I think we should definitely retain the short statement saying that extra caution should be taken with Primary sources... but the bulk of the section (especially the parts that do not relate to the concept of OR) would be elsewhere. Blueboar (talk) 00:35, 28 December 2009 (UTC)

"No original research + notable sources" prohibits evidence, prohibits Nullius in Verba.

This rule commits Wikipedia to argument from authority, and forbids argument from evidence. On controversial topics, argument from authority is unlikely to be true.

Until Khruschev's secret speech, all notable authority agreed the Moscow trials were fair and just, even though there was overwhelming and undeniable evidence of fraud and false confessions extracted by torture.

Similarly for the daycare/child molestation hysteria of the '80s. All notable authority agreed the trials were just, and even though everyone today knows that the child molestation trials were fraudulent, with the children's testimony coerced, it is arguably still the case all notable authority still says the trials were just, in the sense that are unwilling to retract or contradict what notable authority previously said - that what past notable authority said about those trials has not been contradicted by more modern notable authorities. Even though not one of them would today say the trials were legitimate, they are unwilling to call them fraudulent and the testimony coerced. As a result the Gerald Amirault page in Wikipedia is bizarre and severely misleading - because the no original research rule prohibits the page from presenting the obvious and compelling evidence for Gerald Amirault's innocence, since such evidence is being systematically ignored by notable authority. —Preceding unsigned comment added by James A. Donald (talk • contribs) 08:51, 19 December 2009 (UTC)

Suppose The New York Times editors, located in New York, assert that Castro is popular.

Suppose that some traveller visits Cuba, talks to Cubans in private after plying them with alcohol, reports that every Cuban questioned in this manner says he hates Castro.

Clearly the second is better evidence than the first. If the editors of the New York Times think Castro popular, how would they know?

In the global warming debate, and the various debates on communism, the practical effect of these rules has been to prohibit compelling evidence, while endorsing assertions made from on high by people who have no basis for their beliefs.

For a real life example, the Climategate files reveal that the Hadcrut surface temperature was made up, and that though thousands of scientists have endorsed the Hadcrut surface temperature graphs, none of them had any basis for doing so, since the evidence for this graph, if it ever existed, had been lost or destroyed.

We should not only ask if a source is notable, but also whether it has any basis for believing that which it believes. Obviously those who believe the Hadrut surface temperature graph is bunkum have a basis for their belief. Those who believe it is based on temperatures recorded by weather stations, while doubtless notable, have no basis for their beliefs.

The practical effect of this rule induces a bias in Wikipedia. This rule prohibits editors from challenging a notable source by asking "And how do they know?", prohibits them from supporting one source over another, on the basis that one source explains how he knows the truth, and the other source, no matter how notable, does not explain, and has no obvious way of knowing the truth.

Science education, both informal and formal, is full of “check it out for yourself”. You don’t get to say that you understand a mathematical theorem unless you’ve actually gone through the proof – probably regenerated the proof as an exercise based on the instructor’s clues. You don’t take anybody’s word for it. There is no trust in math. Zero. No need for it. And in programming, you have to write the damn program for yourself. You don’t take anybody’s word that it works. And in physics, you’re not learning physics unless you do the labs, and see for yourself. Otherwise you’re just doing not-very-rigorous mathematics.

What the Climategate files reveals is not science, but religion. Scientists replicate. Synods build consensus. The Climategate files show a synod in action. Wikipedia rules automatically favor religion over science. —Preceding unsigned comment added by James A. Donald (talk • contribs) 08:09, 19 December 2009 (UTC)

Your examples illustrate the problem with the approach you are advocating. If I go to Cuba and lot's of Cubans tell me something that contradicts the New York Times then I may choose to believe the people I have talked to over the paper I have read. However, if I came on Wikipedia and said "I know this because a load of Cubans told me" then there is no reason for anyone to believe me. I could be making the whole thing up. Whereas, we know that the New York Times is at least trying to maintain a reputation for journalistic integrity. Yaris678 (talk) 10:30, 19 December 2009 (UTC)

If they were trying to maintain a reputation for journalistic integrity, they would provide an explanation of how they know Castro is popular. Since they don't provide evidence or explanation, except the kind of evidence and explanation that Wikipedia provides, we know they are lying. And we likewise know Wikipedia is lying.

A few hours ago <http://www.washingtoncitypaper.com/blogs/citydesk/2009/12/22/washington-post-sits-on-eyewitness-account/> there was a big snowball fight in Washington. A cop pulled a gun, wandered around screaming abuse, and arrested someone. He was photographed by a Washington Post reporter, and videoed. The video of him went on you tube, the photograph of him waving a gun at snowballers, taken by a Washington Post reporter appeared on the Washington Post website - to be swiftly pulled, and replaced by a pile of lies - supposedly he did not pull the gun, and supposedly the snowballers were anarchists threatening violence. Wikipedia rules are that we should treat the pile of lies respectfully, but the video and the now pulled picture constitutes original research. James A. Donald (talk) 03:58, 24 December 2009 (UTC)

If they do a news story on Cuba it will be based on sources (in the journalistic sense) they have checked against other sources, etc. Yaris678 (talk) 10:30, 19 December 2009 (UTC)

This is rarely the case. With politically sensitive stories the facts are usually pulled out of the rectums of the editors, not derived from the observations of reporters, as in the Washington Post snowball story that happened a couple of hours ago as I write this.

Recently we observed Nature and Science issuing reports on climategate that we know to be untrue thanks to "original research" (reading the climategate files ourselves.) Wikipedia rules are to go with the doubtless peer reviewed Nature and Science version. James A. Donald (talk) 03:58, 24 December 2009 (UTC)

If they do an editorial, it will be more a matter of opinion, and that is why editorials are reliable sources only to confirm someones opinion - not to confirm a fact. To me, the more interesting question is "how do we decide what is a reliable source?", this may vary from subject to subject, but I think Wikipedia gets it right most times, through a process of discussion. Yaris678 (talk) 10:30, 19 December 2009 (UTC)

There are,however, articles in which the talk pages indicate that the choice of accepting or not accepting a source is in the hands of a group of true believers who hold the balance in any consensus. Or where the article has become a clear advocacy article in itself (vide Climate change denial) making a bit of a farce about WP processes. Collect (talk) 13:00, 19 December 2009 (UTC)

In which case we have a potential violation of the WP:NPOV policy. But that is not a good reason to change this policy. Blueboar (talk) 12:46, 22 December 2009 (UTC)

The problem is not the point of view, but the facts, thus WP:NPOV is irrelevant. If authority is evidence, but evidence is not evidence, then this is automatically apt to generate untruth on politically sensitive matters, since controversy most commonly arises when official truth differs from what is observed, as in the youtube gun-to-a-snowball-fight controversy that occurred a few hours before I wrote this. Authority says one thing, youtube video and pulled newpaper reporter photo says the other thing James A. Donald (talk) 03:58, 24 December 2009 (UTC)

"Climategate files reveal that the Hadcrut surface temperature was made up". If you have reliable sources that agree with this interpretation, wonderful! If your sources aren't reliable, then... you can't draw that conclusion on Wikipedia. See, it's simple. Fences&Windows 00:02, 23 December 2009 (UTC)

But obviously no "reliable source" will ever agree with this "interpretation", even though the evidence is undeniable, just has before Khruschev's secret speech, no "reliable source" would ever admit that the Moscow trials were fraudulent. That "reliable sources" deny the undeniable is evidence about "reliable sources", not evidence about the correct interpretation of the climategate files.James A. Donald (talk) 03:58, 24 December 2009 (UTC)

If it was as easy to shut down debate as you are claiming then the Wikipedia project would have a serious problem. It isn't and it doesn't.

Of course in anything where there is a mainstream view, an alternative view might not be seen as respectable and it might take time for people to realise that the alternative view is true (assuming it is true). Wikipedia's job is not to speed this process along. It must merely reflect it. Of course where there are notable alternative views on something these should be given but Wikipedia editors are not in the position to judge that these alternative views are the truth, unless and until reliable sources do. Yaris678 (talk) 21:48, 26 December 2009 (UTC)

Plot elements

In this edit I added the words "or plot elements" to the sentence For example, an article about a novel may cite passages from the novel to describe the plot or plot elements, but any interpretation of those passages needs a secondary source. Crum375 undid my edit with the edit summary I think that picking out certain "elements" from inside a primary source can be WP:OR, so it's not a good example.

I think that argument is flawed because writing about plot elements is the same as writing about the plot itself. Writing the plot is also a picking out the important from the trivial. Anything can be OR, but that is no reason to assume it is, and no reason to stop editing in a way that is not OR. Plot elements are just that: "elements of the plot", and so may obviously included in this sentence. Debresser (talk) 12:33, 22 December 2009 (UTC)

I would agree that writing out a plot could also be OR in some cases, if it effectively creates a specific interpretation of the primary source. In general, any use of primary sources must be done extremely carefully, since it is often used by POV-pushers to promote their own view. It is much better to let secondary sources interpret, analyze or summarize primary sources. Crum375 (talk) 12:40, 22 December 2009 (UTC)

That I without argument. I do not see that you disagree with my main point that plot elements are more or less on the same level as the plot itself. So may I conclude that you more or less agree with that? Debresser (talk) 12:45, 22 December 2009 (UTC)

Not at all. I think that describing a plot from a primary source can easily become OR, but picking out elements from the plot is even riskier, and should not be used as example of appropriate use of a primary source. Crum375 (talk) 12:57, 22 December 2009 (UTC)

I do not see any inherent reasons why that would be more risky. Debresser (talk) 14:24, 22 December 2009 (UTC)

It would be riskier if "elements" are allowed, if they are selectively picked out of the whole, as that would be more likely to promote a specific interpretation. If the entire plot is summarized in a neutral fashion, it would generally be OK. Crum375 (talk) 14:32, 22 December 2009 (UTC)

I have no problem with the addition... for one thing our job is to summarize the plot (not to simply retell it), and that will necessarily entail discussing some plot elements and not discussing others. As long as we avoid analysis (without a secondary source) we should be fine as far as this policy goes. Blueboar (talk) 12:52, 22 December 2009 (UTC)

The problem with "summarizing" is that people often misuse it to highlight specific aspects, while ignoring or downplaying others. Thus it effectively becomes their own interpretation of the primary source, which violates NOR. By specifying that "elements" of the primary source can be described, we seem to be in approval of this selective interpretation. Crum375 (talk) 12:57, 22 December 2009 (UTC)

That again is a problem in which plot and plot elements are equal. Debresser (talk) 14:26, 22 December 2009 (UTC)

We already have numerous (numerous) articles that discuss plot elements at length (plenty of embedded or standalone list-of-character articles). Yes, there's OR to be concerned about when deciding what characters to cover in any detail (eg a truly minor character is treated as major and given too much coverage), but in practical application, these are all based on the primary source with rationale consensus over time to avoid abusing the primary source. Mind you, I believe that "plot" implies "plot elements" as well (it's our summary of the narrative in some form), but I see no problem explicitly stating it in case someone thinks that a work's plot needs to be carefully made but character descriptions can go hog-wild away from NOR. --MASEM (t) 14:12, 22 December 2009 (UTC)

The problem with the example list is that it doesn't allow room for explanation, and implies that all listed activities are patently acceptable. I think we need to prune that long list considerably, focusing on perhaps a few key examples which are the least problematic. The other side of the problem is that by having a long list we are implying that anything not on it is therefore unacceptable, which is also wrong. Crum375 (talk) 14:22, 22 December 2009 (UTC)

Given that the issue of primary sources is pretty much fundamental for fiction coverage, and there has been long-standing problems with editors on how they treat fiction from an encyclopedic view, I think it is completely fair to expand it further here, but then point to WP:WAF as to how the primary source can be worked in such articles. There's language "including but not limited to:" that can be added as needed to imply that the cases listed here are not the extent this policy is to be taken, but the most common ones that are encountered in NOR problems. --MASEM (t) 14:30, 22 December 2009 (UTC)

Are we talking about the same thing, Crum375? I am talking about characters and organisations in books and films e.g. This issue is mainly relevant to fiction, as MASEM rightfully noted. Debresser (talk) 14:32, 22 December 2009 (UTC)

Yes, most plots, esp. for films and plays, are non-controversial and should be no problem to summarize neutrally. It would be more complex novels, for example, which may have different interpretations possible, and we need to discourage people from promoting their personal opinions, for example by focusing on selective items. Crum375 (talk) 14:57, 22 December 2009 (UTC)

I would disagree with this edit. The plot summary is not the same as specific plot elements. A plot summary is a straight forward description of the scenes as they occur. While describing plot elements requires picking and choosing from specific moments of the plot. These should require sourcing to those specific moments. In particular, character descriptions should require sourcing, either to the plot itself or to another reliable source. For example, List of Tokyo Mew Mew characters is a featured character list. Each statement made about the characters is sourced to a specific point in its manga series or its anime adaptation in which the exact statement occurs (i.e. one character refers to another as "smart" or expresses their own feelings) without interpretation. To verify the entire plot of a work, you'd have to read/watch the whole thing. For plot elements, one should be able to point to the specific instance supporting it, making it easier for others to validate it. If a specific point can't be found, then it is likely WP:OR. -- AnmaFinotera (talk · contribs) 14:56, 22 December 2009 (UTC)

Let me tell you why I made this edit. I was editing the article Jagernaut, trying to add sources to it, to prevent it from being deleted. Since the term Jagernaut and other terms from the fictional Saga of the Skolian Empire need to be properly introduced, I was looking for some definitions. I found very clear definitions in the first book of that series. Surely this is not original research. When I had a look if this is reflected in Wikipedia policy, I found the word "plot" specified, but "plot elements" not. Debresser (talk) 15:28, 22 December 2009 (UTC) Perhaps we could say "definitions of plot elements" or "descriptions of plot elements"? Debresser (talk) 15:30, 22 December 2009 (UTC)

That demonstrates the problem exactly. You are trying to keep a poorly sourced article from being deleted, and you try to do so by changing a fundamental policy to try to allow primary sources to be interpreted to promote one's personal opinion. I think we should focus on limiting the way that primary sources can be used, not expanding them. Crum375 (talk) 15:36, 22 December 2009 (UTC)

The first part of your post is completely besides the point and includes claims that remain to be decided upon by the community, and the second is simply mistaken. We should define what constitutes OR and what not, regardless of whether that limits or expands. Debresser (talk) 15:39, 22 December 2009 (UTC)

If you were quoting from the book or paraphrasing it, its not original research, but does need a citation. -- AnmaFinotera (talk · contribs) 15:35, 22 December 2009 (UTC)

That is what I did, and that is what I want to see reflected in this policy. Debresser (talk) 15:39, 22 December 2009 (UTC)

I would say your problem is WP:Notability, not OR. You need to establish that Jagernauts are notable, by referencing secondary sources that talk about Jagernauts. Once you do that, you can cite the work of fiction itself to establish what the "in world" definition of a Jagernaut is. Blueboar (talk) 15:41, 22 December 2009 (UTC)

True, but that is irrelevant to this discussion. Debresser (talk) 15:43, 22 December 2009 (UTC)

Not really. Adding primary cites from the novel doesn't make it notable, which seems to be the real issue. Defining it with the primary source doesn't make it notable. If you have no third-party, reliable sources discussing it, it is still unnotable. -- AnmaFinotera (talk · contribs) 15:49, 22 December 2009 (UTC)

I didn't come here to ask advice about the notability of this article. This discussion is about original research in sourcing plot elements from a primary source. Why are you getting mixed up? Debresser (talk) 21:54, 22 December 2009 (UTC)

Ah, but if you want to save the article, you should focus on establishing notability and not plot elements. Also, perhaps it's just a question of terminology, but I would say you are talking about "character discription" rather than "plot elements". Blueboar (talk) 23:36, 22 December 2009 (UTC)

It's perhaps hard to completely separate the issue of notability in discussing the article, or even in discussing original research. It's true primary sources can be used as sources for some purposes, but not for an entire article. WP:OR states "If no reliable third-party sources can be found on an article topic, Wikipedia should not have an article about it." What third-party reliable sources are there? Besides that question, if we imagined that there were no WP:N standards, an article based on a primary source alone would still seem to be a problem. OR requires that no original thought, argument, idea or opinion be advanced. The article is advancing the thought/argument/idea/opinion of the Wikipedia editor, an analytic and evaluative claim, that the subject has encyclopedic merit. Additionally, while Schellenberg would appear to be a RS regarding Primary Inversion, he would not be a RS regarding "Jagernaut." To be a RS it must be "directly related to the topic of the article, and that directly supports the information as it is presented." Schellenberg's mention of Jagernauts is more in the nature of "passing comments." Шизомби (talk) 00:58, 23 December 2009 (UTC)

I'll repeat myself: I did not come here to discuss the notability of any article. I am asking about the OR policy as it pertains to plot elements sourced to a primary source. Please stop this deviation from the subject. Debresser (talk) 01:21, 23 December 2009 (UTC)

I understand you want to save your article and the many other unsourced derivative articles in Category:Saga of the Skolian Empire; it's unfortunate it didn't get addressed earlier. However, besides noting how it isn't really possible to separate original research and notability completely, I did note issues of OR the article has at present: (1) "If no reliable third-party sources can be found on an article topic, Wikipedia should not have an article about it." (2) creating a subject and advancing a personal opinion regarding the merit of the subject is OR and (3) extracting passing comments from a source that do not directly support the subject is OR. That doesn't strike me as "deviation." The problem here appears it could be "perpetuat[ing] disputes by sticking to an allegation or viewpoint long after the consensus of the community has rejected it, repeating it almost without end, and refusing to acknowledge others' input or their own error" WP:IDIDNTHEARTHAT. The OR policy is that writing an article using primary sources alone is a problem. I tried some internet searches and some university databases, and came up with a couple newspaper articles, and some capsule reviews in Library Journal which don't mention Jagernauts even in passing (nor, apparently, does Asaro's own website!). Maybe you could ask a reference librarian for help in finding reliable sources. You'd think with so many books there'd be many instead of next to none; it's not hard to find them for Dune or even (gag) Mission Earth. If you want to propose changing the policy, here's a good place to discuss it, or the WP:VPP. At least a couple here are still not sure what all you intend to cover by "plot elements"? Шизомби (talk) 04:43, 23 December 2009 (UTC)

Wikipedia:Manual_of_Style_(writing_about_fiction)#Plot_summaries and Wikipedia:WikiProject_Films/Style_guidelines#Plot have some things to say regarding plots. "Plot elements" could refer to many different things in relation to the Plot_(narrative), not sure if something specific was in mind or if it was intended to be general. It could be aspects of dramatic structure, themes, etc. In summarizing the plot, it's likely one would end up touching on elements consciously or not. Unless one of them is controversial or disputed IGF by another editor, citing would not seem especially necessary within the plot summary. A plot summary is to my mind not that different than a WP:Lede. However, if following the plot summary there is a discussion of the themes or whatever, then one would want citations to appear there. Both the lede and plot summary of e.g. North by Northwest could be said to refer to "themes," such as the innocent man wrongly accused, but they don't call attention to them as themes, they just report. When the coverage is more substantive in North by Northwest#Themes and motifs, there are citations. It's not a great article, but that structure seems reasonable to me. Шизомби (talk) 16:21, 22 December 2009 (UTC)

"Rely mainly"

It's not a big deal, but I'd like to remove the word "mainly" that Aladdin Sane restored here. The issue is the sentence, "Wikipedia articles should rely [mainly] on published reliable secondary sources and, to a lesser extent, on tertiary sources."

