Wikipedia talk:No original research

Publisher website links and WP:PRIMARY

Moved to Wikipedia:No original research/Noticeboard § Publisher website links and WP:PRIMARY

– 15:34, 10 May 2021‎ (UTC)

RFC plans

I've started drafting an RFC below^{(this is not even an archiving-robust cross-reference SmokeyJoe (talk) 01:59, 23 December 2022 (UTC))}. I'm struggling to articulate a logical reason (beyond a bias against change, which is not unreasonable) for having PSTS remain on the NOR page. If anyone has an idea, I'd be happy to see it. WhatamIdoing (talk) 00:00, 23 December 2022 (UTC)

• There is one part of PSTS that I think could remain: the warning that Primary sources should be used with caution. This is the only part of PSTS that directly addresses the concept of original research - since it is very easy to (perhaps unintentionally) misuse primary sources in ways that result in original research. Blueboar (talk) 00:35, 23 December 2022 (UTC)

  We had talked above about adding a summary of PSTS to the Wikipedia:No original research#Related policies. This would be an easy way to duplicate that reminder. WhatamIdoing (talk) 00:40, 23 December 2022 (UTC)

@SmokeyJoe, we're discussing the potential RFC question up here. Your comment that Trying to explain WP:NOR without strong emphasis on the need for secondary sources (as defined in the historiological field) is flawed and confusing. might give me something to put in the other column, which would make me feel better. To make sure I've got this right, you're saying that:

• The definition is of original research is: "material—such as facts, allegations, and ideas—for which no reliable, published sources exist."
• It is wrong to tell someone that some bit of "material—such as facts, allegations, and ideas—for which no reliable, published sources exist" is original research unless you first explain to them what a secondary source is (as defined in historiological terms.

Right? So you can't just say "Alleging that Queen Elizabeth was a reptilian alien is what we call 'original research', because no reliable published sources say that she was either reptilian or an alien". You first have to say "Articles must overall use more secondary sources than primary sources", and then go on to say that.

Looking at that, I'm pretty sure that one of us is wrong. WhatamIdoing (talk) 01:01, 23 December 2022 (UTC)

1st dot point No. Original research does not include fake facts. The word choice of “material” is poor. Original research is about information and knowledge.

2nd dot point. To understand original research, one must first understand primary and secondary source distinction. It is not wrong to tell someone things in a confusing order, but it is not good.

On the alleged alien nature of the queen, discussing original research is confusing because WP:V is not even met. There is not a single reliable primary source for this. There is not a single source. Getting into source typing with zero sources is silly. Start with the sources, and then we can discuss whether your article writing is derived from these sources, or is original research on your part. SmokeyJoe (talk) 22:20, 24 December 2022 (UTC)

• The word choice of "material" might be poor, but it is the word that is in the second sentence of the policy. OR does include fake facts. OR == "material for which no reliable, published sources exist". Fake faces are one type of "material for which no reliable, published sources exist".
• Why? What exactly do I need to know about primary and secondary sources distinctions before I can understand that "Queen Elizabeth was a reptilian alien" is an example of "material for which no reliable, published sources exist", which the second sentence of this policy says is called "original research"?
• The whole point of NOR is that it's stuff that can't meet WP:V. There is not a single primary source; there is not a single secondary source; there is not a single tertiary source. That's the point. NOR == material for which no source exists [i.e., in the real world]. That is the literal definition of OR in the second sentence of the policy. The actual definition of OR is given in the second sentence. The actual definition of OR does not mention historiography at all.

WhatamIdoing (talk) 22:36, 27 December 2022 (UTC)

You’ve criticised the policy for having bloat like the redundant definition of primary and secondary sources. I disagree with that, as it think it does a good job of paraphrasing the articles. My criticism of WP:NOR is it’s lead, including the introduction of the vague word “material”. I actually think the policy would be improved by deleting of the entire five paragraph lead. Possibly, we are in agreement that there is a problem of bloat? The answer to bloat is to cut the bloat, not to WP:SPLIT as that would encourage worsening bloat. SmokeyJoe (talk) 02:40, 28 December 2022 (UTC)

OR does include fake facts? I disagree. The meaning of OR does not include fake facts. OR is about an editor being their own secondary source for content writing. A fact fake is a much simpler problem, not merely of something being unverifiable, but of being wrong. I think you have recently adopted a peculiar perception of OR. I would like to know why. I suspect that it has to do with a lax application of WP:NOR to medial articles, and I think this happens because Wikipedia has become very close to the cutting edge of medical science. It’s harder to follow when you are close to the cutting edge. “Harder” does not mean “wrong”. SmokeyJoe (talk) 02:48, 28 December 2022 (UTC)

the second sentence of this policy says is called "original research". Yeah. The second sentence is bunk. The whole five paragraph lead is poor. The problems you seem to be seeing is the lead, not PSTS, not the policy structure NOR-PSTS-SYNTH. SmokeyJoe (talk) 02:50, 28 December 2022 (UTC)

Are you interested in proposing the removal of the definition of OR from this policy? WhatamIdoing (talk) 22:18, 4 January 2023 (UTC)

User:WhatamIdoing, this opened an interesting can of worms. I see pros and cons to a few options:

1. As far as possible, use real world definitions (but they aren’t very good, or in the applicable context)

2. Remove the definition (but there is User:Barkeep49‘s respectable objection below, and failure to define a page title is a page failure. )

3. status quo (it uses poor vague waving to define, it is wordy, it is confusing)

4. Move the definition to its whole section at the bottom

5. Change the title, eg by merging to WP:ATT, so that core content policy *is* written in plain English.

I don’t think the removal of the definition, entirely, is a good idea, if it is the title. I think we should improve the definition, as a Wikipedia term of art, because, vague as it is, it is very deeply entrenched in Wikipedia culture. I think we can all agree on the meaning of “original”. I think more focus should be drawn to the definition of “research”, which points immediately to knowledge, which is above facts, pointing immediately to epistemology. SmokeyJoe (talk) 00:54, 5 January 2023 (UTC)

The whole point of NOR is that it's stuff that can't meet WP:V. Disagree again. We are not even agreeing on what we disagree about. “The whole point of” is a sweeping construct. I think you should be more precise, rather than I try to falsify your statement by pointing out points of NOR that are different to WP:V. SmokeyJoe (talk) 02:53, 28 December 2022 (UTC)

• I'd keep it simple and neutral. A question along the lines of the following should suffice (everything else, including the table and the FAQ, should be moved to the discussion section): should the WP:PST section of the policy be moved to its own policy page, WP:Primary, secondary, and tertiary sources (currently a redirect), with no other changes made to either (except as necessary to fix links, grammar, etc.)? M.Bitton (talk) 02:00, 23 December 2022 (UTC)

  I'd like to, but the previous discussion showed that people really struggle with the very simple question. It took a while for people to grasp that we were starting with one page that says "{{policy}} NOR – PSTS – SYNTH" and that we would end up with two pages, one of which said "{{policy}} NOR – SYNTH", and another of which would say "{{policy}} PSTS". I think there was a fear that PSTS was somehow being demoted, or that it wouldn't really be a policy if it wasn't part of this specific policy. WhatamIdoing (talk) 22:40, 27 December 2022 (UTC)

  It is not sensible to suggest the a reading of SYNTH without having first read PSTS. It is secondary sources that make information out of facts, not editors. SmokeyJoe (talk) 02:38, 28 December 2022 (UTC)

  So if I have a source that says:

  The United Nations' stated objective is to maintain international peace and security.

  and another source that says:

  Since the creation of the UN, there have been 160 wars throughout the world.

  then I'm not going to be able to understand that it would be a SYNTH violation for me to write:

  The United Nations' stated objective is to maintain international peace and security, but since its creation there have been 160 wars throughout the world.

  unless I first read PSTS?

  Do I understand your view correctly? WhatamIdoing (talk) 22:34, 4 January 2023 (UTC)

  I’ll admit that it is possible to understand WP:SYNTH violations without using PSTS, but I wouldn’t call your red text example a SYNTH violation. SmokeyJoe (talk) 08:22, 6 January 2023 (UTC)

  You wouldn't? That's rather dismaying, because I copied that red text straight out of WP:SYNTH. That sentence has been given as the first, simple example of SYNTH since mid-2009. WhatamIdoing (talk) 01:30, 9 January 2023 (UTC)

  I think the example tends a bit stringent. I would prefer to consider WP:SYNTH at the level of paragraphs. It certainly isn’t a “good” combination. SmokeyJoe (talk) 02:19, 9 January 2023 (UTC)

  A lot of misinformation can be written in a single sentence, or even less. SYNTH can happen in very small pieces. WhatamIdoing (talk) 02:33, 9 January 2023 (UTC)

  @Alanscottwalker, I appreciate your addition at Wikipedia talk:No original research#c-Alanscottwalker-20221224141800-Turning one policy page into two policy pages. I particularly want to have sound reasons in that column.

  That said, I'm not sure that anything you've written there is unique or specific to NOR (or if it does, it doesn't explain how). I therefore wonder if it actually supports your view as strongly as you hoped. For example:

  • NOR is about knowledge production, in particular about proper knowledge production for our encyclopedia.
  • NPOV is about knowledge production, in particular about proper knowledge production for our encyclopedia.
  • WP:V is about knowledge production, in particular about proper knowledge production for our encyclopedia.

  Those are all true, right? They seem equally true to me, or perhaps it is an even bigger point for NPOV. Do you think that the other content policies aren't about producing proper knowledge in the encyclopedia? If not, then it's probably not a good idea to implicitly claim that as a unique characteristic for NOR, or as a reason why proper knowledge production needs to be in NOR instead of in some other page.

  I also have some concerns about this:

  • Authors outside Wikipedia produce knowledge through published primary, secondary, and tertiary sources, and in the later two mixing these types of sources together. Wikipedians must generally create text that is original in the authorial sense (no copyright violations and no plagiarism), but not original knowledge. We do this by properly adhering to the three types of sources, prizing secondary sources above the others, so it is vital we have some sense of what they are....By understanding and properly using these three, we avoid original production.

  Tertiary sources don't have to mix the types together (it's not a case of primary+secondary=tertiary, like 1+2=3; many tertiary sources, such as textbooks for children, are written entirely from secondary sources), but leaving that correctable detail aside, it's unclear how it relates to NOR. How does properly prizing secondary sources above the other types help editors avoid adding material that isn't contained in any existing source at all? I can easily see how properly prizing secondary sources above the other types helps us avoid non-neutral, unbalanced articles, helps us avoid giving equal validity to unequal POVs, etc., but how does prizing a secondary source's analysis help us stop making up stuff that no source says? It seems to me that the real value in prizing secondary sources is in NPOV, not in stopping editors from adding "material—such as facts, allegations, and ideas—for which no reliable, published sources exist". No sources means no existing sources at all – not just no secondary sources.

  WhatamIdoing (talk) 02:50, 9 January 2023 (UTC)

1) Knowledge production: That phrase is meant to convey creating knowledge, NOR emphasizes that Wikipedia is aimed at communicating knowledge through original writing but not creating it new. Wikipedia does that summarizing sources, but not all sources are the same, and so therefore their use and usefulness in our summarization writing is different, depending on the source, and in our regularly employing combination of sources of different types, with the continueing aim of not being original. The 3 types of sources is a categorization that is widespread outside Wikipedia, it is not something Wikipedia invented, it is a conceptualization that is useful in research based expository writing, when one is putting sources together to not to create new knowledge. They serve both as exemplars of writing to be mimicked, and a warning of what not to do, eg., you should not be a primary source. All three policies though have some overlap and are designed to be read together, but their emphasis is different. NPOV is not emphasizing the handling of new knowledge creation, it's focus is taking already existing sourced knowledge, and V is not emphasizing creating new knowledge, its focus is about taking a single source.

2) Your phrase, "textbooks for children", clues the clued-in researcher/writer on its usefulness in our writing (not much), and clued-in involves dealing in the three types of sources. Anyone who has seen textbooks for children will also often see primary source material in it too, even if it is just a phrase or sentence of an original document (or a pull-out box of someone's quote). But even where a tertiary source only refers to secondary source material, it subtextually encompases the primary source material of the secondary source material. A secondary source being an exemplar of analysis, gives lines our writer is not to cross: not your own analysis, their analysis is what you are to convey. Alanscottwalker (talk) 22:43, 9 January 2023 (UTC)

1. As a point of practical politics, editors are generally very suspicious of anyone who says he's creating knowledge on wiki. I think I know what you mean, but the very idea of producing knowledge here is going to make editors' skin itch.
   I'd like to know more about your implicit statement that NOR's focus is not on existing sourced knowledge. You say that NPOV is about "already existing sourced knowledge". NOR prohibits all content/knowledge in that isn't "already existing" in the real world. If NOR prohibits "material—such as facts, allegations, and ideas—for which no reliable, published sources exist", then how is NOR not about making sure that Wikipedia contains only "existing" knowledge?
   And again: What does the requirement that Wikipedia articles contain only knowledge that already exits in the real world have to do with PSTS? Yes, we've borrowed and adapted the three categories of sources from the real world. But OR is banned with primary sources exactly as much as it's banned with secondary and tertiary sources. So why does PSTS need to be explained specifically in the context of "material—such as facts, allegations, and ideas—for which no reliable, published sources exist", instead of having it explained in the context of writing a decent encyclopedia article?
2. Using a textbook written for children (e.g., for 12 year olds) would not violate NOR. I don't remember whether you were involved in the Wikipedia:Identifying reliable sources (history) proposal, but one of the reasons it failed was because editors refused to agree that a textbook written for even young children was an unreliable source. You'd be better off having most of the article WP:Based upon scholarly books and upper-university textbooks, but a textbook for 12 year olds can be relied upon to correct report that simple facts, such as that Guy Fawkes didn't blow up Parliament, that Abraham Lincoln was the 16th president of the US, and so forth. I'm not sure why straightforward, simple NOR operations ("Don't write that Guy Fawkes blew up Parliament in an article, because that's "material—such as facts, allegations, and ideas—for which no reliable, published sources exist" and thus a NOR policy violation") requires us to prize secondary sources. Does it really? Or is prizing secondary sources less about NOR per se, and more about writing a decent encyclopedia article?

WhatamIdoing (talk) 22:36, 15 January 2023 (UTC)

@Alanscottwalker, I wonder if you are interested in continuing this conversation. WhatamIdoing (talk) 02:44, 24 February 2023 (UTC)

I think, I have already addressed your points, the clue to the focus is the word "No" in the title of this policy, as it has pride of place. And as I already indicated, misusing primary, or secondary, or tertiary sources likely leads to publishing original research, which is a "No" - to even begin to not misuse them, you have to have some understanding of them. -- Alanscottwalker (talk) 04:09, 24 February 2023 (UTC)

@Alanscottwalker, that seems unlikely to me. Did you need to read PSTS before you were able to figure out how to handle sources without introducing "material—such as facts, allegations, and ideas—for which no reliable, published sources exist"? For example, your first edit cited a source. Did you need to study PSTS to know whether you were using the source correctly? WhatamIdoing (talk) 02:26, 26 February 2023 (UTC)

Personal questioning? As should be apparent, without getting personal, writing based on research in sources was not invented by Wikipedia. The conceptualizations of primary, secondary, and tertiary, predates Wikipedia. As already established above, Wikipedia did not invent these concepts, nor did Wikipedia invent the processes of writing - these processes and conceptualizations were already within standard educational models for writing. To the extent there is something new, here, it would be necessarily communicating to each other the encyclopedia process, and have it be replicable and replicated, mimicable and mimicked, in a public wiki for all to see and do, at the same time. -- Alanscottwalker (talk) 08:33, 26 February 2023 (UTC)

On the one hand, you say that "to even begin to not misuse them, you have to have some understanding of them", but on the other hand, you say that you didn't actually have to read PSTS to avoid OR. So why is it essential, in avoiding OR, for editors to read the thing that neither you nor I needed to read? WhatamIdoing (talk) 02:48, 21 March 2023 (UTC)

It's essential because doing reading and research writing is using PST. -- Alanscottwalker (talk) 06:11, 21 March 2023 (UTC)

@Alanscottwalker, I have a question about a different statement you've made in support of keeping PSTS inside NOR, rather than splitting it out to its own independent policy. You say: The rest of the policy does refer to the three types, everytime it mentions sources

I would like you to open the policy page, find the PSTS ===subsection=== and blank it. Do the same for the ==See also== section. Then count up how many times you find these words in the text of the policy:

• primary
• secondary
• tertiary

I'm not sure how to understand your claim. Do you mean to say that there is a secret, unwritten mention of primary, secondary, and tertiary sources, so that where the policy says "independent sources" or "reliable sources" or just plain "sources", we are expected to read "independent primary, secondary, and tertiary sources" or "reliable primary, secondary, and tertiary sources", etc.? WhatamIdoing (talk) 02:55, 21 March 2023 (UTC)

They are upfront, so it is no secret, they are the 3 categories of sources, encompassing all sources. -- Alanscottwalker (talk) 06:24, 21 March 2023 (UTC)

Turning one policy page into two policy pages

The Wikipedia:No original research#Primary, secondary and tertiary sources section is important, but developed on this page almost accidentally, rather than through deliberate intention. Should this section of the policy be split to a separate policy page, with no other changes made to either (except as necessary to fix links, grammar, etc.)?

Should this policy be split into two policies?

Yes, this is a good idea.

No, we should not add this.

"Original research" is about editors making up stuff that isn't in the published reliable sources. Whether the published reliable sources are primary, secondary, or tertiary ("historiographical classification") is not essential to the concept of original research. It doesn't even get mentioned outside of the specific WP:PSTS subsection. The definitions of primary, secondary, and tertiary sources appear in multiple policies and guidelines. Although not closely tied to original research, it is a core concept for Wikipedia:Notability. Sources that are used to claim something in a Wikipedia article, when the sources don't directly support that claim, are NOR violations. It does not matter whether the misused sources are primary, secondary, or tertiary. Misuse of sources will remain 100% banned, just like it is today. Splitting the page will have the effect of making the key WP:SYNTH section more prominent in this policy.