Articles should always rely on secondary sources. That doesn't mean they can't use primary sources, of course, as the policy makes clear—and depending on context, some articles may use primary sources heavily—but they should never rely on them, in the sense of being based on them, or of using them to show notability. The structure and foundation of articles should always consist of secondary sources, even where primary sources supply a lot of the detail.

To add the word "mainly" to the sentence risks giving the impression that articles may sometimes rely on primary sources. SlimVirgin ^{TALK contribs} 01:29, 31 December 2009 (UTC)

The example of Par Ohmsford came up in the next thread down: an article about a fictional character that will have to be written from primary sources, because its subject is obscure enough that there will be no secondary coverage in literature reviews. — Carl (CBM · talk) 02:51, 31 December 2009 (UTC)

Perhaps we should remove "rely" entirely... what about "Articles should always be founded upon on secondary sources."? Blueboar (talk) 02:59, 31 December 2009 (UTC)

As for the Par Ohmsford example... if this fictional character has never been discussed in secondary sources... I seriously have to question whether the character notable enough to have an article on its own? The material should probably be merged into the article on the work of fiction where he appears. But that is an issue for WP:NOTE and AFD, and does not involve a WP:NOR issue. Blueboar (talk) 03:04, 31 December 2009 (UTC)

If the change is made, a reader would first come across "Wikipedia articles should rely on published reliable secondary sources and, to a lesser extent, on tertiary sources." This could, at first, be interpreted to mean that no part of an article, not even a single sentence, can rely on primary sources. After several paragraphs, the reader will eventually discover this interpretation is not what is intended, but in the mean time, the misinterpretation will have colored how the reader interprets the policy, so the reader will have to start from the top and re-read everything. We should not intentionally structure documents so readers have to read them several times to understand them. --Jc3s5h (talk) 03:01, 31 December 2009 (UTC)

What about "should be organized around," instead of "should rely on," or "should be founded on" (Blueboar's suggestion), or "based on"? SlimVirgin ^{TALK contribs} 04:41, 31 December 2009 (UTC)

That might be a way to go. The present "rely on reliable sources" is awful English anyway. --JN 466 04:53, 31 December 2009 (UTC)

Re Blueboar: the conflict between the policies is exactly the issue. There's no requirement that AFD decisions have to follow the literal wording of WP:N, and there are plenty of articles that would not be deleted at AFD even without any secondary sources. Moreover, length considerations sometimes lead to article splits. So it would be misleading for WP:NOR to claim that every article has to use secondary sources. This was actually the main point of confusion for the Par Ohmsford thread, when an editor didn't realize that an article about a fictional character will probably be written entirely from primary sources, by necessity. — Carl (CBM · talk) 03:28, 31 December 2009 (UTC)

Actually, this discussion reinforces why I am calling for the PSTS section to be moved to its own guideline or policy. While the issues being discussed here do relate to when it is appropriate to use different types of sources, they have no connection to the topic of this policy... they have no connection to the concept of no original research. If we move PSTS to its own page we can deal with this issue properly. Blueboar (talk) 03:56, 31 December 2009 (UTC)

If a topic has not "received significant coverage in reliable sources that are independent of the subject" we should not have an article on it per WP:N. Okay, so AFD sometimes decides otherwise. But this does not mean that we rewrite WP:N to say that "even topics no one has ever written about may deserve an article in Wikipedia". Same here. We describe best practice. --JN 466 04:41, 31 December 2009 (UTC)

(unindent) There are two reasons not to be too insistent on secondary sources as the basis of an article. The first reason is stated right at the top of the main page: "Welcome to Wikipedia, the free encyclopedia..." If a Wikipedia editor notices that there is no article on a worthwhile topic, and the only coverage the editor can get access to is an expensive tertiary source, the editor should go ahead and write the article. Someone else can improve it later.

Second, the definitions of primary and secondary source are vague and vary from discipline, so independent sources that take a sufficiently broad view of the topic are a satisfactory basis for an article, even if the sources are classified as "primary" in some disciplines. --Jc3s5h (talk) 04:01, 31 December 2009 (UTC)

They're not vague at all, but really quite clear. If you see a car accident and write a witness report, it's a primary source. If I come along as a journalist and ask you what you saw, my story is a secondary source. Almost all the sources we deal with on WP fall into those clear categories. SlimVirgin ^{TALK contribs} 04:43, 31 December 2009 (UTC)

By that reasoning, a journal article that reports results of a survey of people which asks people if they lived near power lines, and if anyone in their household experienced various diseases, is a secondary source; the survey forms are the primary sources. I don't think all fields would regard such articles as secondary sources. --Jc3s5h (talk) 05:16, 31 December 2009 (UTC)

How about a wording like this: "The bulk of any Wikipedia article should rely on published reliable secondary sources and, to a lesser extent, on tertiary sources."
It should be clear that any article should mainly rely on secondary and tertiary sources. Tertiary sources may be useful to help establish due weight, primary sources can be cited additionally to support points already made in secondary sources.
The PSTS discussion belongs in WP:NOR because the very act of selection from a body of primary sources, when performed by a Wikipedian, constitutes an act of original research. For example, if I write an article about Goethe and describe only the most obscure works Goethe has written, works that are rarely given prominence in secondary sources, then I have engaged in original research, even if I only make descriptive statements about these primary sources: because my selection reflects a particular analysis of Goethe's overall oeuvre – in this case one that is at variance with the secondary literature. --JN 466 04:35, 31 December 2009 (UTC)

Yes, exactly. SlimVirgin ^{TALK contribs} 04:44, 31 December 2009 (UTC)

Just to add my opinion: I feel some qualifier like "mainly" should be in here. If that would be "the bulk of", as proposed above" or the old "mainly" is not important. Debresser (talk) 06:28, 31 December 2009 (UTC)

Take a look at User:Yaris678/PSTS. I think it deals with two issues raised here quite effectively:

The initial sentences prevent the misunderstanding of "mainly use".
The section on User:Yaris678/PSTS#Complex source categorisation deals well with the different use of "primary" in scientific circles.

Yaris678 (talk) 11:05, 31 December 2009 (UTC)

"...the very act of selection from a body of primary sources, when performed by a Wikipedian, constitutes an act of original research." It can constitute an act of OR... but so can the selection from a body of secondary sources. The act of OR is in a Wikipedian purposefully selecting and excluding sources to support what he/she has written, not in the type of sources that she/he has selected or excluded. As I said, it is possible to write an aricle that has no OR at all... and use nothing but primary sources. Not easy... but possible. Blueboar (talk) 14:30, 31 December 2009 (UTC)

I think the issues of "selective use" OR and PSTS are closely related. In general, the closer the source is to the "raw data", the easier it is to misinterpret it (or interpret it in a "creative" way), intentionally or unintentionally. Primary sources are the easiest to abuse, since they are so close to the raw data that it can be trivial for an editor to pull out some factoid (or ignore another) to advance a position, but as Blueboar says, the danger of selective use OR remains for all sources. Crum375 (talk) 15:04, 31 December 2009 (UTC)

I'm throwing out a bit of caution here: as NOR is policy, emphasizing any requirement for secondary sources (even with the word "should") is going to seen as acknowledging WP:N at the policy level, which most here will recognize that that will rekindle inclusionists/deletionists debates (WP:N is likely only ever going to be as good as a guideline). I understand the intent, and I think most would recognize the difference between a "good" article that may have 2-3 deep secondary sources scattered among 30 primary ones and a "poor" article that simply grabs a couple tenacious secondary references to appear to avoid original thought.

I don't know if the reverse advise would be the better option: "Wikipedia articles should not be based only or mainly on primary sources, but instead use primary sources along with secondary and tertiary sources to avoid the appearance of 'original research'". --MASEM (t) 14:56, 31 December 2009 (UTC)

The reliance on secondary sources is not only for notability reasons. Primary sources are close to the "raw data", and as such are all too easy to misinterpret, intentionally or unintentionally, per above messages, e.g. by selective use of some components to advance a position. Secondary sources, with their greater "distance" from the raw data in primary sources, help us gain perspective about that data, based on a reliable published source. Crum375 (talk) 15:16, 31 December 2009 (UTC)

I removed the sole focus on notability in the first sentence. Other ideas welcome. Crum375 (talk) 15:40, 31 December 2009 (UTC)

Agree, reads fine by me. --JN 466 00:21, 1 January 2010 (UTC)

I feel strongly that at some point in PSTS it should allow for the idea of an "uncontentious uncited fact" in an article. I'm sorry I've not read the above discussion, nor the edits to the document since (WP:DGAF and I have a headache). My only point is that every other guideline and policy I've read allows for this possibility of an "uncontentious fact" at some point. For example WP:Citing sources states, "Sources should be cited when adding material that is challenged or likely to be challenged, when quoting someone, when adding material to the biography of a living person, and when uploading an image." There's an allowance there, as worded, that some statements don't need to be cited to a source, specifically because they are uncontentious fact. Examples of what happens if PSTS requires every statement to be cited to a source:

Hamlet: "The Tragedy of Hamlet, Prince of Denmark, or more simply Hamlet, is a tragedy by William Shakespeare, believed to have been written between 1599 and 1601." The statement is uncited and needs to go (this article is an FA).
Banana: "Banana is the common name for herbaceous plants of the genus Musa and for the fruit they produce." and "Banana plants are of the family Musaceae." The statements are uncited and need to go.

My point is, when policy documents become so straitjacketed that common sense is not allowed anymore, the document becomes non-useful to editors, WP:IAR starts looking nice to cite instead of PSTS. Not allowing for the possibility of "uncontentious uncited factual statements" in articles reduces the usefulness of the PSTS document, not increases it. I note it does make it more useful to trolls who like to go around slapping {{fact}} tags on uncontentious statements to cause contention for the fun of it, because now (without "mainly" or some other allowance in the document) every uncited statement gets to be decorated with a useless {{fact}} tag and policy can be cited for doing so (I suppose I should WP:BEANS myself at this point).

Not only that, without "mainly" or some other allowance for uncontentious facts, the document no longer reflects consensus as practiced in the encyclopedia (see the above two examples for evidence). There is a palpable feel for what consensus is when reading the above two articles: Some statements need a primary source cite, some a secondary, some a tertiary, and some none at all. —Aladdin Sane (talk) 15:46, 1 January 2010 (UTC)

Again, we seem to be confused between "sourcing an article" and "sourcing a statement within an article"... an article needs to be based upon secondary sources... a statement within an article can be cited to any type of source (or as Aladin correctly states, in some cases can be left unsourced). Blueboar (talk) 15:53, 1 January 2010 (UTC)

I think you've drawn the correct distinction, and I hope it all worked out OK. —Aladdin Sane (talk) 19:26, 3 January 2010 (UTC)

Are "plot summaries" made by WP editors Original Research?

There is a discussion going on about whether literary plot summaries need sources over at the Wikipedia:Content noticeboard (initiated 30 Dec 2009). Here is the link to the page: Wikipedia:Content_noticeboard#References_for_plot_summaries.3F.

Please consider weighing in if you have a view on the appropriate Wikipedia policy for this class of WP articles. N2e (talk) 01:37, 31 December 2009 (UTC)

Fairly well answered there. Blueboar (talk) 02:41, 31 December 2009 (UTC)

Thanks for your involvement Blueboar. I see the useful discussion there, but don't see that it really clarified anything: there is ambiguity and less-than-full consensus on plot summaries. Furthermore, it gets even fuzzier when one considers, as some of the commenters pointed out, that we are not talking here about an unsourced plot summary, but rather a character description of a literary character who appears in multiple novels of a series, all with no references whatsoever in the WP article.

Is there any way we could make it more clear and explicit in WP policy (e.g., WP:V and/or WP:OR)? For example, if I am wrong and there is a consensus that plot summaries need not be sourced, as some assert, could we not add to WP:OR and WP:V some exception (or set of exceptions):

Exception: WP articles that consist entirely of plot summaries need not be sourced.
Exception: WP articles that consist of character synopses of a character from a book or literary series need not follow the guideline of no original research

In other words, in my view, the pointing to a bunch of previous discussions doesn't cut it. We ought to boil down whatever the de facto policy is into de jure policy within WP:V and WP:OR. This would help any editor, new or experienced, better deal with how WP is evolving without stumbling into an undocumented area like literary plot summaries and failing to know that there is a special standard, and stumble into it only after patiently asking for citations and then finally deleting unsourced material. N2e (talk) 13:35, 3 January 2010 (UTC)

It seems to me that Wiki welcomes and indeed requires us to summarize the contents of a nonfiction book or article; the same should apply to a piece of fiction--that is, the editor should identify the main characters and summarize the plot. Rjensen (talk) 14:38, 3 January 2010 (UTC)

I would also word an exception differently. It is not that plot summaries don't need to be sourced but that basic plot info can be sourced back to the work of fiction itself and that people who have access to the work should be able to confirm it for themselves without much trouble.--76.66.189.121 (talk) 03:58, 4 January 2010 (UTC)

Right (as in agree). It's "nice", I think, to cite a plot summary back to the primary source, but not required I feel, due to the many eyeballs theory: lots of people have gone over the article "Hamlet", and I'm pretty sure when reading it that it is fairly accurate. Good enough for the article's purposes. But here's the thing you bring up: When we kindly cite our source for a plot summary, should we cite the primary source or the secondary source? I ask the question rhetorically because I seem to catch many editors asserting "Wikipedia only allows secondary sources". The primary source for the plot summary for the play Hamlet is the play Hamlet. The secondary source for that is the CliffsNotes version. Which would we prefer the plot summary be written from? —Aladdin Sane (talk) 09:22, 4 January 2010 (UTC)

The logic of the "no original research" rule is that it refers to private research that is not visible to others (because it's not published) and cannot be replicated in trivial fashion. That is, it is hard to verify someone's unpublished OR. A summary of a public movie, book, TV show, play etc. is easily replicated in trivial non-controversial fashion: read or watch it yourself, and no one needs special lab equipment or access to unpublished data. The OR rule should apply to "unpublished" texts or un-released movies that people do not have easy access to. Rjensen (talk) 09:48, 4 January 2010 (UTC)

Rjensen, I don’t think that is a full picture on WP:NOR. For example, WP:SYNTH is part of WP:NOR. You are not allowed to combine information from sources to reach or imply a conclusion that is not reached in any reliable source - even though the combining of sources is done entirely in public on Wikipedia. I was concerned about WP:SYNTH at one time, reasoning that the whole of Wikipedia is a synthesis, and raised it on this talk page. I was persuaded that the sort of synthesis that is fine comes under the heading "good editing" and is not really called synthesis in Wikipedia. Original synthesis is where the distinction between the conclusion you are drawing and any conclusions in the sources is non-trivial. Similarly, plot summaries come under the heading "good editing". Yaris678 (talk) 13:51, 4 January 2010 (UTC)