NOR is about knowledge production, in particular about proper knowledge production for our encyclopedia. Wikipedia is generally text based knowledge. Authors outside Wikipedia produce knowledge through published primary, secondary, and tertiary sources, and in the later two mixing these types of sources together. Wikipedians must generally create text that is original in the authorial sense (no copyright violations and no plagiarism), but not original knowledge. We do this by properly adhering to the three types of sources, prizing secondary sources above the others, so it is vital we have some sense of what they are. We prize secondary sources because that is where analysis happens. We must understand primary sources because that is the basis for secondary sources (and we sometimes use primary sources ourselves). We must understand tertiary sources because that is the template for our articles' collectionary and summary purpose (and we sometimes use them ourselves), although our articles cannot contain anything "new". By understanding and properly using these three, we avoid original production. These concepts are central to this policy; we must understand this world of knowledge outside the pedia, and how we are the same but also different in the process of knowledge production (careful research in the three types, mixing them together appropriately, creation of original writing based on them, but not original knowledge).-- Alanscottwalker (talk) 14:18, 24 December 2022 (UTC) The rest of the policy does refer to the three types, everytime it mentions sources, our whole process is the labor of putting together (and omitting) and representing together (and omitting) the three types appropriately to make good encyclopedia articles (original in writing but not original in substance). -- Alanscottwalker (talk) 15:57, 24 December 2022 (UTC) This text developed organically here, so we just shouldn't change its location. Idea: dedicate a page for this, and transclude it here. —SmokeyJoe (talk) 01:57, 23 December 2022 (UTC) We do ourselves no favors by further balkanizing policy. The three policies V, NPOV and NOR must be read together, we say in each policy, but too often they are already balkanized. The last thing Wikipedia needs is yet another policy. -- Alanscottwalker (talk) 14:18, 24 December 2022 (UTC)

Questions

Frequently asked questions:

How did this happen?

In 2004, after a discussion on the mailing list about an editor who wanted to use Wikipedia to promote his new idea about physics, this policy was updated to say "Wikipedia is not the place for original research such as "new" theories. Wikipedia is not a primary source."  Later, someone though it would be helpful to have a definition of "primary source", and eventually instruction creep took over, and the one sentence has turned into 800 words with 7 explanatory footnotes, and 7 sources.

Won't splitting the content between two pages gut the policy?

No. Every sentence currently part of this policy will continue to be part of a policy.

Won't this make PSTS stop being a policy?

No. Both pages will have the policy tag at the top.

Won't this make SYNTH stop being a policy?

No. Both pages will have the policy tag at the top.

Do you seriously mean just cutting and pasting some text to a different page, and nothing else really changes?

Yes, with the caveat that the new page will need a basic introductory sentence, and we'll need to fix a few links or similar details. The WP:PSTS and other shortcuts will point to the new page. If you are interested in the details, see this sandbox.

What would the new policy page be called?

The new page could be located at Wikipedia:Primary, secondary, and tertiary sources (an existing redirect to the current section). If you have better ideas, please add them in your comments.

Your questions:

• Would WP:PSTS be a new "Core content policy" or "Other content policy"? a455bcd9 (Antoine) (talk) 14:58, 24 December 2022 (UTC)

Comments

• RFCs are discussions, not votes. Add your comments here!
• Oppose. WP:PSTS is the intellectual basis for WP:NOR and it’s removal would leave WP:NOR with a hole at its centre. If *any* restructure of core content policy is a good idea, it is the merging of WP:NOR and WP:V (see WP:A), which preserves WP:PSTS as the intellectual foundation of the combination. No good content doesn’t have both primary sources and secondary sources, and in balance. Trying to explain WP:NOR without strong emphasis on the need for secondary sources (as defined in the historiological field) is flawed and confusing. —SmokeyJoe (talk) 00:39, 23 December 2022 (UTC)

  Later, someone though it would be helpful to have a definition of "primary source", and eventually instruction creep took over, and the one sentence has turned into 800 words with 7 explanatory footnotes, and 7 sources.

  Instruction creep, and general bloat, is a problem in any instruction. Concise is good. Redundancy in instructions is bad. The more words, the more likely none will be read.

  Regarding the “primary source” definition bloat, the answer is to strip it back, remove anything redundant with the article primary source. Keep core policy concise, do not fork and pander to the bloat. SmokeyJoe (talk) 00:45, 23 December 2022 (UTC)

  On re-reading the policy, I disagree that there is too much instruction creep in creating a new definition of "primary source" and "secondary source". WP:PSTS prominently links to the mainspace articles, and paraphrases from the articles. SmokeyJoe (talk) 22:56, 24 December 2022 (UTC)

WAID, please sign your posts. —SmokeyJoe (talk) 00:47, 23 December 2022 (UTC)

If you want this RFC to be a private draft for now, do it in a userspace subpage. If it is here on the policy talk page, it is open, now. —SmokeyJoe (talk) 00:48, 23 December 2022 (UTC)

@SmokeyJoe, please look at the coffeeroll-colored tag at the top of this section that says "This draft RfC is not yet open for comments. Please discuss changes to the format of this RfC on the talk page, but do not comment on the topic of the RfC itself until it opens".

The reason the community created this tag is so that RFCs could be drafted in public. That's because it's hard to draft a discussion privately and still follow the advice at WP:RFCBRIEF that says "It may be helpful to discuss your planned RfC question on the talk page before starting the RfC, to see whether other editors have ideas for making it clearer or more concise." Please remove your comments for now (and this one, too, if you'd like), and wait until we have a clear question. If you have advice on the question itself, then please scroll up one section and join Blueboar and me on whether this question could be improved. If you just want to talk about the subject itself, you're always welcome on my own talk page. WhatamIdoing (talk) 00:52, 23 December 2022 (UTC)

If this draft is not open for comments, take it off this page. Put it on a subpage, preferably in your userspace. Possibly make it a project page so that it can have its own talk page.

I don’t respect your right to put that tag on a section and have it respected. No ownership of talk page or talk page sections.

I don’t agree to remove my comments, and would object to you removing yours. Instead, put the owned draft on its own page, with its own watchlisting and history of edits. In general, I think if something is worth an RfC, it is worth its own page.

You could link section to talk page threads.

I’m all for refining a question to be a better fairer question. SmokeyJoe (talk) 01:28, 23 December 2022 (UTC)

@SmokeyJoe, WP:RFCBRIEF says to "discuss your planned RfC question on the talk page before starting the RfC". Please tell me how I am to discuss my planned RfC question on the talk page if I take it off this talk page. WhatamIdoing (talk) 01:50, 23 December 2022 (UTC)

By putting the draft RfC on the page WP:No original research/2022 RfC to spinout WP:PSTS. Use WT:No original research/2022 RfC to spinout WP:PSTS to discuss it. I can also think of other possible ways that don’t involve implied ownership. SmokeyJoe (talk) 01:54, 23 December 2022 (UTC)

• I'm very open to this idea and I expect to be supporting this proposal when the drafting is done.—S Marshall T/C 00:31, 24 December 2022 (UTC)

• Compare the sentence "Original research" is about editors making up stuff that isn't in the published reliable sources. with Readers must be able to check that any of the information within Wikipedia articles is not just made up. If that is the purpose of NOR, then what is the difference between NOR and V? I think this sentence is somewhat confusing.--Paul Siebert (talk) 04:27, 24 December 2022 (UTC)
  • I think OR is a spinout of V. It's encountered so often and has so many facets that it has its own page but actually it's a case of V, not a truly separate thing. An attempt to merge the two at WP:ATT was historically unsuccessful.—S Marshall T/C 10:39, 24 December 2022 (UTC)

    You think OR is a spinout of V? That’s an odd thing to think. Why do you think it? I think your quite differently. WP:V is about facts. WP:NOR is about information, and whether others have though the information was worth publishing in reliable sources. WP:V is the simple requirement that Wikipedia has its facts right. WP:NOR is about ensuring that editors are not weaving new information, by synthesising new information from selected facts. This is no subset of WP:V. In historiography, this is an old and mature discipline of source typing, and an encyclopedia is well considered squarely an historiographical document. It certainly is not well considered science or journalism.

    WP:ATT was pretty good, but there was a failure in change management. SmokeyJoe (talk) 21:26, 24 December 2022 (UTC)

    Actually, if you trace the history of the two policies you will find that it happened the other way around... NOR came first, and WP:V was a spin out from that. Not that it matters... the two are intimately linked. Blueboar (talk) 23:54, 24 December 2022 (UTC)

    • • Yes, that's what I think. They're two sides of the same coin. If I was Ruling Tyrant of Wikipedia, then I'd reduce both policies to a single line each:
        WP:V: Make sure that each article says the same thing that the sources say. Use citations to show how you've done this.
        WP:NOR: Make sure that each article doesn't say something the sources don't say.
        There would be supplementary guidelines that explain everything else, and all wrangles about primary, secondary and tertiary sources would be banned until Wikipedians can coalesce on a single definition of each of those words and then give a short, clear explanation of why they matter.—S Marshall T/C 23:17, 24 December 2022 (UTC)

        Pithy.