Plots can be referenced to either the primary source (so long as what you write is purely descriptive), or to a reliable secondary source that has discussed the plot (probably not Cliff Notes though). SlimVirgin ^{TALK contribs} 10:35, 4 January 2010 (UTC)

To expand on SV's comment... what is cited depends on what is said in the article. A simple summary of the action that takes place in a book can be cited to the book (or movie, play, etc). It is not OR to summarize the action of a work, as such a summary is easily verifiable by anyone who reads the book. This is considered so obvious that we usually assume the citation (ie we don't bother to actually cite it.) However, if you go beyond a simple summary of the action, especially if you include any sort of analitical, conclusionary, or interpretive statement into your summary... then you need a secondary source. Blueboar (talk) 14:27, 4 January 2010 (UTC)

I agree with Yaris678--I was not trying to coved the whole OR policy, just what is relevant in this discussion. Professor are annoyed by Cliff Notes/Sparks/Monarch because they give too good a summary of the book and the student can thereby escape reading it. That makes Cliff Notes/Sparks/Monarch rather good secondary sources for Wiki purposes, since they give professional summaries of the book.Rjensen (talk) 14:55, 4 January 2010 (UTC)

Are the Cliff Notes articles written by named people so we can check they're experts? I haven't looked at one for a long time, so I can't remember, but I do recall that, whenever I've read a Cliff Notes entry about something I know about, it hasn't been very accurate or comprehensive, so I wouldn't recommend them. SlimVirgin ^{TALK contribs} 15:00, 4 January 2010 (UTC)

I just checked a Cliff Notes' entry about a book I'm familiar with, and it was anonymous, not to mention inaccurate and badly written. If that's a typical standard, I'd say they definitely shouldn't be used as sources. SlimVirgin ^{TALK contribs} 15:06, 4 January 2010 (UTC)

Let's not get side tracked... the issue isn't whether Cliff Notes are reliable... the issue is whether the book itself is a reliable source for a summary of its plot. It is. In fact, I would argue that the book itself is the most reliable source for a basic plot summary. I think we have a consensus that Plot summaries that stick to simply summarizing the action of a work are not OR. Blueboar (talk) 15:22, 4 January 2010 (UTC)

I'd have to disagree. It's better to stick to reliable secondary sources for everything, including plot summaries. For notable works of fiction, there will, in fact, be many such reliable sources. This would include book reviews at a minimum. Jayjg ^(talk) 18:34, 4 January 2010 (UTC)

I mostly agree with Jayjg in principle, but not in practice. A reliable secondary source, such as a professional book review published in a major newspaper, is usually a better summary than what an anonymous Wikipedia editor would produce. However, it would be harder to use it without copyright violations than an original summary by a Wikipedia editor.

Also, reliable secondary sources don't always do a good job. A somewhat related example is the New York Times publication on their web site of the audio from the FDNY tapes for 9/11, together with a Times transcription of the same. It was evident that the transcriber was completely unfamiliar with fire department terminology, and he/she just made up names for units that were difficult to hear. --Jc3s5h (talk) 19:07, 4 January 2010 (UTC)

Most notable books will have had multiple book reviews, so the issue of copyright violation shouldn't be so difficult to deal with. Also, a transcription of a audio source is essentially a primary source. Jayjg ^(talk) 21:05, 4 January 2010 (UTC)

Agree with Blueboar. —Aladdin Sane (talk) 20:27, 4 January 2010 (UTC)

Where the plot is a complex one, editors need to check with secondary sources (good ones) to make sure they're highlighting the important plot elements, and not missing out something vital, or highlighting something irrelevant that they've misunderstood. Ideally, both primary and secondary sources will be consulted. SlimVirgin ^{TALK contribs} 21:00, 4 January 2010 (UTC)

Yes, that's the key reason why secondary sources are preferred; to ensure that the important plot elements are highlighted, undue weight isn't given to minor elements, etc. Jayjg ^(talk) 21:05, 4 January 2010 (UTC)

Best practice is to check for summaries in secondary sources first, and where good secondary sources are available, to use them and cite them. Especially so if it is a well-known work by a major author that has a substantial amount of scholarly secondary literature devoted to it. Bypassing available secondary sources and making one's own summary from having read the book/watched the film/listened to the opera etc. would be OR. --JN 466 21:15, 4 January 2010 (UTC)

It's not OR necessarily. Saying "John is the strongest character in the novel." does need to be cited, but if the work itself says that he is indeed the strongest character then that is good enough. It's only OR when the work doesn't say it, ie one assumes it because he becomes the champion therefore he must be the strongest. The latter is synthesis. The former is part of summarizing the primary work, something we do everywhere in Wikipedia, not just plot.陣内 _Jinnai 21:29, 4 January 2010 (UTC)

Signpost Policy Report

Responses by Yaris678
On the 6th of December I suggested that WP:PSTS should be a separate guideline, rather than part of WP:NOR. This suggestion got some positive response so I have created a draft guideline at User:Yaris678/PSTS. This is discussed further at User talk:Yaris678/PSTS.

Responses by Bob K31416

I don't think that the editors who determine what goes into WP:NOR, sufficiently consider the needs of editors who are not familiar with it. Resistance to changes that would help correct this is exemplified by a recent discussion here. This was the sentence in question, which is currently in WP:SYNTH, that was proposed to be put in the lead where its general applicability and existence would be more evident.

Carefully summarizing or rephrasing source material without changing its meaning or implication is not a violation of this policy, it is good editing.

--Bob K31416 (talk) 17:58, 7 January 2010 (UTC)

Responses by next user

Responses by next user

Responses by next user

A summary of your comments on our No original research policy will be featured in one of the upcoming Policy Reports in the Signpost. If it helps, monthly changes to this page are available at WP:Update/1/Content policy changes, July 2009 to December 2009. Any question you want to tackle would be fine, including: Can you summarize the page? How has the page changed over the last few months? Did the changes involve some compromising or negotiation? Would the page work better if it were shorter (or longer)? Is this page "enforced" in some useful and consistent way? Was this page shaped more by people's reactions to day-to-day issues or by exceptional cases, for instance at ArbCom? Does the policy document reality, or present ideal goals for content, or something in between? Does this page contradict or overlap other policies or guidelines?

A paradox of modern democracies is that voters generally have a low opinion of national politicians, but tend to trust and re-elect their own representatives. I think the same thing goes on with policy pages ... some people^[who?] distrust policy pages in general but like the pages that they keep up with. The weekly Policy Report aims to let people look at policy pages through the eyes of the people who work on the page. - Dank (push to talk) 15:26, 3 January 2010 (UTC)

"Carefully summarizing..." paragraph- better in lead or synth?

SlimVirgin, It seems that the ideas in the paragraph are more general than just synth and would be useful in the lead, if it is properly phrased to be more general. In that regard, perhaps you could expand on your edit summary remark, "moving this back to where it was, instead of in the lead, where it doesn't make sense out of context". --Bob K31416 (talk) 16:39, 4 January 2010 (UTC)

It makes no sense to say in the lead, "Carefully summarizing or rephrasing source material without changing its meaning or implication is not a violation of this policy, it is good editing," because no one suggested it might be. That para was written for the SYN section, where it makes sense as serving to limit the application of SYN. SlimVirgin ^{TALK contribs} 16:47, 4 January 2010 (UTC)

Re "That para was written for the SYN section, where it makes sense as serving to limit the application of SYN." - I think the same limit is useful to mention regarding the application of WP:NOR in general, and that is well served by the same sentence, in my opinion. Perhaps you would agree to having just the one sentence, "Carefully summarizing..." in the lead? --Bob K31416 (talk) 19:17, 4 January 2010 (UTC)

I don't mind it if it makes sense and actually adds meaning, but that particularly sentence really doesn't make sense in the lead. There is no suggestion in the lead that carefully summarizing source material might be OR, so to jump in and say that it isn't looks odd. It's like adding, "editing WP with a coffee and a cigarette isn't OR; reading a book so you understand the topic isn't OR; taking your dog for a walk instead of editing also isn't OR." But if it's rewritten in some way, it might be okay. SlimVirgin ^{TALK contribs} 19:24, 4 January 2010 (UTC)

actually, I think the first of those (editing with coffee and cigarettes) might be OR, but that may be because I'm an ex-smoker...

I don't think a restrictive comment is appropriate in the lead, but we could add something affirmative, e.g.:

Citing sources and avoiding original research are inextricably linked. To demonstrate that you are not presenting original research, you must cite reliable sources that are directly related to the topic of the article, and that directly support the information as it is presented. You may paraphrase or summarize those sources if you do so carefully, without misrepresenting their position.

we might want to reconstruct it so it's not second person text, though. --Ludwigs2 19:45, 4 January 2010 (UTC)

That would be better, but still a little strange. It's not that we may summarize sources. Rather, we must. I can't really see the point of expanding the lead any further. SlimVirgin ^{TALK contribs} 21:43, 4 January 2010 (UTC)

That would be:

Citing sources and avoiding original research are inextricably linked. To demonstrate that you are not presenting original research, you must cite reliable sources that are directly related to the topic of the article, and that directly support the information as it is presented. Sources should be paraphrased or summarized carefully, without misrepresenting their positions.

which sounds better anyway. but I can see your point about it not really being necessary. --Ludwigs2 21:49, 4 January 2010 (UTC)

I thought it would be useful in the lead so that editors who don't understand WP:NOR very well, don't misinterpret it to mean, for example, that one can't put in summaries without the summaries appearing in a reliable source. Also, they may be inhibited about rephrasing that is not really close to the wording in the source. Or an editor who doesn't understand WP:NOR well enough may challenge an edit on these grounds that the summary doesn't appear in a reliable source or the rephrasing is not close enough to the wording of the source, and the challenged editor who also doesn't understand WP:NOR that well may not be able to defend the edit. To use one of your phrases, but applied to NOR in general, "it makes sense as serving to limit the application of" NOR. --Bob K31416 (talk) 21:44, 5 January 2010 (UTC)

"Also, they may be inhibited about rephrasing that is not really close to the wording in the source."

I do see that as a real problem. There are editors who interpret SYN absurdly strictly, so that people are often forced to copy what the source says word for word because they daren't paraphrase. I would like to see a sentence or two about not being too rigid about it, provided we can do it without undermining what's important about SYN. SlimVirgin ^{TALK contribs} 22:16, 5 January 2010 (UTC)

In the interests of transparency, I currently have an FAC on the boil where one of the reviewers said I had violated SYN, based on (as I see it) too strict an interpretation. To avoid any apparent COI, I'd therefore have to wait until the FAC is over before I could involve myself in writing anything about this for the policy. But that example is just the latest in a long line of problematic SYN interpretations I've seen, so I'm not responding here because of the FAC. SlimVirgin ^{TALK contribs} 22:21, 5 January 2010 (UTC)

But the FAC hasn't stopped you from editing synth here, for example. --Bob K31416 (talk) 01:57, 6 January 2010 (UTC)

Sniping aside, it would be useful to spell out what shouldn't be undermined in SYN. If you ask me (which no one has, I know

) SYN is a very difficult section; a bit like trying to legislate common sense. I think it's clear what it's trying to stop - the creation of a novel idea from two or more well-established ideas through sheer proximity (what I'd normally call juxtapositional induction) - but that kind of thing is a very normal part of human thought, and happens in more or less innocuous ways on every page of wikipedia. The trick is figuring out how to squelch the noxious forms of it without getting in the way of the innocuous ones (and thus making editing on wikipedia essentially impossible). So what exactly do we want to stop with this passage? --Ludwigs2 02:59, 6 January 2010 (UTC)

(to Bob) Wow, I fixed the writing. Thanks for the AGF! SlimVirgin ^{TALK contribs} 12:46, 6 January 2010 (UTC)

The answer is simple: we are allowed synthesis, as long as it does not "advance a position". The trick is for editors to assess the synthesis and determine (by consensus) whether it tends to favor one side in a dispute, or creates a negative implication about a living person (in a WP:BLP situation). That is the key: if there is no NPOV or BLP issue, there is (generally) no synth restriction. Crum375 (talk) 03:09, 6 January 2010 (UTC)

Good-bye Ludwigs2, good-bye Crum375, good-bye SlimVirgin. I won't be pursuing this. Regards, --Bob K31416 (talk) 04:00, 6 January 2010 (UTC)

Well put, Ludwig, with the idea of juxtapositional induction. If we could write the policies using that kind of vocabulary, things would be much clearer, but people would complain. So we have to simplify it, but in doing that problems arise because the vocabulary we use is imprecise. SlimVirgin ^{TALK contribs} 12:52, 6 January 2010 (UTC)

Lets look again at the policy lead, which is what Bob was talking about. The first paragraph proscribes "...any unpublished analysis or synthesis of published material that serves to advance a position." There is clearly the possibility that someone could misinterpret this to imply that summarising a source is not allowed. Therefore, I think we should make it clear that summarising is actually a good thing. Using the sentence "Carefully summarizing or rephrasing source material without changing its meaning or implication is not a violation of this policy, it is good editing." Seems perfectly logical. Yaris678 (talk) 19:59, 7 January 2010 (UTC)

well... how about something like this, to replace the last line of the first paragraph? "Wikipedia editors should not use wikipedia to publish their own opinions, experiences, arguments, or conclusions, but should carefully summarize or rephrase verifiable source material without changing its meanings or implications." --Ludwigs2 20:16, 7 January 2010 (UTC)

It's OK, but I have three problems with it and they all point to the same thing.

That's quite a long sentence.
I would prefer it if we made it more obvious that we were talking about the limits of the policy.
I like the phrase "good editing". Its feels like a pat on the back.

i.e we should say "This means that Wikipedia is not the place to publish your own opinions, experiences, arguments, or conclusions. However, carefully summarizing or rephrasing source material without changing its meaning or implication is not a violation of this policy, it is good editing."

Yaris678 (talk) 20:39, 7 January 2010 (UTC)

I notice no one has responded to my suggestion above (of 7 Jan). Should I implement it then? Yaris678 (talk) 11:07, 16 January 2010 (UTC)

No. Further down the policy already says "carefully summarizing or rephrasing source material without changing its meaning or implication is not a violation of this policy, it is good editing." We can't put everything in the lead, otherwise it wouldn't be the lead, it would be the policy. --Jc3s5h (talk) 16:06, 16 January 2010 (UTC)

You might want to expand on that. Obviously, not repeating yourself and not saying something in the lead that can be left till later are good. However, we have already established that the lead, as it stands, is potentially misleading. In addition, the advice applies to the whole or NOR policy, not just WP:SYNTH. Someone could easily not read that bit of the policy and get the wrong idea. Or they might read it and think it only applies to WP:SYNTH. Yaris678 (talk) 15:20, 18 January 2010 (UTC)

WP:SYNTH: add the word "new"

I think the word "new" before the word "conclusions" in the first sentence is understood, but unfortunately unexpressed. It is, however, necessary to make it explicit. I would just be bold and add it, but... anyhow... If you have one RS that says "Reese's Peanut Butter Cups taste good because of the chocolate" and another RS that says "Reese's Peanut Butter Cups taste good because of the peanut butter", then it isn't SYNTH to have an introductory sentence that says "Some sources say Reese's taste good because of the chocolate, while others suggest that the peanut butter is the key element". That is simply the style of introductory topic sentence that is customary in English language literature, essays, etc. • Ling.Nut 02:38, 10 January 2010 (UTC)

I may be missing something, but I don't think the word 'new' is needed or desired, since we don't want editors to add their own conclusions, period. In the example you cite, you are correct that it isn't SYNTH to combine elements from the two sources, because in this case there is no position being advanced. SYNTH is prohibited only when used to advance a position, which is clearly not being done here. Crum375 (talk) 02:50, 10 January 2010 (UTC)

Clarity is good, however. The more clarity, the fewer opportunities for gaming the definition. • Ling.Nut 07:39, 10 January 2010 (UTC)

But adding unnecessary words to policies can often add confusion, so it's a fine balance. Crum375 (talk) 12:16, 10 January 2010 (UTC)

It does not matter whether the conclusion is "new" or not... we should not state a conclusion (drawn from a given set of facts) unless a reliable source has stated the same conclusion (drawn from the same set of facts). Blueboar (talk) 14:58, 10 January 2010 (UTC)

Areas of islands

Would it be considered original research to calculate the area of an island from a published map by using a computer to count the pixels within its outline and then doing the simple calculation based on the stated scale of the map? 81.157.197.66 (talk) 01:44, 10 January 2010 (UTC).

I would say yes, that would be OR. Blueboar (talk) 02:55, 10 January 2010 (UTC)

I agree with Blueboar. First, this isn't "simple arithmetic". Second, island coastlines can fluctuate over time or seasonably. Third, some coastlines can be very ragged, which could significantly affect the accuracy of the measurement. Fourth, it's easy to make a mistake in the counting (e.g. a software bug if it's being done programatically, or a counting or multiplication error if done manually). I am sure there are other reasons, but this should suffice for starters. Crum375 (talk) 03:03, 10 January 2010 (UTC)

The possibility of making a mistake does not, to me, seem particularly relevant. It's easy to make a mistake adding up a bunch of numbers. It's easy to make a mistake converting between units. In fact, it's pretty easy to make a mistake in virtually any information added to Wikipedia. The fluctuation of coastlines is not an issue specific to this method. Any area measurement is based on a particular interpretation of where the coastline lies at some particular point in history. A surveyor who publishes a calculated area has to make this judgment, just as does a map-maker when deciding where to draw the island's outline. Finally, the small-scale raggedness of the coastline only has a small proportional effect on area measurement (though it has a wild effect on the measurement of the length of the coastline). There is no reason why an area calculated from a good-quality map should not be accurate to an acceptable number of significant figures. (I'm talking about very small islands, where complexities related to map projections can be ignored.) In fact, this is essentially the way that all area figures are derived. 81.151.231.107 (talk) 14:24, 10 January 2010 (UTC).

To me the reason why this is OR is that the method is novel. Why not just print out the map and calculate the aproximate area using an old fashioned ruler. Blueboar (talk) 14:52, 10 January 2010 (UTC)

It's not especially easy to calculate irregular areas using a ruler, except by drawing a grid and counting squares, which, if the source map is in digitised form, is exactly what the pixel-counting method is doing. 86.150.102.84 (talk) 02:32, 11 January 2010 (UTC)

"Any area measurement is based on a particular interpretation of where the coastline lies at some particular point in history" - interpreting where the coast lies violates WP:OR. We need a reliable source to do that for us. Crum375 (talk) 17:40, 10 January 2010 (UTC)

The map is the reliable source that does this for us. It does so by depicting the coastline. The area information is thereby inherently present in the map so no interpretation or OR is involved. 86.150.102.84 (talk) 02:32, 11 January 2010 (UTC).

The coastline on a map is from one point in time; other maps from other periods could yield different results. The pixelization of the coastline introduces more uncertainty and inaccuracy and depends on the specific algorithm chosen. Counting the pixels and converting them to an area could introduce errors. Choosing one map over another may introduce a bias and produce different results. All these issues are far beyond what we mean by "routine calculations, such as adding numbers, converting units, or calculating a person's age," and well into the realm of original research. The idea of "routine calculations" is that we permit calculations that any reader can easily perform to verify, and this is far beyond it. If the island is notable, and its area is notable, the area will be calculated by a reliable source and we can report it. Crum375 (talk) 03:29, 11 January 2010 (UTC)

(i) Your first point applies equally to any source that gives an area. Different figures published at different times may be different. Yet we do not dismiss them all and say that none are usable. (ii) The calculation of an area by this method is not claimed to be exact to the nearest square metre, but is, with a high quality map at sufficient scale, claimed to be accurate enough for the present purpose. (iii) The choice of one map over another is the choice of one source over another. As with your first point, this applies to sources of any type and has nothing specifically to do with maps or with this method. Conflicting published area figures abound. (iv) The fact that it is relatively difficult for others to verify the figure, given only the source map is, so far, the only valid objection to this method that I can come up with or that has, in my opinion, been presented here. 86.133.245.37 (talk) 02:02, 12 January 2010 (UTC).

If there is not a convenient published source for the area of an island then what you propose should work well, or well enough. Islands are smallish so that issues of map projection aren't major (and if there ever is an issue of map projection then either your algorithm could be adjusted to take that into account or you shouldn't use it.) Also, you're measuring the area, not the coastline, so coastline variations aren't really that big a deal. Further, I assume your goal is to provide a notion of the approximate size of the island, not an ultra-precise measure. If your measure yields the result that Island A is 400 square miles and if somewhere it is stated that Island B is 399 square miles it would be inappropriate for anyone to use the two figures to state that Island A has a greater area than Island B. (Not that this much matters.) On the other hand if some reliable source says Island C has an area of 100 square miles the corresponding comparison is safe. If there is an island that has some characteristic that makes its area vary a great deal over time then probably you should simply state that the area, as computed using map whatever, is approximately x.

Heck, if objections are to be raised then perhaps the better objection is that for a mountainous island your method only computes the area of the island as projected onto a horizontal plane (or onto a sphere at average sea level.) If the island were a perfect cone with a slope of 60 degrees then the actual area of that cone will be more than the area you'd calculate. I'd guess that for geographers the meaning of 'area" is the projection area, not the actual area in 3D so your method would be consistent with what they do. Also, of course, an island on a map that is represented by a small number of pixels is (a) a small island and (b) an island for which your method is probably inappropriate (or the method is not appropriate for that particular island on that particular map.) Which I suspect you already understand. 24.177.114.132 (talk) 04:21, 20 January 2010 (UTC)

"original research" on games..

question, take this senario: a new game was just released, and there was no documentation on its storyline etc online, and you played it, and finished it.. then, would you have to wait for other people to talk about it online, then to cite them, because your is "original research"? or could you just put the "game" as reference? as "in the game when you come to X, see what happens.." or should you upload images (if possible) of the game as proof/reference? or what? Divinity76 (talk) 11:07, 13 January 2010 (UTC)

To some extent, this depends on the individual game... and whether the game play will be the same for all players. Many games are not structured with a set "plot"... The game play will be different for each individual player... Think of something like Roller Coaster Tycoon. You can not say: "You should then place a thrill ride near the top of the hill to draw people to the far side of the park" because that is advice and not a "must do" to move forward in the game.

On the other hand, many games do structure the game play into the equivalent of a plot... If it is a case of "After doing X, the player comes to Y and must do Z", and this is the same for all players, then we can say that the game has a distinct plot (or a distinct plot element) ... and to that extent, the game play is no different than the plot of a movie or a book. We can give a simple description of the "plot element" and cite it to the game itself.

However, in many cases the "plot" is actually too complex to summarize as a simple description... running more along the lines of: "Having achieved X the player can go to A, B, C or D... at A he has a choice of R, S or T or Q... while at B he has a choice of Q, R or T" (etc.) So the question becomes: how detailed do we want to get... while Wikipedia does not worry about "spoilers", we are not a game guide. There is a fine ballance to be achieved here. That said, how detailed you get is more a decision of appropriateness than a case of violating "OR".

What would be definitely be OR is to outline some sort of "optimum strategy" ... as that would be a matter of opinion and something that is not readily apparent from playing the game. Blueboar (talk) 16:33, 13 January 2010 (UTC)

There is a way in which the plot of the game is not like the plot of a book or film. WP:PSTS says "a primary source may be used only to make descriptive claims, the accuracy of which is verifiable by any educated person without specialist knowledge." Ability to play the game to the end could be called specialist knowledge. It's not as straight-forward as reading a book, although arguably some books require a lot of persistence, if nothing else. Yaris678 (talk) 18:43, 13 January 2010 (UTC)

While not everyone would be able to reach the end of a game. There are likely several editors who played the game and won and can confirm the accuracy of the plot. I don't see this as a major issue.--76.66.180.179 (talk) 21:34, 16 January 2010 (UTC)

Re those several editors, see WP:OR: "... Wikipedia is not the place to publish your own opinions, experiences, arguments, or conclusions." Wtmitchell (talk) (earlier Boracay Bill) 23:12, 17 January 2010 (UTC)

I think that is a bit of a red herring. No one is arguing that people should describe game experiences that are specific to them. Indeed, Blueboar made it clear that this should be avoided. But describing the plot of the game, in a way that can be verified by someone else that plays the game, is similar to describing the plot of a book or film. What we are discussing is the extent to which it can be verified and hence the limits to what can be added to Wikipedia. Yaris678 (talk) 13:00, 18 January 2010 (UTC)

I think saying that beating a game in general is "specialized knowledge" is using a wide brush to paint what is likely a more narrow critiera. Certainly there are games that beating it could be considered to have specialized knowledge, but the same could be said for some books in specialized fields that are tailored for experts.

As for what you can say, if its a matter of fact ie, "The player takes control of a red car" that's fine. The line is crossed when it says "The player takes control of a red sports car" unless the game expressly says its a sports car. It is definatly crossed when you say something like "The player takes control of an awesome red sports car." If the game uses the word "awesome" then something like "The player takes control of what the game describes as 'an awesome red sports car.'" -if indeed it uses that exact phrase.陣内 _Jinnai 23:41, 17 January 2010 (UTC)

An example of original research

...gets press coverage. One editor's attempt to tie Burj Khalifa to Goethe is covered in the UAE's The National:[2] Fences&Windows 03:03, 22 January 2010 (UTC)

Deletion of examples of primary sources from PSTS

SV tightened up PSTS and removed quite a number of examples of primary sources. Here's the original list, most of which has been removed: " Other examples include archeological artifacts; photographs; historical documents such as diaries, census results, video or transcripts of surveillance, public hearings, trials, or interviews; tabulated results of surveys or questionnaires; original philosophical works; religious scripture; published notes of laboratory and field experiments or observations written by the person(s) who conducted or observed the experiments; and artistic and fictional works such as poems, scripts, screenplays, novels, motion pictures, videos, and television programs." I thought this list was useful. Since editors can delete material based on its being a primary source, it seems like it might be useful to have these examples so that editors have a clear idea what sources are primary. TimidGuy (talk) 12:32, 20 January 2010 (UTC)

I agree with the removal of the excessive examples because including a long list of examples (as opposed to just a couple) creates a misleading impression (at least for some people) that anything not on the list is not primary, or less likely to be so. We need to describe the principle of what is primary, then give at most a couple of examples, and let editorial common sense take over from there. Crum375 (talk) 13:32, 20 January 2010 (UTC)

Being supported by a primary source is not sufficient reason, by itself, to delete material, although it could be a factor. I hope TimidGuy isn't going around deleting every primary-source-supported statement he/she sees. --Jc3s5h (talk) 14:36, 20 January 2010 (UTC)

The guideline now fails to give any guidance as to if a scientific paper is a primary or secondary source. Since these are very widely-used sources, removing them from the list of examples probably wasn't a good idea. Tim Vickers (talk) 18:07, 21 January 2010 (UTC)

It hasn't been removed. It's in footnote 2. People were complaining about the length and wanted to see the list removed entirely. As a compromise, I moved it to a footnote. SlimVirgin ^{TALK contribs} 18:14, 21 January 2010 (UTC)

(edit conflict) I suppose this might be covered by the entry on "notes" about experiments, but laboratory notes are different from scientific papers. I've changed this to read "descriptions of experiments or observations written by the people who conducted or observed the experiments", which should cover both research papers and lab notes. However, this won't cover the material in the introduction of the paper where the authors review other people's work and describe the field, this material is probably a secondary source, even if the data itself and the authors' interpretation of their data is a primary source. Confusing huh? :) Tim Vickers (talk) 18:18, 21 January 2010 (UTC)

OK, that makes sense. Tim Vickers (talk) 18:20, 21 January 2010 (UTC)

I see that this (IMO reasonable) change has been reverted as being too broad. CBM, if you would, please tell me how my own initial paper about what I personally saw happen in my lab in a novel experiment could possibly be anything other than a primary source. Being an eyewitness to a train wreck in my lab doesn't seem to me to be any different than being an eyewitness to a train wreck down the street. WhatamIdoing (talk) 20:26, 21 January 2010 (UTC)

It's clearly a primary source. To try to address this problem, I replaced "An account of a traffic accident written by a witness is a primary source of information about the accident" (which is not that helpful an example: how many traffic accident reports by witnesses are cited in Wikipedia?) with the far more helpful and apropos "An account of a scientific experiment written by the researchers who conducted the experiment is a primary source of information about the experiment." I also removed the mention of experiments from the footnote, as it's now redundant. Eubulides (talk) 20:36, 21 January 2010 (UTC)

The notes and data are primary sources. However, because of the arcane way that our policies are written, any "interpretation" or "analysis" in your paper has to be counted as a secondary source. Because the policy here says "all interpretive claims, analyses, or synthetic claims about primary sources must be referenced to a secondary source, ...".

Example: suppose you measure heat increasing during an experiment, and you write in your paper, "Because the temperature increased, the reaction is exothermic". If we classified that explanation as a primary source, we have to say something like

According to XXX, the temperature increased because the reaction was exothermic" [1]

But in reality our article would just say

The temperature increased because the reaction is exothermic.[1]

I realize this is stupid, but it's the way the policy page happens to be written: the analysis and interpretation in papers is a secondary source for our purposes. — Carl (CBM · talk) 20:40, 21 January 2010 (UTC)

I am OK with "account of an experiment", but we have to be careful of the point I just raised. — Carl (CBM · talk) 20:42, 21 January 2010 (UTC)

It seems like an unpublished account from a scientist, for example on the scientist's website, is a primary source, whereas an account published in a scientific journal is a secondary source because it comes from the journal and not the scientist. Perhaps something along this line is a way it can be clarified in the policy. --Bob K31416 (talk) 20:47, 21 January 2010 (UTC)

The mere act of publishing an account in a peer-reviewed scientific journal does not magically turn a primary source into a secondary source. What counts is that it's a first-hand account, by the investigator who did the experiment. My edit tried to make this clear. Also, I noticed that the page had zero examples of secondary sources (though it had examples of both primary and tertiary). This was pretty confusing, so I added one. Eubulides (talk) 20:53, 21 January 2010 (UTC)

(edit conflict) I agree with Eubulides.

Also, putting something on your own website is "publishing" it. If a paper is a primary source when I self-publish it, it's still a primary source when it's published in a newspaper, magazine, or scholarly journal. It's the same words.

Bob, based on your comments, I'm going to assume that you're not a researcher. In the sciences, the journal doesn't -- and can't -- know whether or not you're telling lies. They have no way of checking up on your work. The peers are never shown your lab notes. The notion of peer review is more like "Assuming she's telling the truth under "Methods" and "Data", then these "Conclusions" seems reasonable to me (or: Given that she didn't tell us how she did ____, then we have no clue whether these conclusions are reasonable)." Peer review is not magic; it is, in fact, quite fallible. It's just that, in common with what is said about democracy, it's the worst possible system except for all the others. WhatamIdoing (talk) 20:59, 21 January 2010 (UTC)

(edit conflict)Well, that was interesting. I think it's time for me to leave. Pardon my interruption. Bye. --Bob K31416 (talk) 21:26, 21 January 2010 (UTC)

(unindent) I must disagree with Carl's analysis. A journal article by a scientist who conducted an experiment is mostly a primary source. We can use it, with due caution. We can use anything in the paper, whether it is reporting of facts or interpretive. The part in our policy about "all interpretive claims, analyses, or synthetic claims about primary sources must be referenced to a secondary source, ...". means that a Wikipedia editor is not allowed to interpret, analyze, or synthesize information from primary sources to reach a novel result not contained in the sources. Of course, if there is a result that can be derived from the primary sources that is not stated in any of them, we will have to find it in a reliable secondary source before we can put it in an article. But the reason it has to be a secondary source is not a rule we made up, it is a logical consequence that since it is synthesizing primary sources, it is by definition a secondary source. But if there is synthesis already present in the primary (or mostly primary) source, we can just use it. No need to go looking for it in a (mostly) secondary source. --Jc3s5h (talk) 21:16, 21 January 2010 (UTC)

You would think so, but I have seen people claim the opposite: that even if a primary source includes interpretation or analysis we can't use that interpretation or analysis, and instead he have to find a different source for it. That is, the policy does not include the words "not contained in the sources" relating to analysis and interpretation that you included in your post.

Of course we do use journal articles for their interpretation and analysis. Conclusion: the parts of the journals that we use for interpretation and analysis are secondary sources for us. This is fine because the same source can be both a primary source (for data) and a secondary source (for analysis of that data). It would be better if the policy were phrased differently, but things are what they are. — Carl (CBM · talk) 21:45, 21 January 2010 (UTC)

I sort of agree with this reasoning, although it seems that CBM's tweak of the page conveys the opposite impression - that the interpretation section of a research article is primary. Primary and secondary depend upon context (see Primary_sources#Classifying_sources). There's a tendency to oversimplify and say that "research articles = primary" and "review articles = secondary" (see Wikipedia:MEDRS#Definitions, or Eubulides' recent change). The problem is not only Carl's point, but also the point noted by Tim that research articles have substantial introduction/discussion sections. This secondary information is often more up-to-date and focused on a particular topic than review articles (which are more rare and often written by individuals rather than teams). Following Carl's reasoning (if I understand correctly) it should be noted that investigative journalism articles are primary and that scholarly reviews are primary sources for their synthetic conclusions (reviews are primary for everything except their repeats of another articles' statement).