        Your approach has merit. I think WP:5P is a celebrated success of that approach. SmokeyJoe (talk) 23:45, 24 December 2022 (UTC)

        I would support this, if anyone can be bothered to propose it. BilledMammal (talk) 02:51, 21 March 2023 (UTC)

• I am opposed, and I have inputed some reasons in the draft template above. Alanscottwalker (talk) 14:28, 24 December 2022 (UTC)
• Oppose as this approach to the draft. The question should be simple and the exactly changes to policy (given as major that us being asked) should be avoided. If there is general support for making PSTS a core content policy, then a separate discussion can be had to discuss wording. --Masem (t) 00:54, 25 December 2022 (UTC)

Returning the key point

• Whether we split off the PSTS section or not, I do think we should return the point that gave rise to it: That we need to avoid adding anything that would make Wikipedia a Primary source for information.

It was this point that originally caused us to explain what a primary source is (and subsequently what Secondary and Tertiary sources are), and when we removed that point we removed what directly tied PSTS to the broader point behind this policy. Blueboar (talk) 19:23, 25 December 2022 (UTC)

When was this? Wikipedia should not be used as any kind of a source for Wikipedia. SmokeyJoe (talk) 22:08, 25 December 2022 (UTC)

NOR began life to deal with the same question as Wikipedia:What Wikipedia is not#Wikipedia is not a publisher of original thought. NOR's first mention of primary sources was to say that Wikipedia itself isn't (meant to be) a primary source. This was a couple of years before you started editing, and there had been a problem at the time with people thinking that they could publish their own actual research papers here. People were literally posting papers they'd written for school, or had rejected from academic journals, and they thought they'd just put it on this Wikipedia site they'd just heard about so the world could find out about it. Quaint, right? But Wikipedia wasn't well known back then, Wikipedia:Five pillars still hadn't been written, and folks were still sorting out what belonged and what didn't. This was a really l-o-n-g time ago. We actually did need a rule that said Wikipedia is not meant for any facts/information/material/content/or anything else that hadn't already been published elsewhere. That's why this policy exists That's why the definition is about "material—such as facts, allegations, and ideas—for which no reliable, published sources exist", rather than about transforming "facts" into "information", or about getting the correct balance between secondary and non-secondary sources. This policy's raison d'etre is to stop people from writing Wikipedia articles about how they've single-handedly disproven modern physics, even though the evil physics cabals refuse to let them publish their proofs in peer-reviewed scientific journals.

(As for Wikipedia being an invalid source for Wikipedia... If you believe that, then Category:Wikipedia is awaiting your clean-up efforts.) WhatamIdoing (talk) 23:05, 27 December 2022 (UTC)

I find the quaint history interesting, fascinating. I know about the 1990s tendency of amateur wannabe science philosophers posting manifestos to the local university. Wikipedia appeared as an outlet for them, and I completely understands Jimbo’s quote. I think the tendency was cured not by Wikipedias improvement in articulation of purpose, but by the improvement of Wikipedia search engines, which allowed kooks to self-educate and then to find their theories were not new.

The Jimbo quote makes perfect sense in the PSTS lens, even if Jimbo himself couldn’t define a secondary source. WP:5P1, Wikipedia is an encyclopedia. And an encyclopedia is a tertiary source.

You think that we should return to the point that gave rise to PSTS? Ok, I am keen to listen. However, before any action, I think we should also consider the question: what is the message to be given to the newcomers. Core content policies get used for high-language debates amongst old Wikipedians, but their real purpose is starting information for the newcomer who is starting to become serious about contributing to Wikipedia.

The rise of PSTS? Please, share your perspective.

- SmokeyJoe (talk) 23:58, 28 December 2022 (UTC)

Might I suggest that we document this history in a collapsible box on this talk page to help future editors know its roots? (We really should do this on other core P&G pages) Masem (t) 00:02, 29 December 2022 (UTC)

I like this idea, but suggest doing so in a separate page. Self referencing is poor. No document should attempt to describe its own history. SmokeyJoe (talk) 00:27, 29 December 2022 (UTC)

@Masem, there's a FAQ transcluded at the top of this page; it's a separate page, but would be relatively visible. A history of NOR was written in 2007 by SlimVirgin, after the ATT merge failed. That has since been merged to Wikipedia:Core content policies. If you added a question and a brief answer, you could point to the longer story.

I notice that words such as primary and secondary do not appear in SV's telling of the history, even though by that point, the PSTS section looked much like it does today. This does not surprise me, given that PSTS is central to writing a decent encyclopedia article, rather than central to keeping pseudophysics out of Wikipedia. WhatamIdoing (talk) 22:49, 4 January 2023 (UTC)

SV was not comfortable with the historiographical definitions. She seemed to prefer the journalism definitions. I recall challenging her on her choice of definitions, and she said (my loose recollection) that good journalism standards (like good history writing) supports good Wikipedia content. SmokeyJoe (talk) 00:34, 5 January 2023 (UTC)

That definition also allowed her to conflate secondary with independent, and thus claim that a breaking news story, or eyewitness journalism, should be counted as "secondary" for the purposes of WP:GNG. It is one of the few points that she and I completely disagreed with. But overall, I'd say that PSTS is central not to NOR, but to something much bigger than just NOR. WhatamIdoing (talk) 02:53, 5 January 2023 (UTC)

PSTS is central not to NOR, but to something much bigger than just NOR?

Very interesting. SmokeyJoe (talk) 03:28, 5 January 2023 (UTC)

Yes. I call that bigger thing "writing an encyclopedia article". Not making up stuff, whether the made-up stuff about how you're smarter than Einstein or about why you think the local mayor lost the election or about why you believe that mass murder victims are all crisis actors, is important to that bigger goal, but PSTS goes well beyond not making stuff up. PSTS is how you (should) decide what you can say and how much you should say about it. WhatamIdoing (talk) 02:31, 24 February 2023 (UTC)

Implicit synthesis

The policy clearly states that implicit synthesis should not be made by editors. But what about cases where there is implicit synthesis in the source? TFD (talk) 21:21, 7 January 2023 (UTC)

Any synthesis in a source (whether implied or implicit) would not be a violation of our NOR policy… as the synthesis does not Originate with a WP editor. NOR is about what we write, not about what the sources write. Blueboar (talk) 22:11, 7 January 2023 (UTC)

I have had a change of mind… because an implied synthesis could be unintentional. We might see a connection between things in a source that the author didn’t mean to be connected. If so, then we are the ones connecting the dots, not the author of the source. And that is indeed OR. Blueboar (talk) 22:21, 7 January 2023 (UTC)

Specific examples would be helpful. M.Bitton (talk) 22:28, 7 January 2023 (UTC)

A good secondary source should be explicit in the synthesis. It may not be a good source. I’m imaging an example where examples are grouped, and it is implied that grouped examples are similar, but it is never stated why the examples are grouped together. SmokeyJoe (talk) 22:52, 7 January 2023 (UTC)

Not sure… if an author states that A, B and C are all examples of X, wouldn’t the synthesis be explicit? The author is explicitly linking them as examples, even if the author does not explain why they are examples. Blueboar (talk) 23:07, 7 January 2023 (UTC)

An author stating A B C are examples of X is explicit, even if weak on why. An author might put Q R S together, maybe after previous groupings with explicit reasons, and it may be that they are similar, or it may be that they are the leftovers. A reader may infer a reason for the grouping, which would be synthesis by the reader.

Specific examples, would be helpful. SmokeyJoe (talk) 23:19, 7 January 2023 (UTC)

• I think this is about this (permalink). The disputed edit is whether to say "Four officers who responded to the attack killed themselves within seven months", on the basis of this source (archive link). My take is that yes, the Reuters source is reliable for the claim that four officers who responded to the attack died by suicide, and to say so doesn't violate WP:SYNTH because the inference there is in the source. This is because SYNTH stops Wikipedians from reaching novel conclusions, but it doesn't stop sources from reaching them. However, I think it's a problematic edit to make for other reasons, and I'd draw your attention to this essay and the discussion with User:WhatamIdoing at User talk:S Marshall#Disapproving tone which gave rise to it.—S Marshall T/C 00:13, 8 January 2023 (UTC)

  The source makes a primary source observation connecting an event to a later cause of death. This is a primary source in demography. The source, Reuters, and the named author, is not a reliable source for demographic synthesis. Reuters, in my opinion, is a leading quality news reporter for primary source information and standing back from opinion and bias and any other form of secondary source content.

  Statistics is a mature robust academic discipline. Four deaths is not statistically significant.

  The making of a connection, four counts, a demographic connection between an event and later deaths, might not be WP:OR due to a source having done it, but the source is unreliable for any judgement on the reliability of there being a connection.

  # Primary sources that have been reputably published may be used in Wikipedia, but only with care, because it is easy to misuse them.^[e]

  Any exceptional claim would require exceptional sources.

  WP:EXCEPTIONAL applies.