The NOR policy page currently says that an article should not be based primarily on primary sources. A plain secondary source of a primary conclusion, by definition, does not add value, because when value is added the article becomes a primary source for its novel conclusions. Citing a secondary source which parrots a primary source makes the reader do extra work and misleads the reader; it gives the reader little impression of the actual research underlying the claim. For example, a review which cites a small study or two to support a conclusion is not really more reliable than the research it cites. Another example: if I find a review which incidentally cites a metaanalysis, do I cite the review or the metaanalysis? I cite the metaanalysis. If the metaanalysis is used in connection with a novel conclusion, I cite the review. Citing secondary sources rather than the original source can make the editors work harder: if I recall correctly (this was a while ago), at water fluoridation Eubulides cited a "mini-summary" of a huge review to support a key statement. This summary was from a specialized journal not available to anyone else, Eubulides would not email it, and it turned out that after we got ahold of it the "mini-summary" cited another review (Worthington & Clarkson 2003 I believe), which may have cited another article to support its statement. When I've written my own papers, I try to show both the secondary source and its citation whenever possible. Unfortunately this is more work and not conveniently done with footnotes. Incidentally, Bjorn Lomborg was widely blasted for his use of secondary sources, which made it harder for people to check his sources, even though he was meticulous about page numbers. Obviously it's not really feasible to cite a lot of research articles on many areas, but it's not necessarily a bad thing if the topic is small. II | (t - c) 23:53, 21 January 2010 (UTC)

Carl, the conclusions I draw from my data are still primary. For example: We'll run an experiment on methods of making friends. To half the people in the lunchroom, I'll talk about baking natural sourdough bread. To the other half, I'll talk about baking chocolate cake. We'll record their responses, and control for reasonable factors, such as the presence of incubators in the victim's -- I mean, the participant's lab.

I'll present data -- overall, people are more responsive to chocolate cake than to sourdough bread; the range of responses is broader for bread than for cake -- and I'll conclude something (probably that anyone who's ever cleaned a fungal infection out of an incubator is unlikely to be cordial if you even mention bread baking).

Every bit of this is a primary source, not just the data. I'm not getting my conclusions from another party, and therefore the conclusions are still part of my first-person, first-party, primary source. WhatamIdoing (talk) 00:23, 22 January 2010 (UTC)

Responded below. However, let me say that even if a new history book includes an original theory about the U.S. Civil War, no new history book can possibly be a primary source about the Civil War. — Carl (CBM · talk) 13:30, 22 January 2010 (UTC)

True: a modern book is not a primary source for the war. It is, however, a primary source for the author's new theory about the war. WhatamIdoing (talk) 22:47, 25 January 2010 (UTC)

It might be worth mentioning the fact that a given publication might be both primary and secondary, e.g., the many "case report and review of the literature" papers for rare diseases.

Additionally, if we want to be on firmer ground with review articles, we could say that specifically narrative reviews are secondary sources. A meta-analysis or at least certain types of systematic reviews could be understood as a primary source. WhatamIdoing (talk) 01:20, 22 January 2010 (UTC)

I agree with WhatamIdoing: an original research article is not a secondary source for its own conclusions. It is a primary source for its own conclusions, just as a witness in a court case is a primary source for the witness's own conclusions. A research article's "related work" section may be a secondary source for other researchers' work, but that's a completely different matter. (These "related work" sections are typically less reliable than review articles, because they're not the focus of their authors or their reviewers.) Eubulides (talk) 06:58, 22 January 2010 (UTC)

That is a reasonable position, it just doesn't fit with "Any interpretation of primary source material requires a reliable secondary source for that interpretation" from the policy. Clearly the data in a research paper is a primary source. If we declare a researcher's conclusions about that data to also be a primary source, then we cannot use the researcher's own interpretation as a source for the interpretation of the data. Note that the sentence above, which I copied from the policy right now, has no caveats or exceptions: it directly says that any interpretation of primary source data has to come from a secondary source. The simplest solution to make the written policy match actual practice is to treat interpretive sections in research papers as secondary sources. If you would like to treat them as primary sources, then the sentence I quoted above will need to be qualified somehow. — Carl (CBM · talk) 13:28, 22 January 2010 (UTC)

Also, if we are too liberal in the scope of "primary source", we run into issues with "Do not base articles entirely on primary sources.". There are plenty of articles that are sourced only to peer-reviewed journal articles, and that is not a deletion concern. — Carl (CBM · talk) 13:33, 22 January 2010 (UTC)

Sigh... Once again, we are sidetracked by focusing on whether something is Primary or Secondary... when the important thing is whether we are using the source appropriately or misusing the source in a way that constitutes OR. It does not matter whether a scientific journal article is Primary or Secondary (or a little of both)... because OR is not determined by whether the source is Primary or Secondary. It is not OR to cite a primary source. It is not OR to discuss what a scientist says in a published journal. It is not OR to discuss a researcher's published interpretation of his or her data... although we should attribute that interpretation to make it clear that the interpretation being discussed is that of the researcher. The whole point of NOR is that we may not include our interpretations. Yes, we need to be careful, because it is easy to cross the line between discussing the researcher's interpretation and discussing our own interpretation. But as long as we are careful, and don't cross that line, there is no problem. Blueboar (talk) 15:22, 22 January 2010 (UTC)

Yes, that's the point, but it's not what the page actually says. The PSTS section is very emphatic that interpretation must come from a "secondary" source. I would be happy to see that fixed, but at the very least we have to be realistic about what the policy actually says. — Carl (CBM · talk) 15:34, 22 January 2010 (UTC)

Well... yes and no... the policy allows for descriptive statements about what is contained in Primary sources... so to the extent that a Primary source contains an interpretation, it is appropriate to give a basic description of that interpretation (and I would say that part of that description should contain clear attribution so the reader knows who's interpretation we are describing). What the policy is trying to clarify is the fine line that exists between mearly describing (ie summarizing) what the source says (which might include an interpretation of something else) and going beyond the source and inserting our own interpretation of what the source says. Blueboar (talk) 17:28, 22 January 2010 (UTC)

That's a very good point, Blueboar. I've tried to re-focus that part of the policy on the novelty of interpretations, to deal with this issue. Tim Vickers (talk) 17:49, 22 January 2010 (UTC)

Tim Vickers: well done. --Jc3s5h (talk) 19:08, 22 January 2010 (UTC)

A description of an interpretation isn't an interpretation; it's a description. Primary sources can't be used in an interpretive way. We can only use them to say, "A says x." But we also need a secondary source to tell us that it matters that A says x, something that A alone can't act as a source for. A alone cannot be used to say, "This is what I say, and furthermore what I say is important." That is to misuse the primary source -- to use it as though it's a secondary source. That's the point of this section. SlimVirgin ^{TALK contribs} 19:51, 22 January 2010 (UTC)

The two versions have the same meaning, so I'm not worried which we use (although I think focusing on novelty is clearer). They are the same since any source that makes a new interpretation of a primary source is, by definition, a secondary source. Therefore you can either say, "no new interpretations of primary sources" or say "all interpretations of primary sources must be cited to secondary sources" - the two formulations are exactly equivalent. Tim Vickers (talk) 20:28, 22 January 2010 (UTC)

I can't quite see what would be meant by "new" or "novel". It's any interpretive (analytic or synthetic) use of a primary source that's problematic. What this means is we can't use a primary source to say: "Scientists have shown that x," because it's just one study that claims to show x, and there's no indication that's it's a study that's taken seriously. That's misusing a primary source to make an interpretive claim. It's not connected to whether it's new or not, unless I'm misunderstanding the point. We also have to avoid a primary source to say more specifically, "Scientists at University X showed in 2010 that ...," unless we have secondary sources telling us that that study is worth mentioning. Again, to use the primary source alone would be original research—meaning a Wikipedian substituting himself for a secondary source, and turning Wikipedia into a secondary source. SlimVirgin ^{TALK contribs} 20:37, 22 January 2010 (UTC)

"Novel" means "not previously published elsewhere". It is equally problematic to make novel interpretations of secondary or tertiary sources - "The Encyclopedia Britannica says that the Earth is not perfectly round,^{Ref to EB} which implies a pushing force at the poles." It is new interpretations made by a Wikipedia editors that are the problem, not what classification of source they happen to be based on. Tim Vickers (talk) 21:11, 22 January 2010 (UTC)

Agree, regarding "novel." But the point about primary sources is that they shouldn't be used for any interpretive (analytic or synthetic or evaluative) claims at all, unlike secondary sources, which can be so used. See below. SlimVirgin ^{TALK contribs} 21:17, 22 January 2010 (UTC)

More examples

Re Blueboar: Here is a more concrete example. Suppose that I am writing an article on some notable mathematical topic X. Suppose that I want to include the fact that Joe Smith proved that every X has property Y. Normally, I would do so like this:

Every X has property Y (Smith 2000).

However this goes against the "descriptive" limitation in NOR, which SlimVirgin summarizes as "We can only use [primary sources] to say, 'A says x'." According to that interpretation of NOR, my article would have to say,

According to Smith (2000), every X is a Y.

That not only fails to cover our actual practice, but it's genuinely bad. The second phrasing strongly suggests to our reader that there is a reason to doubt the claim, or that other people claim other things, and thus presents an NPOV problem if nobody actually doubts the result is correct. Moreover, there is no way to integrate method (2) into a list of properties of X (if property Y is just one of many). So, to the extent that the NOR policy says the second way must be used, it's wrong. I agree with you that the problem is caused by trying to shoehorn things into "primary" and "secondary" in the first place. — Carl (CBM · talk) 20:34, 22 January 2010 (UTC)

You must not use Smith alone to say "Every X has property Y (Smith 2000). That's misusing a primary source. You need a secondary source to show that Smith is worth mentioning in this regard. Smith may have said it, but may be talking nonsense, or may be a non-notable source, or may be out of date, or it may not be the whole story -- and the Wikipedian adding it might not have realized this. Hence the importance of secondary sources. SlimVirgin ^{TALK contribs} 20:41, 22 January 2010 (UTC)

No, you are misunderstanding the role of secondary sources. We need secondary sources to establish the notability of an article topic. Once the topic is notable, we do not need secondary sources to establish the worth of every piece of information included in the article. The decision about what to include is completely a matter of editorial discretion, subject to NPOV and the goal of writing an encyclopedia. In particular, we routinely include material sourced to journal articles without even thinking to look for a secondary source that says the same thing. Not all of the material that belongs on Wikipedia can be found in textbooks or other secondary sources. — Carl (CBM · talk) 20:48, 22 January 2010 (UTC)

For an entirely non-contentious point, it would be okay to use a primary source in the way you describe. But for anything that might be contentious, or anything that moves from the specific to the general, you need a secondary source that discusses it. For example:

Professor Smith (2000) says that every X is a Y.
The Giant Turnip Society (2010) says that God is a giant turnip and that SlimVirgin is his representative on Earth.

In both cases, you have to give the reader some reason to believe both (a) that Professor Smith and the Giant Turnip Society are people we should listen to, and (b) that the specific claims they make about X or God or SlimVirgin are claims regarded by others (not just by a Wikipedian) as worthy of mention. The essence of the NOR policy is that Wikipedia itself should not be a primary or secondary source. It should be a tertiary source. SlimVirgin ^{TALK contribs} 21:00, 22 January 2010 (UTC)

Here is an even more specific example. The featured article Germanium says,

Germanium has been detected in the atmosphere of Jupiter [48]

The reference is to a journal paper that analyzed data from a Voyager probe. The article does not say,

According to Kunde et al., Germanium has been detected in the atmosphere of Jupiter [48]

The paper referenced is certainly a primary source. Our (featured) article does not make any attempt to show why we should care who these people are, or why the topic of Jupiter's atmosphere is worthy of mention in an article on Germanium. The claims you are making are simply not accurate regarding our actual use of journal articles on Wikipedia. — Carl (CBM · talk) 21:07, 22 January 2010 (UTC)

It doesn't have to say, "according to X" in the text. It does need a secondary source. A primary source can be used alone if it's entirely non-contentious, but if it's a widely accepted point, there will be a secondary source anyway. It needn't actually be attributed to a secondary source, but it must be attributable. If no secondary source can be found for a point you want to make, you're engaged in OR. The question then is whether it's a trivial point that no one is going to challenge, which is an issue for WP:V. But if you're challenged and can't produce a secondary source, you would have to remove it. SlimVirgin ^{TALK contribs} 21:13, 22 January 2010 (UTC)

Everything in Wikipedia must be verifiable, but it is simply false that everything in Wikipedia must be verifiable via a secondary source. No policy has ever taken that position, and it is completely at odds with general practice. You're also incorrect that the Germanium article would "have to remove" a fact such as this if a secondary source could not be provided; the source provided, a peer-reviewed paper announcing that Germanium was detected around Jupiter, completely meets the requirements of WP:V. If you do not understand our actual best practices regarding the use of journal articles, you should think twice about writing about them in the NOR policy. — Carl (CBM · talk) 21:21, 22 January 2010 (UTC)

This policy has always taken that position, and you've always argued against it. Are you seriously arguing that, if only one primary source says X was detected near Planet Y, and no other source existed, it would still be regarded as credible? No. Sorry, but that's just wrong. I can assure you that if someone submitted a claim at FA that was challenged, and it could be sourced only to one primary source, and no other source existed that even alluded to it, it would be removed. SlimVirgin ^{TALK contribs} 21:28, 22 January 2010 (UTC)

Yes, a peer-reviewed paper is generally considered credible unless there is a reason to think otherwise. The article I pointed is an FA. However - and this is also important - the standards at FA are often higher than policy requires, and so simply because something would not pass FA does not mean that it violates policy. But the example I provided is an FA anyway. — Carl (CBM · talk) 23:16, 22 January 2010 (UTC)

But the claim in the primary source in the FA is attributable to a secondary source. If it is not, it could be challenged as OR, if it is a contentious point. Note the difference between attributed and attributability; verified and verifiable. SlimVirgin ^{TALK contribs} 23:55, 22 January 2010 (UTC)

No. There is nothing in WP:NOR nor WP:V that requires everything to be attributable to a secondary source. If there was such a rule, I am sure it would already be mentioned in either WP:V or PSTS. — Carl (CBM · talk) 00:02, 23 January 2010 (UTC)

There is nothing that requires attributability to a primary source per se, but the source must be appropriate for the situation. If the situation is contentious, you need a secondary source. Same for a new discovery which is only mentioned on a primary source — its importance (if any) and perspective would require a secondary source, one that writes about the primary source. Crum375 (talk) 00:12, 23 January 2010 (UTC)

You're arguing this way because you do a lot of work using primary sources. But I guarantee you that, if someone else were doing it in a way you disagreed with, and you asked for a secondary source to support his point, and found that none existed -- that not one single source anywhere in the world had seen fit to repeat the information in the primary source -- you would insist it be removed, and it wouldn't simply be a notability issue. It would be because it was OR. SlimVirgin ^{TALK contribs} 00:11, 23 January 2010 (UTC)

That sounds like an NPOV issue. Content within an article is never a notability issue; "notability" only refers to article topics. — Carl (CBM · talk) 00:45, 23 January 2010 (UTC)

Re Crum. To take a specific example: there is no requirement that a new scientific discovery must be discussed in secondary sources before it is cited to primary sources here. The article topic must be notable, but if there is some important new result that is only cited to primary sources, it is pretty much always included despite not being attributable to secondary sources. Here "important" means in the consensus of the editors who are evaluating the page for NPOV. Examples:

When a new theorem is proved, or an important new experimental result is obtained, it is usually included in the article here once it is verifiable, regardless whether it is in a secondary source yet.
When a person dies, we wait until their death is verifiable, but not until it is mentioned in secondary sources. In academia, passings are often announced via email lists and university news services. We try to wait until there is some reliable, public announcement, but it will rare to even have a newspaper story for months after the death.

— Carl (CBM · talk) 00:45, 23 January 2010 (UTC)

Random break

I think we have two issues here.

WP:NOR does not talk about contentiousness. The nearest thing is limiting itself to things that do not serve to advance a new position, but that relates to things not explicitly stated in sources. The existence of Germanium in the atmosphere of Jupiter is stated in the source.
The paper on Germanium in the atmosphere will have been peer-reviewed. This means that some non-authors have agreed with what it is saying.

Perhaps the rules on primary sources should be slackened off on non-contentious statements from peer-reviewed papers. Yaris678 (talk) 22:07, 22 January 2010 (UTC)

(unindent) SlimVirgin's standard is unrealistic. Of course a particular journal article or the like (whether primary or secondary) can be shown to be not credible, either by being contradicted by better sources, or by being shown to be unsupported by any other source on a matter where support would be expected. But SlimVirgin is essentially saying "don't add anything unless it is clear it would win any challenge that might come along". That is unnecessary. Being stated in a primary source on an uncontentious matter is sufficient to include in Wikipedia, even if it does not guarantee it will win any discussion that might happen on the article's talk page.