  - SmokeyJoe (talk) 01:18, 8 January 2023 (UTC)

• Sorry, SmokeyJoe. I know I keep disagreeing with you and I swear it's not personal. We've known each other on-Wiki for a long time and agreed a lot over the years. But I can't accept what you say here.
  On your first point, Reuters is reporting figures from the District of Columbia Police Department. The District of Columbia Police Department is a reliable source for the cause and number of deaths of its own staff, and Reuters is an editorially independent secondary source.
  Counting suicides is not demographic synthesis.
  Around 2,000 police officers were deployed on 6 January 2020. Between that date and the Reuters report on 3 August 2021, four of those police officers killed themselves. Well, age-standardized suicide rates in the US for a comparable period are 14.5 per 100,000 (according to List of countries by suicide rate which is in turn based on WHO data). So contrary to your statement that "four deaths is not statistically significant", it is in fact quite a few standard deviations from the mean.
  The problem is that the Reuters article doesn't go further: it just counts the four suicides. In doing so it is blatantly inviting the reader to draw conclusions about the effect of the capitol attack on police officers' mental health, but it doesn't provide the proper paper by an academic statistician that we would consider ideal.
  But we can't rule out Reuters as a news source for articles about law and order in the US.—S Marshall T/C 12:30, 8 January 2023 (UTC)

  With such determined different perspectives from amateurs, there’s no way Wikipedia can work. SmokeyJoe (talk) 12:38, 8 January 2023 (UTC)

• And yet somehow Wikipedia does work… amazing! Blueboar (talk) 13:13, 8 January 2023 (UTC)

  That's not statistically significant? Says who?

  In the US, a little more than four out of each 30,000 people died of suicide over the course of the 12 months of 2020. Among the officers surviving the attack, four out of less than 1,000 died of suicide over the course of just 7 months. That's about a 50x difference. I haven't actually calculated the p-value, but I'm confident just from a glance that a 50x difference on these numbers is statistically significant.

  More to the point, if a source says something, then it's not OR. It could be UNDUE; it could be a NOT violation; it could be an inappropriate source; it could be the kind of source that isn't really usable in practice; it could be a source that is misused (e.g., it needs WP:INTEXT attribution), but the one thing that content taken directly from a generally reliable, published source is not is "material—such as facts, allegations, and ideas—for which no reliable, published sources exist" – and, for better or worse, that is exactly the definition of OR. WhatamIdoing (talk) 02:30, 9 January 2023 (UTC)

  Four is a small number.

  Using google scho r to search: capitol attack and police suicide, since 2022, yields many . This is a serious matter, and Wikipedia should proceed carefully. It is not an OR matter. SmokeyJoe (talk) 03:13, 9 January 2023 (UTC)

  This is why no original research is foundational to Wikipedia and why it predates even the requirement for referencing. Before we had developed the ideas behind ref templates etc., Wikipedia still knew that it was not an outlet for original thought. That means we have to accept the conclusions of reliable sources. If the reliable sources ascribe event A to cause B, in the preponderance of their weight and reliability of course, we should also accept that as fact. Original research is original speculation, or connecting dots without a foundation, or creating a chain of implications. It also could be trying to limit the boundaries of what is reliable in a novel way. For example, in the situation where someone has conducted a statistical analysis to determine whether the capitol police suicides, we have to accept what sources ascribe that to. If there are notable minority viewpoints we may consider those in an attributed, contextualized way. But doing math to verify the claims made by reliable sources for editorial evaluation and using that to influence the article is a kind of synthetic thought that goes beyond the bounds of encyclopedic summarization. We simply need to present what the scientific, or journalistic authorities have reported. Andre🚐 03:53, 9 January 2023 (UTC)

• Yes, exactly, but what's interesting in this case is that the source doesn't explicitly reach a conclusion. It gives you the numbers and implies the conclusion that these suicide rates are abnormal. That conclusion is not unreasonable because mathematically they are abnormal (as verifiable by Wikipedians doing basic arithmetic that would pass WP:CALC); so I would say we can give the numbers from the source.—S Marshall T/C 08:45, 9 January 2023 (UTC)

  I concur. Andre🚐 22:33, 9 January 2023 (UTC)

  I concur with what you have written. Reporting reported data is absolutely fine. Reuters is somewhat a secondary source because it contextualises the data, even if it doesn’t do much more.

  What I mean is that the source is unworthy to claim a connection. It may be random. The p-value is not reported, and even then it may be cherry picking of extraordinary data from a large data set (a selection bias). There may be a non-causal correlation (eg maybe the capitol attacks happened in part due to a recent decline in respect afforded to police). There are indeed quality secondary sources addressing the connection, but the Reuters report is not one of them. The unusual slip towards comment by Reuters might be attributed to Reuters staff and the author being well aware of the quality secondary sources. SmokeyJoe (talk) 00:38, 10 January 2023 (UTC)

• The Reuters report definitely is an example of implicit synthesis (connecting these police suicides with the Jan 6 riot). However, it does not violate our WP:NOR policy because it is Reuters making that implication, not a Wikipedian. We can talk about other reasons to omit/keep the information, but NOR isn’t in play here. Blueboar (talk) 13:19, 9 January 2023 (UTC)

• I agree what Reuters did doesn't violate Wikipedia's SYNTH rules since the implied causal relationship comes right from the source. However, I also agree that this is a great example of an implied claim that is also an exceptional claim. Per WP:V exceptional claims requires multiple high-quality sources. In this case that the implied claim would need exceptionally high quality sources and a normal news source, even a good one like Reuters, isn't at that level. It's OK for a news source to beg a question (or conclusion) like this but an encyclopedia shouldn't. Springee (talk) 18:37, 9 January 2023 (UTC)

  But the majority of the reliable sources do make the same conclusion. Andre🚐 01:56, 10 January 2023 (UTC)

In which case, we probably should cite those sources instead of Reuters. Blueboar (talk) 02:39, 10 January 2023 (UTC)

Additionally, we should check to see if any sources dispute the association or note that no causal relationship has been established. Springee (talk) 03:49, 10 January 2023 (UTC)

This factcheck.org article suggests that two of the 4 suicides are likely causal (and if I read correctly, later declared deaths in the line of duty) while the other two were much later and may not be causally linked. [1] Springee (talk) 04:00, 10 January 2023 (UTC)

@TFD The policy, WP:NOR, as currently written, is clear that the source must explicitly support the article content:

The best practice is to research the most reliable sources on the topic and summarize what they say in your own words, with each statement in the article being verifiable in a source that makes that statement explicitly.;

Even with well-sourced material, if you use it out of context, or to reach or imply a conclusion not directly and explicitly supported by the source, you are engaging in original research;

Do not combine material from multiple sources to reach or imply a conclusion not explicitly stated by any source. Similarly, do not combine different parts of one source to reach or imply a conclusion not explicitly stated by the source.;

A source "directly supports" a given piece of material if the information is present explicitly in the source so that using this source to support the material is not a violation of this policy against original research.

Rotary Engine (was Ryk72) ^talk 22:43, 13 January 2023 (UTC)

WP:SYNTHNOTSUMMARY. The quoted policy says that the best practice is to do X, and says reaching a conclusion not directly supported, it is OR. It does not say that simply summarizing conclusions made by sources is OR. In the case of the suicides, the sources aren't saying, they are simply stating statistics (WP:CALC) and we summarize their descriptions. Sources are perhaps also inferring or implying stuff but we are not. We are simply summarizing. Andre🚐 22:55, 13 January 2023 (UTC)

One way to look at it is that making the (implied) statement needs to be directly supported by the source. The source did not directly make that statement and so is not sufficient to support including that implied statement. North8000 (talk) 01:51, 10 January 2023 (UTC)

That is contrary to my understanding. Andre🚐 01:57, 10 January 2023 (UTC)

Well if there are multiple sources that say something similar isn't there one that says it more directly ? (In which case my point would be moot) Or even a source that reports that others postulate a connection.North8000 (talk) 02:09, 10 January 2023 (UTC)

If there are a series of reliable sources that all make the same implied connection, and none that say otherwise, that is sufficient to state it as fact in Wikivoice. Andre🚐 02:52, 10 January 2023 (UTC)

Per my comment above, Factcheck.org only links the two suicides that occurred within a month of Jan 6. Springee (talk) 04:02, 10 January 2023 (UTC)

You don't need to make the comment twice. I don't see how the Factcheck.org link is germane to the discussion of original research. That probably belongs on an article talk page. Andre🚐 17:56, 10 January 2023 (UTC)

My sincere and deepest apologies for thinking to post the comment twice. I'm sorry you had to take the extra time to read it a second time. The reason why it is relevant is you suggested that if no sources disagree... well at least one does. Springee (talk) 04:10, 11 January 2023 (UTC)

I don't mind reading it but it fragments the discussion. That source doesn't really say what you say it says. It's older than some sources, and it was updated later with a note on the bottom, and it is presenting a nuanced point that really doesn't contradict the other sources. In fact, it specifically avoids taking a position: We take no position in the debate over whom to include in the deaths from the riot Andre🚐 04:21, 11 January 2023 (UTC)

The discussion was already fragmented so I decided to reply to both fragments. Since we are quoting things, "On Aug. 2, the Washington Post reported: “Authorities drew no connection between the riot and his death. An official familiar with the investigation said Hashida had struggles beyond Jan. 6 that could have played a role.”" The quote you include actually applies to all the deaths they discuss, including Ashli Babbitt and not just the suicides. Springee (talk) 04:34, 11 January 2023 (UTC)

That was outdated, and it's just a reference to another Washington Post article that is outdated. This has nothing to do with original research at all. Nor does it contradict the claims other sources that are also reliable that Jan 6 played a role in the suicide. Andre🚐 04:43, 11 January 2023 (UTC)

Outdated? So what newer articles make it clear there was a causal relationship and how did they prove it. I guess that doesn't need to be answered here vs on the article talk page but it seems you are accepting of questionable correlation. Incidentally, we certainly can say that some politicians associated the suicides but that was likely for political rather than evidentiary reasons. Why would Wikipedia editors want to put an implied conclusion in our article, don't know. I'm sure we can find sources that say who made the associations instead. Springee (talk) 05:08, 11 January 2023 (UTC)

It's not original research to not check the "proof" of what sources did. They just say stuff and we say the same stuff. We do not need to check their reasons or their conclusions or implying conclusions or making any conclusions. It's simple. The sources say X number of people died from Y, we just say the same thing. We are not implying a conclusion. Andre🚐 22:57, 13 January 2023 (UTC)

The Reuters example is an edge case, and edge cases make for bad policy. This is an example where we would need to discuss. "The source connects these facts." "No, it doesn't explicitly do that, read carefully." If we treat this connection as a more extraordinary claim, then we'd expect more than one source to say it. Discussion is always going to be important, and it's impossible to create a policy that settles every argument in advance. Broadly speaking, it's not WP:OR when a reliable source says it, but it might be WP:OR if I draw a controversial interpretation from that source. Shooterwalker (talk) 15:36, 10 January 2023 (UTC)

• This does not seem that hard to me, the primary source is things like the death certificate, statements made by people with personal knowledge (decedent, family, doctors) etc.; the secondary, here, is the report not from someone with personal knowledge of the primary fact (but with expertise in reporting events), that placed the fact within the topic (should we decide to use it, the matter is a V issue of sticking to the source, don't misrepresent it); and the decision remaining is an issue of basically NPOV, comparative use of source material. -- Alanscottwalker (talk) 16:01, 10 January 2023 (UTC)

  WP:Secondary does not mean secondhand. For most of Wikipedia's purposes, most news is WP:PRIMARYNEWS. WhatamIdoing (talk) 00:25, 11 January 2023 (UTC)

  As I alluded to earlier, while I agree with this, I think it's worth noting that it is specifically breaking news reporting that should be considered purely primary. As time goes on, major news media and editorial outlets start taking on secondary analysis. It seems to happen rather quickly these days. And many political articles would be lightly sourced indeed if we didn't start considering articles in major media outlets as secondary sources after a while. Andre🚐 00:36, 11 January 2023 (UTC)

  I agree with this. Quickly? I’d say they start in about a week, are well into it in a month, but are not good secondary sources until about a year later.

  In any case, a breaking news topic is the sort of topic that should never be deleted, or AfD-ed, on the basis of being a NOR failure, but should be updated daily with better sources according to the advice of PSTS. (Is this a reason for PSTS to be a mere guideline??). Note here WP:DRAFTIFY#2c specifically excludes “a new topic likely to be of interest to multiple people (such as current affairs topics)” from back door deletion. It is a great strength for Wikipedia, and a means for new editor recruitment, that Wikipedia is up to date, within minutes of release of the first reliable source on a breaking news topic. SmokeyJoe (talk) 04:50, 11 January 2023 (UTC)

  I agree with most of that Andre🚐 04:54, 11 January 2023 (UTC)

  How unconvincing your rote employment of such shibboleths are. Examining the actual information in issue, examining the context in issue, examining the actual sources in issue, is the only useful path, here. Distance matters and always will (distance gives a wider view), while it is the case that newspaper articles from a historical time period are viewed as primary sources that is still a function of distance (but your analysis would rate historians as mere second handers), it is also the case that many newspaper articles of recent events are the only thing providing context for now (and doing it much better than any Wikipedia editor, could possibly do it). Newspaper articles may be the first draft of history but that's fine, because in our recent events articles, that is what we are doing. The unfinished work-in-progress pedia is what we are. (And anyone who is knowledgeable of the state of Wikipedia knows by now, no matter how much anyone may decry it, we are not going to stop the flood of recent events articles in the pedia, our only way forward is finding the best sources that exist now, those that give context.) -- Alanscottwalker (talk) 01:29, 11 January 2023 (UTC)

  I do generally agree with most of this - funny that I'm ending up here at the end of this with this dovetailing nicely with the prior thread Andre🚐 04:57, 11 January 2023 (UTC)

  Secondary sources contain some sort of analysis ("A review of previous reports of house fires in our town indicates that this has been the biggest fire in town for at least the last 20 years"). Secondhand reports may or may not contain any secondary contents. "My neighbor told me that she was afraid that he'd kill her" is secondhand. It is not secondary. WhatamIdoing (talk) 03:34, 14 January 2023 (UTC)

  But that's the point, there's breaking news reporting close to an event, and then we reach a point where they start analyzing and reaching conclusions about what's happening probably about a week later or less. The breaking of breaking news is a pure primary source. The analysis articles may be a mix of primary/secondary and becoming more and more secondary over time until they are mostly. Andre🚐 03:37, 14 January 2023 (UTC)

  I agree. The contents of a newspaper cannot be accepted as pure secondary ("but it's all secondhand information, and the author has expertise in reporting events, and there are journalistic ethics, and..."). They can also not be dismissed as pure primary ("but 19th-century newspapers get treated as primary sources by historians, so why not 21st-century ones, too?!). Most of what appears in my local newspapers are primary sources and should be treated as such by Wikipedians (also, they should mostly not used at all, because "New store opened" or "Mayor presided at city meeting" or "Police arrest drunk drivers on New Year's Eve" are not usually suitable for an encyclopedia article). However, that's only most, and some of it is secondary sources. WhatamIdoing (talk) 03:42, 14 January 2023 (UTC)

• Synthesis only applies when an editor combines two different sources to reach an implication not present in either. If an WP:RS itself combines two datapoints to make an implication, then our default stance should be to assume that that implication has been through their standard fact-checking and accuracy. Editors might object to it because they feel it is undue (especially if it is just a passing mention), but I think it is generally inappropriate to object to it because an editor feels the source is performing invalid synthesis, since that is functionally no different from "well, I think the source is wrong" or "well, the source hasn't convinced me, personally." Synthesis, like original research, is something we are not permitted to do ourselves but which we are supposed to rely on sources for - saying "the implication of these two things is significant and meaningful" is the kind of thing a secondary source is for. --Aquillion (talk) 13:26, 30 March 2023 (UTC)

Theoretical question on if someone published their own secondary source

Sorry if this was asked before or if I missed this mentioned here but i didnt catch it if it was in here. Lets say a PHD student decided to write and publish their own paper to an academic journal then used that as a source on wikipedia.

Would this count as a loophole (the editor being known as the publisher), or would it be accepted provided the paper doesn't get retracted later on? DarmaniLink (talk) 12:54, 8 February 2023 (UTC)

If it isn't on the page and I didn't miss it, we should add something brief about the unlikely-but-possible scenario DarmaniLink (talk) 12:58, 8 February 2023 (UTC)

Meh… see WP:COI. While we don’t disallow it, we also recognize that there is a conflict of interest with a Wikipedian citing their own work… so, at a minimum the editor should disclose that he/she is the author of the paper (not the publisher - that would be the journal) … and, ideally, they would wait and let someone else cite the paper. Also, a lot depends on the specific journal that published the paper and its reputation. Not all journals are equal. Blueboar (talk) 13:27, 8 February 2023 (UTC)

Disclose would require that they out themselves. North8000 (talk) 13:52, 8 February 2023 (UTC)

I presume North8000 is referring to the "General COI" section of the "Conflict of interest" guideline. That section describes possible ways to disclose a general COI, but does not explain how to decide if a COI exists.