A further point is that if something appears in a primary source published by a reliable publisher, it is certainly not original research and any criteria for excluding it should be covered in some other policy. --Jc3s5h (talk) 22:11, 22 January 2010 (UTC)

I said they're fine for uncontentious points. But if its use is challenged, and if no secondary source exists, you would have to remove it, because picking and choosing which primary sources to use without secondary-source back-up, is OR, for obvious reasons. SlimVirgin ^{TALK contribs} 22:11, 22 January 2010 (UTC)

"Picking and choosing which primary sources to use without secondary-source back-up, is OR, for obvious reasons" is incorrect. Picking and choosing sources is source-based research, not original research. It can be well-done, or badly done. It can be done in a neutral manner, or a partisan manner. It is wrong to indicate that any kind of bad writing in Wikipedia is original research. --Jc3s5h (talk) 22:19, 22 January 2010 (UTC)

That is incorrect. Any use of primary sources to present or create significant new information not already covered by secondary sources is original research, because even though the primary source may have been reviewed by peers, we still need a secondary source writing about that primary source and putting it in perspective for us. Otherwise, we'd be acting as a secondary source ourselves, and that's not our mission — Wikipedia is a tertiary source. Crum375 (talk) 22:32, 22 January 2010 (UTC)

I disagree with Crum375. Original research only occurs when an editor creates significant new information. This policy expands this to include information that appears to be original research because the editor didn't provide a citation. If the information already exists in sources that can be cited in Wikipedia, it isn't original research. It would be original research to give the information a different degree of certainty than could be inferred either by the explicit statements in the source, or the nature of a source. --Jc3s5h (talk) 22:50, 22 January 2010 (UTC)

If a statement is challenged or likely to be challenged, it requires a source. If we only rely on a primary source to create or present significant new information, we are creating an implication (stated or just implied) that this new information is significant, or important. But this is the role of a secondary source, to (among others) provide context for primary sources and put them in perspective. If we do it ourselves, we are engaging in original research. Crum375 (talk) 23:00, 22 January 2010 (UTC)

We use editorial judgement, enlightened by and based on sources, to decide what to put into articles. However, in the end there is always editorial discretion to decide which things to include. There is no basis in written policy or in practice for the idea that WP:V requires secondary sources for challenged material. — Carl (CBM · talk) 23:16, 22 January 2010 (UTC)

I do understand Slim's point about the notability of the source... but that is a WP:NPOV (and specifically WP:UNDUE) issue, and not a WP:NOR issue. Saying "Professor Smith says that every X is a Y (cite to where Smith says every X is a Y)" may well be giving Smith undue weight, but it isn't Original Research to note that Smith says it.

This is yet another reason why I support the idea of spinning PSTS off into its own Policy/Guideline page... there are issues that relate to PSTS that do not relate to the concept of "No Original Research", and they are out of place and confusing to people when discussed here. Blueboar (talk) 23:37, 22 January 2010 (UTC)

It is an NOR issue, because the Wikipedian is selecting primary sources, rather than selecting secondary sources who discuss them. We are a tertiary source. Our work should almost always be at secondary source level. I'm finding it quite disturbing that anyone is arguing that this is not a key part of OR. SlimVirgin ^{TALK contribs} 23:45, 22 January 2010 (UTC)

Slim... I am finding it disturbing that anyone would argue that the notability of a source is an NOR issue. I do not undersand how the mear selection of a source is any more original research than the selection of a different source. Please explain your position more. Blueboar (talk) 00:13, 23 January 2010 (UTC)

I gave you an example below. Jesus, animal advocate. Very notable source. Says some things that clearly put him in that category. According to you, I don't need a secondary source confirming that this is the correct interpretation of what he said, and that other people agree. According to you, Jesus just needs to be notable, and I only need to stick closely to what he says. But otherwise, it's fine to add him to various animal welfare or animal rights articles as a primary source. SlimVirgin ^{TALK contribs} 00:18, 23 January 2010 (UTC)

No, I am not saying that Jesus just needs to be notable... I am saying that if Jesus says X about animals, then it is not Original Research for us to include a simple descrive statement to the effect that he says X about animals. That does not mean that we should sescribe what Jesus says about animals (there are other policies and guidelines, such as WP:UNDUE, that apply) I am only saying that it is not an OR violation to do so. Blueboar (talk) 02:13, 23 January 2010 (UTC)

This is basic NOR, a classic example of it, adding what Jesus said to animal rights, animal welfare, cruelty to animals, and whatever else, using the New Testament as the source. What could be more notable than that? There is no violation of UNDUE in terms of the notability of the source. But there is a violation of the NOR policy, because only you are saying Jesus's views matter. You won't find him in any AR article, at least not any scholarly ones. Ditto animal welfare. Should we start adding what people think his views on homosexuality were to Gay? As I said, this would be classic OR, because it's a Wikipedian picking and choosing which primary sources to use without reference to secondary sources, turning WP into a secondary source instead of a tertiary one. This is absolutely at the core of the NOR concept. SlimVirgin ^{TALK contribs} 02:43, 23 January 2010 (UTC)

I'm finding it quite disturbing that you don't understand our standard practice of citing journal articles in all sorts of science articles. So maybe our disturbeds can offset. — Carl (CBM · talk) 23:49, 22 January 2010 (UTC)

Yes, I have seen it, and it's not good practice. This is a problem that's very specific to science articles, in my limited experience of them. SlimVirgin ^{TALK contribs} 23:52, 22 January 2010 (UTC)

When the general policy fails to describe the actual best practices employed by careful editors, the problem is with the policy, not with the editors. It's not as if I am referring to edits made by new users or people who are not sensitive to "original research". — Carl (CBM · talk) 23:55, 22 January 2010 (UTC)

Jesus often said things that made him sound like an animal rights advocate. We used to get lots of newbies trying to add that to articles, citing the New Testament. Is the only problem with this whether Jesus is notable, in your view, Blueboar? You're arguing that it's not OR? SlimVirgin ^{TALK contribs} 23:50, 22 January 2010 (UTC)

Re Blueboar: I agree it is an NPOV issue, of course. On the other hand, I have never seen "non-notable source" in a policy document and I don't think it has any meaning. "Notable" is a term of art for article topics. In any case, the Germanium example is better (in the math example I was trying to refer to a theorem that had been proved in the paper, but I see how it was easy to misread what I wrote in that example). — Carl (CBM · talk) 23:49, 22 January 2010 (UTC)

Slim... what I am saying is this: When it comes to the concept of No Original Research, it does not matter whether Jesus (or anyone else) is notable or not. If X says Y, then it is not Original Research for us to say that X says Y. The reason why it is not OR is because, in the situations I am talking about, noting that X says Y is a simple descriptive statement directly supported by the source, and as the policy already notes, even a primary source may be used to support a simple descriptive statement of this nature. Of course this does not mean that we must discuss the fact that X says Y, or even that we should (indeed it might even be that we shouldn't... there are, after all, other policies and guidelines, such as WP:UNDUE, that need to be considered)... But, it is not an OR violation to do so. Blueboar (talk) 02:13, 23 January 2010 (UTC)

The fact that X says Y is not Original Research, but inclusion of that material when it is a primary source is usually an example of Original Research. For example, Jesus said in Luke 19:27 "But these enemies of mine, who did not want me to reign over them, bring them here and slay them in my presence." Including that bald statement in the Capital punishment article would be OR. Jayjg ^(talk) 02:58, 24 January 2010 (UTC)

Actually, no it isn't. For example, it could be used to support a simple descriptive statement such as... "Capital punishment has been mentioned in many sources through antiquity, such as Babylonian Legal Codes <cite translation of Hammarabi>, Egyptian texts <cite Egytian text> and in the New Testiment of the Bible <cite Luke 19:27>.. As long as we stop there, we appropriately use the source to support a purely descriptive statement about the source. No OR involved. Blueboar (talk) 03:46, 24 January 2010 (UTC)

Still more examples

Carl asked me to look in here again to clarify my earlier comment that "an original research article is not a secondary source for its own conclusions", and asked me whether that would mean a straight-up newspaper article (say this Reuters piece) would be primary or secondary for its conclusion (in this case, that an arms deal may be imminent). Wow, am I the wrong guy to ask questions like this! I hardly ever edit articles on current events and have little feel for how policy applies in subtle current-event cases. But my kneejerk reaction is that the piece in question is a secondary source about the potential arms deal, as the reporter is not directly involved in the deal.

Turnabout is fair play, so I'd like to ask a question back. Here's a typical problem for articles I help edit. Although there is no cure for autism there are scores of autism therapies, most of them with no scientific evidence. New therapies seem to crop up once every month or two: there will be a primary source, or series of primary sources by one research group, that promote it, and often the therapies are too new to have been reviewed.

Here's a recent example of this sort of problem. A series of edits to Autism therapies created a new section Qigong Sensory Training that cited (among other things) Silva et al. 2009 (PMID 19708471), an original research paper that concludes that the severity of autism is reduced by a therapy based on the theory of traditional Chinese medicine that massage alters the body's energy field. This paper was published in The American Journal of Occupational Therapy, a peer-reviewed journal not noted for publishing autism research. Although the paper is too new to have been reviewed by secondary sources, it's published by a research group whose earlier paper on the same subject has one known reliable review, Levy & Hyman 2008 (PMID 18775371); this review says only "Sensory differences are frequently described by parents, but do not figure prominently in the DSM IV criteria for diagnosis of autism. Therapeutic approaches in this category include massage and Aroma Therapy." (Translation: "These therapies have no evidence vis a vis autism, as their studies are not measuring autism symptoms.") Would it be original research for Autism therapies to cite this new source in support of the claim "In the Qigong massage program, parents are trained to give their child a daily massage, under the theory that autism is due to an impairment of the sensory and autonomic nervous system. A 2009 controlled trial of 46 children found that it significantly reduced autistic behavior and improved social and language skills."? Eubulides (talk) 00:49, 23 January 2010 (UTC)

I don't feel it would be "original research", because the text is directly citing a reasonable source and not interpreting or synthesizing. On the other hand, it is not obvious that the source should be included, and the best people to decide that are people familiar with the article. If you feel the study is important enough to include, you can always hedge more, for example by replacing "found" with "concluded" or "suggested". It's really more of a question about whether the editors of the article feel that the inclusion of the material passes WP:UNDUE. Deciding that requires you to decide the "prominence" of this research in relation to the field, which is something that has to be done on a case by case basis. — Carl (CBM · talk) 00:57, 23 January 2010 (UTC)

(ec) Any paper like that is a primary source. To use them, we ideally need secondary sources who discuss them, and preferably academic sources, because these research teams send press releases to newspapers, and junior reporters write them up if they have nothing else to do. We could fill whole articles with lists of research discoveries. What's ideally needed are scholarly secondary sources who tell us that the research is worth mentioning. But CBM and others on this page disagree with me, so this is my interpretation of the policy.

On the autism page, if I were editing it, I might allow that primary source on its own if no one objected. If there was an objection, I would ask to see scholarly secondary sources, or if it's too new for that, then multiple high-quality newspapers who've written serious articles about it, not just brief mentions. SlimVirgin ^{TALK contribs} 00:57, 23 January 2010 (UTC)

But the question is not "should we make such a statement", but rather "if/when we reject this statement, are we rejecting it specifically because it violates OR"?

I don't think that Eubulides' example indicates original research: The statement doesn't add anything up or apply it to a context that isn't intended by the source itself. But I normally wouldn't include a statement like that -- because I thought it was unDUE emphasis on a WP:FRINGEy idea, not because it exceeded its sources. (In fact, it is a marvel of restraint compared to the sources.) WhatamIdoing (talk) 03:05, 23 January 2010 (UTC)

The problem with this discussion (the one above, not this portion) is that it assumes the content policies are entirely separable: "this isn't OR, it's UNDUE" etc. But they're not. The concepts each of them relies on are interdependent, which is why we introduce each one as needing to be read along with the others. The heart of the issue here is a Wikipedian deciding that primary source X matters, when it should be a secondary source who decides that. That's the idea that's being discussed here. UNDUE and FRINGE are built on that -- a Wikipedian placing undue emphasis on something secondary sources don't mention, or don't mention much. SlimVirgin ^{TALK contribs} 03:40, 23 January 2010 (UTC)

In the end, our own editorial discretion decides what belongs in articles and what does not. Sometimes primary sources are included despite not being mentioned in secondary sources. Sometimes, secondary sources are left out because editorial discretion says that including them at all would give them undue weight. The Wikipedian editors of an article are the ones who decide, as a group, which sources "matter" and which do not. Sources are a crucial guide, but each editor also brings their own personal experience, and between all of them a neutral version is hammered out.

The point of having separate policies is to differentiate between issues. WP:OR is about introducing new ideas that are not found in sources. WP:UNDUE is about deciding which sourced ideas should be included, and which should not be included. The point of "notability" is to decide what article topics should exist, independent of the actual content of the articles. These are distinct issues, which is why the policies are not merged. — Carl (CBM · talk) 03:50, 23 January 2010 (UTC)

I disagree. OR is not only about "introducing new ideas that are not found in sources". OR includes a situation where a Wikipedian decides that some obscure experimental paper, which no secondary source has found important or significant enough to interpret or write about, should be written about in Wikipedia. This activity, by an anonymous editor with no known credentials and with no editorial oversight by credentialed professionals, who decides that in his own professional opinion this primary source is valid and important, with no support from any secondary source, constitutes original research, and violates our core policies. The goal of Wikipedia, as a tertiary source, is to summarize existing secondary sources, while possibly adding details from primary sources. Unfortunately, many editors miss this point, and believe they are a secondary source, interpreting primary sources. This is a crucial point, which this policy attempts to address. Crum375 (talk) 04:34, 23 January 2010 (UTC)

Let me add another related aspect of OR based on primary sources. If uncredentialed anonymous Wikipedia editors take a primary source and highlight parts of it over others, or highlight one primary source over another, without support of a secondary source which provides us with this perspective, it would also constitute OR, despite the fact that all the material is individually sourced. The crucial point is that we must depend on secondary sources for context and perspective. If we are the ones picking and choosing which primary sources (or parts thereof) to use, instead of the secondary sources, we are engaged in original research. Crum375 (talk) 04:47, 23 January 2010 (UTC)

"OR includes a situation where a Wikipedian decides that some obscure experimental paper, which no secondary source has found important or significant enough to interpret or write about, should be written about in Wikipedia." — no, that is WP:UNDUE, which is a completely separate issue. Deciding which sources to use and how much weight to give them is "source based research" and is inherent in writing our articles. Indeed, the choice of which secondary sources to use has all the problems that you are attributing to the choice of primary sources, so it is not really about primary/secondary. Editors may use all resources available to them — primary sources, secondary sources, personal opinions, talk page discussions, etc., when deciding what to include in an article. But not all of this may actually be inserted into the article itself. — Carl (CBM · talk) 05:15, 23 January 2010 (UTC)

WP:UNDUE, as presented inside NPOV, has to do with neutrality, when there is some kind of dispute, and different points of view need to be fairly represented, or fringe minorities need to be ignored. In this situation, there may be no dispute, and no neutrality issues at all. Except, an anonymous Wikipedia editor, decides that his own unknown credentials and personal knowledge allow him to promote one primary source over another, or a part thereof. Or, he may decide that an obscure research paper which no secondary source considered important enough to describe or interpret, is significant in his opinion, based on his own personal knowledge. This is OR, and not UNDUE, not a neutrality issue. Crum375 (talk) 05:25, 23 January 2010 (UTC)

No; WP:UNDUE is about presenting views in proportion to their "prominence" in the literature. As WP:UNDUE says, "Undue weight applies to more than just viewpoints. An article should not give undue weight to any aspects of the subject but should strive to treat each aspect with a weight appropriate to its significance to the subject." It often happens that there is consensus that a primary source is sufficiently significant to mention in an ariticle, even though the primary source is not mentioned in any secondary sources.

In any case, accurately citing an obscure research paper is not "original research", it is simply sourced-based research. It may be that the "prominence" of the paper is low enough that it should not be mentioned at all (e.g. a fringe view), but it is not a matter of "orginality". — Carl (CBM · talk) 05:35, 23 January 2010 (UTC)

UNDUE, as part of NPOV, focuses on neutrality issues. There is no neutrality issue when an editor decides that some primary source, never reviewed by any secondary source, should be presented on WP. Similarly if an editor decides to highlight a part of a primary source, with no guidance for perspective from a secondary source. The main point is that the editor is relying on his own uncredentialed personal expertise to decide significance, or importance, of primary material. This is original research, not (necessarily) POV pushing. WP is a tertiary source, among other reasons, because its editors are entrusted to summarize what secondary sources are saying about primary sources, not to analyze primary sources, or their significance, on their own, based on their personal knowledge. Crum375 (talk) 05:46, 23 January 2010 (UTC)

WP:NOR applies to the material that is actually added to articles. It does not apply to the mental processes by which editors create articles. As Blueboar says above:

If X says Y, then it is not Original Research for us to say that X says Y. The reason why it is not OR is because, in the situations I am talking about, noting that X says Y is a simple descriptive statement directly supported by the source, and as the policy already notes, even a primary source may be used to support a simple descriptive statement of this nature.