I think the relevant section to decide if a COI exists when a subject matter expert cites a paper written by the expert is the "Citing yourself" section of the guideline. As I read that section, having your username in the page edit history is sufficient if the edit is "within reason, but only if it is relevant, conforms to the content policies, including WP:SELFPUB, and is not excessive." Of course, if one of the other sections about various kinds of COI apply, and are more stringent about disclosure, the other section should be followed. For example, if the editor believed that citing the paper would lead to it being cited in journals, and improve the editor's Author Impact Factor and thus improve the editor's likelihood of obtaining a desirable job, a clear conflict of interest would exist. Jc3s5h (talk) 18:47, 8 February 2023 (UTC)

thanks, i thought this might had already been covered but now i know for sure and know more too DarmaniLink (talk) 19:12, 8 February 2023 (UTC)

Actually I was just making the observation that an editor disclosing that they are the author is the editor outing themselves. North8000 (talk) 19:51, 8 February 2023 (UTC)

@DarmaniLink, you might be interested in Wikipedia:Party and person, Wikipedia:Identifying and using primary sources, and Wikipedia:Identifying and using self-published works. It is possible to write a secondary source, self-publish it (e.g., on a blog – this does happen), and then come to Wikipedia and try to cite it under WP:SELFCITE. It is uncommon, but not impossible, and sometimes (rarely) it's a valid source. WhatamIdoing (talk) 02:41, 24 February 2023 (UTC)

RFC on clarification to this proposal

I have started a RFC at WP:VPP asking for clarification of the OR policy regarding the use of maps and charts. This is related to a couple of threads that are already on this talk page where such clarifications were discussed. Dave (talk) 05:51, 19 March 2023 (UTC)

This is now at Wikipedia:Requests for comment/Using maps as sources. So far, 52 editors have made a total of 348 comments in this three-part RFC. Your participation would be welcome. WhatamIdoing (talk) 02:58, 21 March 2023 (UTC)

The RfC has been expanded since announced here. The proposals are now:

• Proposal 1: original research
  • Alternative to proposal 1
• Proposal 2: What can be cited to a map?
  • Proposal 2a: reliable sourcing
  • Proposal 2b: image layers
• Proposal 3: history
• Proposal 4: notability
  • Proposal 4a: Notability - Significant coverage
  • Proposal 4b: Notability - Sourced only to maps
• Proposal 5: Reliability

New proposals are marked in bold. BilledMammal (talk) 23:47, 27 March 2023 (UTC)

RfC on clarification of WP:CALC for costliest tornadoes

The following discussion is an archived record of a request for comment. Please do not modify it. No further edits should be made to this discussion. A summary of the conclusions reached follows.

Option 2; there is a rough consensus that calculating tornado costliness based off of NOAA, generally due to issues with NOAA itself, does not fall under WP:CALC but more so WP:OR. Costliness of a tornado must have a reliable secondary source attributed to the fact. InvadingInvader (userpage, talk) 06:35, 3 May 2023 (UTC)

Clarification as requested by Elijahandskip: Most editors seem to think that due to data issues with NOAA itself, that calculating ranks of tornado damage within a year without a non-NOAA source would violate WP:OR. Editors should reference a non-NOAA secondary source when claiming a tornado as the Xth-costliest. InvadingInvader (userpage, talk) 06:58, 3 May 2023 (UTC)

---

There has been disagreement between editors on multiple occasions whether or not the following situation is original research (not allowed) or if it falls under basic and routine calculations:

Most tornadoes in the United States are given a damage total provided by the National Oceanic and Atmospheric Administration (NOAA). Based on the those damage totals, a list of the top ten costliest tornadoes of that year is created. An example would be Tornadoes of 2022#Costliest United States tornadoes. NOAA does not provide a straight list of the costliest tornadoes of the year. This means there is no explicit source saying what the top ten costliest tornadoes of the year are as it was derived from the provided damage totals. Are Wikipedia articles allowed to say X tornado was the (1st/2nd/3rd ect..) costliest tornado of the year under a basic and routine calculation (looking at which numbers are larger than other numbers) or does it fall under original research as no source explicitly states the list?

• Option 1: It falls under WP:CALC as a basic and routine calculation.
• Option 2: It falls under original research as no source explicitly states the list.
• Option 3: Other - Should be described in detail by the editor.

Elijahandskip (talk) 23:06, 19 March 2023 (UTC)

Discussion (tornadoes)

• Option 1 - The way I look at this is similar to how Wikipedia charts are able to sort information. A really good example is List of the deadliest tropical cyclones. The charts allow the reader to sort the chart in increasing or decreasing order in terms of how many deaths occurred. Each of the deaths are sourced, but one source does not specifically state whether X event was deadlier than Y event. The computer just sorts the numbers as asked to by the reader. In this circumstance, no source says whether X damage total was higher than Y damage total, but the reader can visually see that $5 is greater than $1. So I believe this falls under routine calculations. Elijahandskip (talk) 23:06, 19 March 2023 (UTC)
• Option 2/3 - But in a way that can be fixed through the in-article wording. If the NOAA doesn't provide a damage estimate for all tornadoes, and doesn't itself publish a list of costliest tornados, it would be OR to produce a ranking that makes it seem like there are estimates assigned to all tornadoes. It could be fixed, however, by framing it as something more like "among the tornadoes NOAA published a damage estimate for, this was the second costliest of 2022". — Rhododendrites ^talk \\ 00:09, 20 March 2023 (UTC)

What about in infoboxes like Tornado outbreak of March 5–7, 2022#Macksburg–Winterset–Norwalk–Newton, Iowa, which was the costliest tornado of 2022 and has a news article backing that up as well. Instead of it being mentioned in the article, it is a comment following the damage total in the infobox. Elijahandskip (talk) 01:01, 20 March 2023 (UTC)

Looks fine to me. If a secondary source has gone through NOAA data and made the superlative claim themselves, I don't think anyone here would complain about including it? — Rhododendrites ^talk \\ 01:38, 20 March 2023 (UTC)

Also a side note: Rhododendrites all tornadoes technically get a damage estimate, but some are left as "$0" which obviously isn't accurate. A good example is the 2021 Western Kentucky tornado. NOAA breaks up their tornado reports per county. The tornado was on the ground for 165 miles, so it crossed through a lot of counties (I think 15 in total). In the tornado reports, only 1 county has a damage total marked for the tornado that isn't "$0". So its official damage total is >$25,000. Obviously the tornado caused way more than $25,000 (destroyed thousands of buildings), but the damage total hasn't been formally finalized yet. That's the problem. Your wording 100% fixes the error that can be given to a reader, especially since not every tornado will have a true damage estimate. I will note though, per US Law, NOAA is the only source of official US weather data, meaning even if the damage total is marked at "$0", that technically still is an official damage estimate. But I do agree saying something that let's the reader understand that not every tornado has a finalized damage total is the best course of action. Elijahandskip (talk) 01:21, 20 March 2023 (UTC)

• Question Does the NOAA have a set of guidelines they use when determining what is/is not likely tornado damage? Alternatively do they at least have a dedicated division to compute these figures so the same people are likely to compute these figures for all the tornadoes in a year? IMHO that's the key. If there's something in place to ensure that the same criteria was used to compute damages from a tornado in Florida as there is from one in Minnesota that happened 6 months apart, I don't see the problem. However, for example, if the NOAA were to simply ask the individual states to provide these figures, there's likely not consistency in how those figures were computed. Dave (talk) 00:26, 20 March 2023 (UTC)

Moabdave, let me try to answer some of your question. Yes, NOAA surveys storm damage to determine what is from a tornado or from straight-line winds and such. All the tornado reports are put together by the National Weather Service's local branch for that area and every tornado report can be found on the Storm Event Database (currently contains data from January 1950 to December 2022). For a reference of how a typical tornado report looks, here is one for the deadliest tornado of last year ([2]). Damage totals are split by "Property damage" and "Crop damage". In this case, the report says the tornado caused $75.00M or $75 million in property damage. That is the same for tornadoes dating back to 1950, which is when official US Government records began on tornadoes. Wikipedia has articles for every year (Tornadoes of [year]) with information going back through 1950, with articles being made for years before 1950 as well. Hope that answers your question and helps answer anyone else who had a similar question. Elijahandskip (talk) 01:13, 20 March 2023 (UTC)

That's the only concern I see with it. IMHO, Option1 is fine provided there is consistency in how the figures are computed. I read the concerns so far raised by others and accept those as well, but from what I read they can be addressed by a footnote to clarify the scope of the data. Dave (talk) 16:44, 20 March 2023 (UTC)

@Moabdave: the answer isn't actually that clear. While the reports may have damage totals listed on the surface, the summary text (if one is even given) may say that the damage encompasses the entire event. In some cases, the damage might be tabulated in another listing's summary but not added to the specific event file. The database seems clear-cut at a glance, but there are nuances to understanding the human errors it is riddled with. When it comes to total damage from an event, namely the billion-dollar disaster list, everything is lumped together. ~ Cyclonebiskit (chat) 02:52, 21 April 2023 (UTC)