Maybe he has explained things better than I have. — Carl (CBM · talk) 05:54, 23 January 2010 (UTC)

Yes, OR applies to material which is added to articles. Therefore, if we add that "X said Y", where X is a primary source which no secondary source has deemed important or significant enough to mention, we are using our own personal knowledge to decide that "X said Y" is important and significant and should be added to the article. By adding it to the article, we are effectively telling the readers, "this is important, this is significant", without any source for that effective statement. Similarly when we highlight one part of a primary source over another, again, we are implying "this part is important, that part isn't", with no source for that implication. We must rely on secondary sources to add perspective and context for information from primary sources; if we do so ourselves, based on our own personal knowledge, we are engaging in OR. Crum375 (talk) 06:12, 23 January 2010 (UTC)

So Crum375, what do you think of using personal research to ignore sources? For example, I am revising the Tropical year article because it (correctly) has a {{Refimprove}} template. So with some pointers from another editor, I looked in a well known secondary source, the Explanatory Supplement to the Ephemeris, and it gives an equation for the tropical year that I would like to use. Unfortunately, it labels the equation as the "equinox-to-equinox" tropical year, and I happen to know by doing some calculus with the other equations in that book that it is really the "mean" tropical year. So, based on my personal research, I'm just going to ignore that book and try to find a better one. --Jc3s5h (talk) 06:51, 23 January 2010 (UTC)

(Hey, by sheer coincidence I read that article for the first time yesterday, what are the odds for that?) I think every editor must use his common sense and knowledge in the background and the talk pages, but not necessarily in article space. This is similar to police detectives using all available resources and knowledge about a criminal case to help them find and convict the real culprits, though not all evidence is admissible in court. In your specific case, if you believe (or strongly suspect) a primary source (not mentioned by any secondary source) is wrong, it's easy: ignore it (after talk page discussion). If it's a highly reliable, much cited, secondary source which you believe is wrong, and you have no reference for it being wrong (except your own math), and it's not a contentious issue, then you could say: X says Y[note 3]. In the footnote, you can briefly explain that although Y is well accepted, the math seems wrong. I would be hesitant to do even that (in that situation), unless I got talk page consensus for it from people I trusted. As another possibility, you could leave an invisible edit note describing your perceived discrepancy, perhaps pointing to the relevant talk page discussion of the issue. So as bottom line, we must use our common sense. Crum375 (talk) 13:30, 23 January 2010 (UTC)

A simple way to test Crum375, Eubulides, and SlimVirgin's interpretation of the original research policy, which is that it is original research to give "primary sources" undue weight, as opposed to Carl/Blueboar/WhatamIdoing/myself's interpretation that original research is citing a source which doesn't really say what you think it says, is to look at the WP:NOR page. Surely, if it was part of the NOR policy that original research includes "hyping up" obscure information, the page would reflect that? Yet I do not see anything which really says that. The closest thing is the statement that "Wikipedia articles should be based on reliable, published secondary sources and, to a lesser extent, on tertiary sources" ([3]). Yet this definition of primary sources was until recently extremely narrow and did not include research articles. A couple months ago the closest thing was "published notes of laboratory and field experiments..." [4] This has been extremely recently broadened by Tim Vickers into "similarly, a scientific paper is a primary source about the experiments performed by the authors and their conclusions" [5]. As I've pointed out before, if a scientific paper and its conclusions are primary sources, then in a relevant sense so are most pieces of journalism which tie together data in the form of various documents to come up with original conclusions, and so are reviews which analyze their raw data (research articles) to come up with original conclusions. Even ignoring the fact that there's no real discussion in WP:NOR of using secondary sources to evaluate the notability of a particular source for inclusion, implicit in Crum375/SlimVirgin's argument that a secondary source is necessary to establish the "notability" is that looking at the secondary coverage is the main and most relevant way to figure out if something merits inclusion. That's not right because just about everything generates some secondary coverage, and many trivial things generate the most secondary coverage. II | (t - c) 07:53, 23 January 2010 (UTC)

Taken to extreme, the "everything's NOR" argument is pretty silly: A new treatment has been tested for a rare disease. Exactly one publication mentions it -- a primary source on a small clinical trial. Does anyone want to go on record as saying that a simple descriptive statement like "____ has been tested for this rare disease" violates WP:NOR? Does anyone think that any rational person would consider a potential treatment to be unrelated to the disease it's supposed to be treating, on the rather credentialist grounds that the fact hasn't (yet) been anointed by a secondary source?

Depending on the circumstances, I might consider it undue weight on a minor fact, but in writing a sentence like that, the Wikipedia isn't inventing any "original" ideas; s/he is merely accurately describing a verifiable fact. WhatamIdoing (talk) 08:35, 23 January 2010 (UTC)

Incidentally, there is also a question of how much actual laboratory work is done by technicians separate from the author. See, e.g. The Invisible Technician, similarly, What Does It Mean To Be An Author? The Intersection of Credit, Contribution and Collaboration in Science which on page 7 notes that technicians are historically not identified as coauthors. II | (t - c) 11:30, 23 January 2010 (UTC)

To Crum375: first of all, there's obviously a dispute here, and so saying there isn't and reverting [6] smacks of bad-faith and an odd sense of entitlement. Second, please engage my above point that your interpretation of NOR is not in the WP:NOR page. If it is, where is this interpretation? Since the interpretation isn't in the page, it's not part of the policy. This is something a group is trying to add now - and it's looking like it's not going to succeed, regardless of aggressive tactics. II | (t - c) 19:37, 23 January 2010 (UTC)

There is no "entitlement", because I didn't write any of the material you removed, and very little of the policy as a whole. There is also no dispute about it; the issues being addressed here are not (afaict) the ones in what you removed. As far as "my interpretation of OR", what I describe here is my understanding of OR, as written and practiced on WP. If you believe something in the written policy contradicts it, please let me know. Crum375 (talk) 19:55, 23 January 2010 (UTC)

The entitlement comes in the fact that you reverted without bothering to engage the person you're reverting; it suggests you think it is beneath you to actually discuss with certain people. Standard practice, as written in policy, is to discuss before reverting or at least concurrently (see Wikipedia:3RR#How_experienced_editors_avoid_being_dragged_into_edit_wars). The dispute is actually about this particular sentence to a large degree; the sentence you reintroduced was introduced by Tim a couple days ago [7] and is different than the longstanding language which I restored. This dispute is partly about classifying research articles as primary and then saying we can't use their interpretive conclusions - see, for example, CBM's earlier note that "the parts of the journals that we use for interpretation and analysis are secondary sources for us. This is fine because the same source can be both a primary source (for data) and a secondary source (for analysis of that data)..." As far as your interpretation of NOR, I think that your interpretation does contradict the policy as it isn't in the WP:NOR. Thus, it's not part of the WP:NOR policy - it is in your mind. I can't point to something in the written policy which says your interpretation isn't valid because the policy is largely positive (OR is X), not negative (OR is not Y, Z). Thus, the burden is on you to provide something in WP:NOR which supports your interpretation. If you want to revert again, please respond before. Please allow me to respond and bring up anything which I don't think you've really engaged. II | (t - c) 20:14, 23 January 2010 (UTC)

I agree with Crum375, nobody in this discussion has objected to the data and conclusions of a scientific paper being described as a primary source. Jumping into the middle of an ongoing discussion and reverting something we all agree on isn't really very helpful. Tim Vickers (talk) 20:21, 23 January 2010 (UTC)

not just science articles

The issues that we are discussing may be highlighted by our difficulty in classifying Scientific materials... but they relate to non-scientific materials as well. Consider the following from the field of History... Papal Bulls... These pronouncements are normally classified by historians as Primary sources. Yet it is not at all uncommon for the author (a Pope) to include his interpretation of previous Papal Bulls and Church doctrine in making such pronouncements. In other words, these Primary sources often contain analysis and interpretation of other Primary sources. So is such a document really a Primary source?

As I see it... that question is actually irrelevant. It does not really matter whether a given Papal Bull is primary or secondary. There are appropriate ways to use the source, and there are inappropriate ways to use the source. OR is determined by HOW we use the source, not the fact THAT we use it. Blueboar (talk) 15:38, 23 January 2010 (UTC)

Anybody, whether scientist, historian or reporter, writing about another document, is a secondary source for that document. However, not all secondary sources are created equal, and this is where our common sense and judgment as editors come in. The farther the secondary source is from its primary, the more reliable it becomes, as its ability to form an objective perspective increases. So if I am writing about my own previous work, or my colleagues', it would be less valuable (for perspective) than someone else, unrelated to us, doing it. Of course the qualifications and expertise of each source have to be considered also, and added into the equation. As Wikipedia editors, we need context and perspective for anything we write about, and a primary source gives us very little, if any, while a secondary source can provide valuable input, but it too must be scrutinized for distance from the data it reports on. And "distance" can be measured in terms of time, or independence from the original writer(s), or any other parameter which adds perspective and objectivity. Crum375 (talk) 16:19, 23 January 2010 (UTC)

Absolutely... but the question is... what has any of this to do with the concept of "No original research"? Blueboar (talk) 17:56, 23 January 2010 (UTC)

NOR is directly related to the above. We all understand the simple application of NOR: if there is no source that the moon is made of granola, and we add it to Wikipedia based on our own personal knowledge, it's OR. But the more subtle type of OR, which many editors misunderstand, has to do with the proper use of existing sources, by adding implications which have never been published, based on our own personal knowledge. So if a court record (primary source) said that Professor X was caught driving drunk (and no secondary source mentions it), and a diligent Wikipedian finds it and adds it to X's article, that editor is using his personal knowledge that that event is important and relevant to the article. Similarly for picking primary research papers which are not mentioned by any secondary sources, or highlighting parts of primary sources while excluding others, etc. The point is that in order to understand when we are adding our original knowledge by implication, i.e. performing "implied" original research, we need to understand that for obtaining perspective about an issue, we need a secondary source, and we need to understand the crucial issue that only secondary source provides us with perspective for a primary source, e.g. a reliable sourcing for what's important and relevant, and what's not. So understanding NOR and the classification of sources are closely interrelated. Crum375 (talk) 19:00, 23 January 2010 (UTC)

No, that court example is not OR. It is likely a violation of WP:UNDUE, WP:BLP, and other policies. But adding a sentence "X was convicted of drunk driving" along with a full citation to a published court record is not original research, it is simply source-based research. We expect editors to go out, find sources, and use them to write articles. This is only original research if the content of the article goes beyond the content of the sources. The importance of the source is irrelevant to the issue of OR. — Carl (CBM · talk) 19:06, 23 January 2010 (UTC)

You are missing the point. Yes, it also violates UNDUE, BLP, etc., but all of that stems from the editor's "personal knowledge" that that primary source is important and relevant to the article, vs. some other item which said that Prof. X's daughter's soccer team won a game. In general, it is original research to use our own personal knowledge to add information to Wikipedia without proper sourcing. That can be to write that the moon is made of granola, or to imply, based on a primary paper we have just read, that some chemical compound which was found there is important (e.g. vs. another). In the latter case, it is our own personal knowledge that the information in the primary paper is important and relevant that we are adding to the article, but to do so properly (in a contentious case) we need a secondary source to put that paper and its implications in perspective and tell us that it's important and why. To do so by relying on our own personal knowledge would convert us into a secondary source. Providing perspective, esp. in contentious situations, even implicitly, requires a source, or else it's OR. Crum375 (talk) 19:21, 23 January 2010 (UTC)

Crum375 - see my above request for engagement. II | (t - c) 19:37, 23 January 2010 (UTC)

(unindent) A US court case is a bad example because a court verdict can be argued to be a secondary source. The jury evaluated the testimony (which, because of the availability of official transcripts, meet this policy's definition of a primary source) together with documentary evidence to reach a conclusion. The (usually written) verdict was further reviewed by the judge, in conjunction with the written statutes and written court opinions before reaching the final decision. --Jc3s5h (talk) 19:29, 23 January 2010 (UTC)

No, a court ruling is a primary source for the specific case being litigated, since the judge personally reviewed the evidence, the jury verdict (if any) and the applicable laws, decided the outcome, and wrote the ruling. A law journal, or even a newspaper, reporting on the ruling would be a secondary source. You need distance and perspective to be a secondary source, and if you were involved in creating the published data, or controlled its outcome, you don't have that distance or perspective. Crum375 (talk) 19:39, 23 January 2010 (UTC)

I acknowledge that the court verdict could be argued either way, but not for the reason stated by Crum375. Publications which are based on synthesizing, organizing, and evaluating other sources are secondary sources. Only when a new conclusion that cannot be derived from the sources without a significant new insight is it a primary conclusion. Einstein's special theory of relativity would be an example of a significant new conclusion; even though he didn't do any experiments, his conclusion went well beyond what anyone else was able to derive from the existing sources. A source that makes choice among possible decisions, each of which is a plausible choice given the sources, is a secondary source. For example, a movie review is a secondary source; in almost all cases, "I liked it" or "I didn't like it" are plausible outcomes for the review given the contents of the movie. --Jc3s5h (talk) 19:55, 23 January 2010 (UTC)

The argument that the court case is a primary source would be that the judge and jury observed the demeanor of the witnesses, not just the words written in the transcript. --Jc3s5h (talk) 19:55, 23 January 2010 (UTC)

In the court example, if no information is added to the article that does not appear in the sources cited, then there is no "original research", because no information has been added to the article without proper sourcing.

The mental processes that we use to decide what sourced information to add to articles do not themselves have to be sourced, which is why we don't require sources for talk page conversations. Your arguments about "implicitly providing perspective" by choosing which sources to use goes far beyond every written policy and are is odds with the common practices of many experienced, careful editors — and in that sense it is simply wrong. Researching sources and then deciding which of them to use is the essence of source-based research, and it is exactly what all of our policies tell editors to do. — Carl (CBM · talk) 19:33, 23 January 2010 (UTC)

"Researching sources and then deciding which of them to use is the essence of source-based research": in general, you are correct. But if the sources are primary, and no secondary source has reviewed them and put them in perspective for us, by deciding which ones to use we are effectively promoting some over others. In non-contentious cases, it would be OK, but in other cases, esp. when challenged, that effective statement, "this bit is important, that bit is not", would have to be supported by a secondary source, or else we'd be engaging in original research. Crum375 (talk) 20:03, 23 January 2010 (UTC)

The "contentious" issue is a red herring: there is no text in NOR that says "primary sources are OK, unless the issue is contentious". In fact, the word "contentious" does not appear in the NOR policy at all. This makes sense, because "contentious" is an issue with neutrality, which is covered by NPOV, not by this policy. The NOR policy allows primary sources to be used to cite descriptive claims, with no restriction to "non-contentious" claims. — Carl (CBM · talk) 20:16, 23 January 2010 (UTC)

"Contentious" just means that we need to be more careful, because every word counts. In a contentious situation, it is much more likely that some editor will use a primary source improperly, to promote his "understanding". So when there is some dispute, or BLP issues, we need to be extra alert not to imply things about material in primary sources which are not supported by secondary sources. Crum375 (talk) 20:58, 23 January 2010 (UTC)

We are not talking about adding new implications. We are talking about using primary sources for descriptive claims such as "Germanium has been detected in the atmosphere of Jupiter", which the NOR policy goes out of its way to permit. — Carl (CBM · talk) 21:06, 23 January 2010 (UTC)

If that statement is non-controversial, then yes, adding a simple descriptive claim based on a primary source would probably be OK. But if all secondary sources say X, and you have a primary source disputing X, you'll need to provide a secondary source to do any more than a mere mention of it. Or possibly put it in a footnote, pending availability of an appropriate secondary source. Crum375 (talk) 21:17, 23 January 2010 (UTC)

As I was saying, NOR has no language whatsoever about different handling of "controversial" topics (that's what NPOV is for). There is no "probably" about it: primary sources are permitted, by NOR, to be used to make descriptive claims. This creeping "probably" is why I am engaging in this discussion still.

It's not true that primary sources are actually bad, but are allowed if nobody complains. For certain types of claims, as explained in NOR, primary sources are perfectly valid and acceptable from the NOR perspective, with no further caveats and no "probably" about it. In some cases there will be other policies that apply, but NOR is not a catch-all for improper use of sources. — Carl (CBM · talk) 21:31, 23 January 2010 (UTC)

Primary sources can be used "descriptively". But we don't really define what that means, and leave it to editorial judgment. The way I see it, we may not interpret them in any way, and may not use them selectively (highlight one part and downplay another). If we must summarize them, the summary must be "descriptive", i.e. if there is some dispute about what the source says exactly, we need to be extra cautious not to introduce bias or our own interpretation in the summary. This is why a secondary source is so important: it provides us with a reliable source which has interpreted the primary material and put it in perspective for us. Crum375 (talk) 21:47, 23 January 2010 (UTC)

I definitely have to agree with Carl here. Going to the court house and searching a public record is no different than going to a library or accessing a website on the internet. Yes, if we go beyond what the court record tells us, then we get into OR territory... but simply saying "Professor X was conficted of drunk driving on Jan. 2 1999 (cite to court record)" is not OR.... it is sourced based research. There probably are a whole bunch of good reasons why we should not include that simple statement, but WP:NOR isn't one of them. Blueboar (talk) 19:36, 23 January 2010 (UTC)

I disagree. Yes, the actual data we obtain from a reliable source is not in question on its own. It is its inclusion in the article which is the key. The mere fact we include it conveys importance, above and beyond many other (potentially) available sources and related items. So by including it, we are effectively saying, "it's important", with no secondary source to back up that implied statement. Similarly if we pick one part out of a primary source and not another part: we are basing that selectivity on our own personal knowledge, "one is important, the other is not (trust me)". In a contentious situation, we need a secondary source to provide perspective and support for that effective statement, or else we are relying on our own personal knowledge for it. Crum375 (talk) 19:46, 23 January 2010 (UTC)

Crum, you seem to be saying that we may never cite a primary source... but the NOR policy explicitly states that reliable primary sources may be used in Wikipedia. So can you explain under what circumstances you think it is appropriate to cite a primary source? Blueboar (talk) 20:37, 23 January 2010 (UTC)

No, reliable primary sources are fine, and I personally use them a lot (in many areas they are better and more accurate for basic information than the secondary ones). The point is that we need to be extremely careful when using them so as not to imply things they don't say, and this is very tricky. For non-contentious non-BLP issues it's easier, and less critical, but in contentious situations, and esp. BLP, one must be extra alert. Of course, the article should be based on at least one secondary source, but for adding details we can use reliable primary sources carefully. Crum375 (talk) 20:44, 23 January 2010 (UTC)

Just to add, when I use primary sources, I prefer to quote their own summaries of themselves (if there are any short ones) vs. my own, to minimize the risk of interpretation or selectivity. And in general, primaries are best used when they are already mentioned by the secondaries, in which case we can either just include the reference to them, or if we need to expand beyond the secondary and rely on them, we should try not to add anything which could be perceived as advancing a position. Crum375 (talk) 20:53, 23 January 2010 (UTC)

It's the last part ("primaries are best used when they are already mentioned by the secondaries") that is not anywhere NOR. If we required all our primary sources to also be in secondary sources, we could just cite the secondary sources, and there would be no need to cite the primary sources in the first place. The NOR policy simply says that primary sources can be used to source descriptive claims, etc. — Carl (CBM · talk) 21:05, 23 January 2010 (UTC)

This discussion is unfortunately too fragmented, and I have already replied to this above. Primaries can be used for their "raw data", when we are not adding our own interpretation to them, i.e. used "descriptively". This is very hard to do without introducing our own interpretation, so relying on a secondary source for this perspective is much preferred. Crum375 (talk) 21:57, 23 January 2010 (UTC)

NOR does not say that secondary sources are preferred over primary for descriptive claims. In practice it is not actually very difficult to cite a primary source without introducing an interpretation. Examples: "Germanium has been detected in the atmosphere of Jupiter [1]", "The United States Treasury collected approximately $2.52 trillion in tax revenue in FY 2008 [2]", "Wiles' paper on Fermat's last theorem was published in Annals of Mathematics in 1995". — Carl (CBM · talk) 23:25, 23 January 2010 (UTC)

Crum... I am glad that you understand that primary sources are fine, but must be used with caution. so how is saying "Professor X was conficted of drunk driving on Jan. 2 1999 (cite to court record)" not a simple descriptive statement?... how does it go beyond the source? How is it OR? Blueboar (talk) 23:39, 23 January 2010 (UTC)

Yes, it is a descriptive statement, but there may be many other such published tidbits about Prof. X in numerous primary records, in court rulings, other government records, and other archives. If a Wikipedian decides that, based on his personal knowledge, this particular court record is important and significant for Prof. X's article, while many others are not, despite the fact that no secondary source has bothered to mention it, that wiki editor would be engaged in original research, because he is the only one supplying the perspective, i.e. that this incident is important in Prof. X's life or career, and the others are not. Now, we do have BLP rules and other policies, but the essential point remains the same for all types of articles: when we as anonymous uncredentialed Wikipedians promote one primary source over another, or one part of a primary report over another, we are implying that the selected material is more important than the others. If we have no secondary source to base that conclusion on, we are violating NOR. Crum375 (talk) 03:17, 24 January 2010 (UTC)

I'm sorry, but I can not accept the argument that deciding what should and should not be discussed in an article violates WP:NOR. No... I can accept challenging something like the Prof. X conviction because it is TRIVIA... I can accept challenging it under UNDUE... and I am sure there are other policies and guidelines that apply... but not NOR.

And I definitely don't think the fact that a bit of information comes from a primary source matters in determining whether the bit of information is "important". Consider another form of Primary source that an editor might find in a records office and use in an article... a death certificate. Is it a NOR violation to say that someone died on a given date, and to cite the person's death certificate? Surely the person's death date is important to note in the article... the fact that we got this information from a primary source does not change that fact. Blueboar (talk) 04:05, 24 January 2010 (UTC)

Policy wording and rules cannot replace common sense. It is certainly OR to use our own personal knowledge to interpret or provide perspective for a primary source which no secondary source has reviewed. But if, as you say, we have a primary source which says someone died on date X (and lets assume we have a secondary source telling us the person is dead), and we are fairly sure this is the correct person and the date makes sense, we may use it, unless it conflicts with secondary sources (in which case we may note the conflict in a footnote, for example). The point is that primary sources which have not been reviewed by secondary sources are a danger zone, where we have to tread very carefully. If we start adding our own personal knowledge into the article to either interpret these sources or place them in perspective ourselves, we'd be violating NOR. If we are careful not to add anything controversial, e.g. selectively pick and choose unreviewed primary sources or parts thereof to advance a position, we'd generally be OK. Crum375 (talk) 04:55, 24 January 2010 (UTC)

actually, I do see the issue here, because this happens all the time on fringe articles. For example, you'll have some article on a (really weird) theory R which was promoted by professor X, and some editor will come along to edit in some scandalous personal information about X (affairs, alcoholism, institutionalization, arrests - whatever applies) or edit in some "it's a stupid idea" comment from some person who would otherwise not be considered a reliable source on the topic (QuackWatch, skeptics journals, and Marvin Gardner are three of the usual suspects here). The obvious intent is to leverage the promoter's scandals or the reputation of a noteworthy but unrelated individual as a back-handed means of critiquing the original weird theory. but that's pure OR: why should the scandals of a promoter matter to the theory? why should just any old scientist be qualified to give a critique? non-neutral information cannot be treated as innocuous. I mean, if you walk up to a husband and wife at a cocktail party and say "Oh, I saw you and your girlfriend Julie at the theater the other day; how was the movie?", you cannot honestly claim that you were just presenting a fact.

really, I think this whole primary/secondary/tertiary thing has become a major source of wikilawyering. the main point is that we want to present information in a way that is not misleading. Whether primary or secondary, sources need to be presented clearly and truly, and need to be balanced with respect to their prominence and applicability to the subject. --Ludwigs2 05:32, 24 January 2010 (UTC)

Re: adding back-handed critiques by mentioning scandals... I completely agree that doing this is not acceptable, but I disagree with your contention that that the reason it is unacceptable is that it is OR. Back-handed critics are a form of POV coatracking, not a form of OR. (What is funny is that trying to deal with coatracking issues in the NOR policy is itself coatracking.) Blueboar (talk) 17:07, 24 January 2010 (UTC)

well, I'm not sure I'd personally file it under coatrack, but I do see your point. similar issues. --Ludwigs2 06:28, 25 January 2010 (UTC)

Of course, the "court record" example is a little extreme. The more realistic example of the sort of primary source we are talking about here is a peer-reviewed journal article that is widely available to anyone with university library access. — Carl (CBM · talk) 19:40, 23 January 2010 (UTC)

An OI

It's my contention here that File:Nsw_patriots_colours.svg and it's peers are OI would you agree? Gnevin (talk) 20:45, 22 January 2010 (UTC)

"Comparison of" articles

Ttiotsw (talk · contribs) seems to think NOR prohibits comparisons between subjects in Wikipedia articles unless one of the article's editors cites a reliable source that has already made each comparison described in the article. See Talk:Conservapedia#Original research. If comparison is deemed prohibited synthesis, it would appear that this policy justifies deletion of most of the comparison pages on Wikipedia. Is this the case? --Damian Yerrick (talk | stalk) 22:12, 25 January 2010 (UTC)

If a reliable source has already mentioned the item we are comparing to in relation to the article topic, and the comparison is trivial and non-controversial (i.e. not "advancing a position"), then I see no problem with it. But if it's controversial, or creates a new interpretation or implication, or no reliable source mentions the compared item in relation to the article subject, then it could violate OR. Crum375 (talk) 22:26, 25 January 2010 (UTC)

In addition, if it's a simple and non-controversial numerical comparison, such as the number of casualties resulting from an earthquake, or the height of a building, we generally allow it. Crum375 (talk) 22:29, 25 January 2010 (UTC)

That's correct. You can't take something from Smith and something from Jones and compare them, Smith (or Jones) needs to do the comparison. See Wikipedia:Articles for deletion/Comparison between Roman and Han Empires (2nd nomination) which ended with a tiny stub and the article being worked on in the article incubator at Wikipedia:Article Incubator/Comparison between Roman and Han Empires. Dougweller (talk) 22:36, 25 January 2010 (UTC)

Doug, if you'll expand your statement to allow a third person, Williams, to make the comparison of Smith and Jones, then that's exactly what I think. Wikipedia editors shouldn't be declaring that Smith and Jones ought to be compared, but they can describe someone else's comparison. WhatamIdoing (talk) 22:55, 25 January 2010 (UTC)

So under this interpretation, how do Comparison of HTML editors and other "big table with green means yes and red means no" articles get a free pass under SYNTH? As I understand it, for each feature, you'd need to have a single reliable source stating that HTML editor A has a feature and HTML editor B lacks it. --Damian Yerrick (talk | stalk) 23:03, 25 January 2010 (UTC)

No, if the features or parameters being compared are clear-cut, they would be similar to the numerical comparison I mentioned above, esp. if there is no controversy about them. If it's more nuanced, like "ease of use", then yes, they would need a reliable source comparing them. Crum375 (talk) 23:19, 25 January 2010 (UTC)

With Comparison of HTML editors there is clear grounds for comparison, either because sources make are making the comparisons explicitly, or sources are (at least) pointing out that the elements are in competition with each other or in some other way designed to be comparable. the real problem comes when the act of comparison involves a de facto assertion by an editor that a comparison should be made. I just ran across this problem on Comparison of the Han and Roman empires - the two entities seem comparable, but there are very few sources that actually do the comparison, so the entire article came down to an editor juxtaposing research about one against research about the other. very subtle form of synth. --Ludwigs2 23:53, 25 January 2010 (UTC)

Wikipeidia is not Consumer Reports or a buyer's guide... I don't think it is not our job to compare competing products or programs. As for OR... I agree with the idea that generally you need a source that has made the comparison. We would have to be very very careful in making any exceptions here. Blueboar (talk) 01:06, 26 January 2010 (UTC)

I'll admit COI here: I find it very useful as a reader/consumer to have comparisons such as these on WP, esp. for computer related subjects. But of course we still need to carefully abide by NOR, so only straightforward clear-cut features should be compared, and each has to come from reliable source individually. The more nuanced comparisons ("better", "easier", etc.), or anything controversial, all have to come from a reliable sources doing the comparison for us. Crum375 (talk) 01:19, 26 January 2010 (UTC)

Crum, what you are talking about makes me very uneasy. I think such comparisons are OR... but I could see that they might constitute something that rates a blanket IAR invocation. I do understand your comment about these pages being useful... but useful and encyclopedic are not the same thing. This is why I am also not at all sure that Wikipedia is the right venue for such comparisons. Has anyone talked to the Wikimedia folks about creating a sister project for this sort of stuff? Blueboar (talk) 01:34, 26 January 2010 (UTC)

I think the operative phrase is "advancing a position". As long as we are just taking reliably published tidbits and putting them side by side in tabular form, without creating or implying anything new by that comparison beyond convenience for our readers, we're OK, and we provide a huge service to people actually wanting to properly learn (and compare) these subjects. Once we start creating new information, or new implications, or otherwise treading into disputed or controversial areas, we need to strictly rely on published reliable sources for the comparison. Yes, maybe there can be a dedicated wiki for it someday, but for the time being WP is doing a fairly good job. But we need to be sure we stay within NOR bounds. Crum375 (talk) 01:45, 26 January 2010 (UTC)

And to clarify, this is not just for consumer-related subjects. Here is a list of buildings which clearly compares items. The key again is that as long as no position is being advanced, the compared parameters are clear-cut and well-defined, and reliable sources are individually used for each item, the comparison is merely a convenience for our readers, not a way to push some subtle agenda. Crum375 (talk) 02:21, 26 January 2010 (UTC)

Hmmm... I think the list of buildings article is boarderline on the OR... but more encyclopedic. Another thing that may make the building article a bit more acceptable to me is that it uses "List of" in the title instead of using "Comparison of" ... To my mind, the word "list" does not invite us to do anything with the information listed. It is simply an organization of information. The word "comparison" however, does ask us to do something. It asks us to compare the items. There is a value judgement that is at least implied when you are asked to compare things (which item is faster vs slower, taller vs shorter, more expensive vs cheaper, more sophisticated vs. less sophisticated, etc.). Blueboar (talk) 04:33, 26 January 2010 (UTC)

I think any table of parameters invites comparison, so clearly this would apply to any information presented in tabular form (unless the complete table is published in a reliable source). So the question is: is it OR to compare? (I am assuming all individual data points are appropriately sourced.) My answer is that it's generally OK, as long as we meet the following criteria: a) the compared parameters must be simple and objective (e.g. building's height, software's vendor, engine's horsepower, disaster's fatalities); b) any controversial comparison must be specifically sourced; and c) the comparison should not create or imply new information. Bottom line: we may not "advance a position" (without a specific reliable source), but we may present reliably sourced, non-controversial information side by side for our readers' convenience. We need to be careful to avoid creating any unsourced information or implication. Crum375 (talk) 05:10, 26 January 2010 (UTC)

If it's established that this parameter is commonly compared in this field, then I don't think that you are required to omit the information for some items in a list simply because you can't find a source that explicitly compares that specific feature for that specific product.

For example: Software reviews routinely compare the price: "Software 1 is free, and Software 2 is not". Let's say that you've got a couple of good sources, and between them, these sources compare the price of ten of them, but not the other two. (This isn't at all uncommon for products that aren't actually on the market at the time the review is written.)

I don't think that NOR requires you to omit the verifiable price of the remaining two, just because it didn't happen to be mentioned in the sources that you're working with. A plain, descriptive statement from a primary source will be acceptable here, because you're not inventing the idea that prices should be compared.

Similarly, if your 'comparison of software' sources say that a piece of software runs only on Windows, and it later becomes available for Mac or Linux, then I don't think that updating the information in the table is an improper WP:SYNTH. WhatamIdoing (talk) 20:04, 28 January 2010 (UTC)

I wondered where Damian Yerrick had wandered off to as they didn't bother to use the talk page for the relevant article !. Just to clarify the misrepresentation. I said that it was a WP:COI for a Wikipedia editor to simply compare Conservapedia policies with Wikipedia. Damian Yerrick said here that certain issues related to the policies of Conservapedia were "Like Wikipedia" and I reverted this here stating that it was WP:OR to compare the two policies unless a WP:RS said this. In the talk section Damian Yerrick went a bit wobbly with this edit here which seems bring alive some old talk innuendo that and implies that I am in some way related to something called "RW" (RationalWiki ?). I very clearly stated here that "As an editor on WP then you have a clear WP:COI to compare the policies on a system that you are an editor of with the competitor system. Again, let someone else comment about this commonality of policy. Then you can add that in and ref that 3rd party.". I also stated very clearly that I have never posted to CP (Conservapedia) nor RW (RationalWiki) but some other editor reverted my clarification so that is irrelevant here anyway. Damian Yerrick hasn't bothered to reply in the relevant talk section but seems to have started to WP:FORUMSHOP policies. So basically folks you've been wasting your time here in replying to Damian Yerrick because they have misrepresented the situation and there is no issue that isn't already handled by WP:OR, WP:RS, WP:COI, oh and, WP:AGF Ttiotsw (talk) 09:09, 27 January 2010 (UTC)

The only reason I put the discussion in an existing section that mentioned RationalWiki is that the existing section was titled "original research", and "original research" was the rationale for the revert. As for forum shopping, that's a response to WP:OTHERSTUFF. I always thought that the right forum for a discussion about a policy interpretation that could result in dozens of AFDs was pages related to the policy, not pages related to a single article. I didn't want to illustrate the point disruptively by copy-and-pasting the issue into dozens of affected articles' talk pages. So I brought the discussion here because I didn't think the policy page was completely clear about what constitutes "advancing a position". The policy page mentions "A and B, therefore C" as the form for synthesis. The case of Conservapedia matches this template: "CP has ENGVAR and WP has ENGVAR, therefore the policies of CP and WP agree on this point." But then so do the comparisons put forth in red-and-green-cells articles. If the COI issue is that Wikipedia is one of the rows in the table, I can probably find Wikipedia in one of those articles too. Other stuff still exists, and I want to resolve the status of other stuff. --Damian Yerrick (talk | stalk) 21:24, 29 January 2010 (UTC)