• It's not original research to say that one number is bigger than another number. We should of course make certain that all the sources use the same methodology and cover comparable areas and time periods. When that's done, WP:CALC is satisfied. But there are other policies and guidelines that might argue against including this information on Wikipedia.—S Marshall T/C 00:38, 20 March 2023 (UTC)
• Option 3, per Rhododendrites. The information itself might be fine, but the presentation needs clear explanations. As an example, it seems apparent from the linked list that these are very broad estimates. A secondary source talking about how those values are determined would be good, as well as a list clearly saying it's a list of NOAA estimations rather than presenting itself as the actual cost. CMD (talk) 02:31, 20 March 2023 (UTC)
• WP:CALC permits editors to say that one number is bigger than another number. Having endorsed #1, I add that outside of the specific, simple context here, there are many pitfalls waiting to trip up the unwary editor. One would need to exercise caution when comparing numbers from different sources, different methodologies, or that might otherwise be non-comparable. It's okay to say that NOAA says this one in 2022 is $25K and that one in 2022 is $24K, so this one is bigger. It's not okay to to say that NOAA gives US$25K for a tornado in Texas in 2009, which is bigger than the estimate from a private insurance company of NGN$11.5 million for a tornado in Nigeria in 2023. WhatamIdoing (talk) 03:09, 21 March 2023 (UTC)
• Option 1/3, an editor can absolutely state that one number is bigger than another, but it needs to be framed correctly as per comments from Rhododendrites and WhatamIdoing. -- LCU ActivelyDisinterested ∆transmissions∆ °co-ords° 19:59, 24 March 2023 (UTC)
• Option 1 with the caveat that we don't mix sources using different methodologies. If we are to create the list ourselves from damage estimates, we need to be explicit that all of the estimates come from one organization, and use them consistently, and be explicit in the text what that singular source is. --Jayron32 12:30, 27 March 2023 (UTC)
• Option 2. There are multiple issues here (some of which can be fixed by tweaks as described above, and some of which can't.) One issue is that not every tornado is tracked by NOAA. In practice I suspect that it is unlikely they would miss a tornedo that would be in the list, but it's still possible. A more serious issue is that a list like this carries the implication that the "most damaging tornado" or "Xth most damaging tornado" is a significant and recognized status, and that these rankings are a useful and meaningful way to examine them. WP:CALC is for mathematical calculations, not for how information is presented, and in particular I'm skeptical about using it for comparisons, which inevitably carry implications related to the items chosen for comparison. (To respond to a comment above, I absolutely do believe that in many situations it is OR / SYNTH to say that one number is bigger than another, since the selection of numbers to compare can be a form of research or synthesis and can carry unsourced implications.) --Aquillion (talk) 09:29, 30 March 2023 (UTC)

Just as a comment: Speaking honestly, yes, the “Xth most damaging tornado” actually does carry a recognized significance that is mentioned both by NOAA and media RS as well as academicly published papers. For example, NOAA has a page specifically on the 10 Costliest U.S. Tornadoes. One of the NWS branches, NWS in Norman, Oklahoma, had a list of the Top Ten Costliest Oklahoma Tornadoes. Numerous RS contain similar things as well. The Weather Channel, NBC News, & a key one being KCRA, which specifically states the costliest tornado of 2022. So in terms of coverage, the costliest nature is 100% a factor used for comparison. Elijahandskip (talk) 12:55, 30 March 2023 (UTC)

If secondary sources cover the costliest tornadoes in a particular way, then we should use those sources with the timeframe and categorization that they use. But I'm skeptical about editors pointing to one arrangement of data to justify a different arrangement of data with potentially different implications. It looks to me like what most RSes cover is the costliest tornadoes of all time, so why not stick to that, and only list costliest tornadoes per year when we have a secondary source showing that a list for that year is relevant? --Aquillion (talk) 13:21, 30 March 2023 (UTC)

NOAA is the source for those secondary sources. A secondary source saying the costliest tornado of X year is using data from NOAA. Why would it not be ok to just use the NOAA data, rather than wait for a source to use it? Everyone (including the general public) has access to the data through the Storm Event Database. That is where RS get the data. That is where finalized tornado reports go as well. Without really saying it, your reasoning somewhat would deprecate NOAA finalized tornado reports, since 99% of the report would be usable on Wikipedia, but this singular section about the tornado’s damage total would not be usable. For instance, in the database, anyone can sort it however they want. In this example, I sorted it to be all time. Would that count as a source for the costliest tornadoes of all time? One could argue that based on the specifications, the U.S. Government just specifically said this tornado’s report caused more damage than this tornado’s report, since it is arranged in that order by their computer system. Where do you think the media get their data? They go to the interactive database, hit sort based on what specifications they want, and boom, they have their data. If an editor needs to wait for RS confirmation, then effectively, NOAA tornado reports would need to be deprecated, aka the U.S. government be deprecated. Elijahandskip (talk) 13:44, 30 March 2023 (UTC)

Why would it not be ok to just use the NOAA data, rather than wait for a source to use it? Because primary data must be contextualized by a secondary source for us to extrapolate meaning from it, because content must be not only verifiable but also comply with NOT, and because we cannot imply a conclusion not directly and explicitly supported by the source. JoelleJay (talk) 02:25, 20 April 2023 (UTC)

By that logic, the US Government (including the White House) can't be used as a source. NOAA is the US government, so we are effectively depricating them by saying we cannot use them as a source and must wait for another source that isn't the US Government to say it. Elijahandskip (talk) 02:31, 20 April 2023 (UTC)

I mean, that depends on the current office /s It doesn't have to be that extreme... We can use the primary sources in order to present information, but we cannot use it to make our own categorizations (such as "list of costliest"). We can use NCEI sources and say "X tornado caused Y dollars in damage" because that's what the source says. We cannot use NCEI to say "Y tornado was the costliest in [year]" because the database does not say that, it simply lists data without providing context. Applying our own interpretation is the primary issue here, not the usage of NOAA sources. ~ Cyclonebiskit (chat) 02:19, 21 April 2023 (UTC)

Technically option #2 Maybe #3 except #3 is unclear. But I think math is not the main issue. NOAA is weather people not economics people and so so making an unattributed statement on economics based on derivations from NOAA data is too much of a stretch. If it was attributed/ explained like "According to NOAA damage estimates....." then it's down to just the math issue and I think that it would be OK. Maybe that's option #3, but option #3 is unclear. Sincerely, North8000 (talk) 15:26, 30 March 2023 (UTC)

• Option 2, Aquillion makes an excellent point. JoelleJay (talk) 02:12, 20 April 2023 (UTC)
• Option 2 – I've been on both sides of this for years and years, and Aquillion's point pushes me to one side. The recent influx of costliest tornado lists to the yearly articles is concerning. We can compile a list of certain values, but we can't impose meaning upon it that isn't presented. We had a rampant issue with "records" being interpreted by users through the National Hurricane Center's database (HURDAT) years ago when no context was given to many of these records. They were just cherrypicked pieces of information with no corroborating source outside of the user's interpretation of the database. Because it breached issues with OR we axed any mentions of records without accompanying text sources clearly stating the information. This same logic needs to be applied to severe weather.
  The NCEI database being brought up by Elijah has some quirks that need to be take into account. For a large chunk of it in earlier years, there are not actual damage estimates rather there are categories for damage ranges. The automated system converted the damage category into the lower bound dollar value and lists that as the damage caused by a specific event. I'm uncertain off the top of my head when the switch to tabulated/estimated values took place, but the methodology within the database itself is inconsistent. The damage estimates are also not available for every event as the NCEI reports are published monthly on the period 3-4 months prior—these reports are imported from the 122 branches of the National Weather Service, not made by the NCEI itself. After the initial publication of NCEI reports, information is rarely, if ever, updated. There's nuance to using/understanding the database and where its information comes from, and making such broad claims of "costliest" is inappropriate without secondary sources. ~ Cyclonebiskit (chat) 02:19, 21 April 2023 (UTC)
• Option 2 – Agree with reasoning given by Cyclonebiskit. Knowing the massive number of inconsistencies as a result of basic human error over the many years of maintaining NCEI Storm Data, I'd be reluctant to use it in so much as an elementary school research project. When we're talking about this costliest tornadoes list, I am in agreement that these lists are concerning given the lack of outside sources and knowing the likely errors in computing the NWS totals in the first place. United States Man (talk) 02:27, 21 April 2023 (UTC)
• Important Comment: For those wishing to remove NOAA damage totals from NCEI, please remember that all tornado damage totals come from NCEI, therefore Wikipedia would no longer accept any information about a tornado's damage total since it all comes from NCEI. Elijahandskip (talk) 02:32, 21 April 2023 (UTC)

I don’t think this has anything to do with the issue at hand. The discussion is on whether it is okay to assemble a separate list from the totals that isn’t explicitly published, not whether the damage totals can be used at all. I believe the totals can be used no problem, but the issue is that you can’t take the totals and assemble your own list. United States Man (talk) 02:49, 21 April 2023 (UTC)

(edit conflict) I don't get where these extreme swings in commentary are coming from, but no one in this discussion has suggested removing usage of NOAA. They're saying to not apply meaning beyond the simple statement of "X is the damage total". And for what it's worth, we don't exclusively use NCEI for tornado damage. We're not bound to NOAA when it comes to sourcing information, we can use other reliable sources to expand upon the topic. ~ Cyclonebiskit (chat) 02:52, 21 April 2023 (UTC)

• Option 2 I've never liked the idea of putting in a chart like this. Many tornadoes don't receive damage figures, while others have incomplete figures. There are too many holes to include it in the article. ChessEric 07:12, 22 April 2023 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Is constructing lists from multiple unrelated sources WP:OR?

With this I am referring to two types of lists:

1. Lists with objective inclusion criteria, but which contain subjective information
2. Lists with subjective inclusion criteria

For an example of #1, List of wars by death toll. It uses different sources for each entry, and each of the different sources uses a different method; some rely on ancient sources which typically provide inflated counts, while others rely on more modern sources - but even more modern sources can differ in methodology. Is it synth to use these different sources, with wildly different methodologies, to tell the reader that more people died in the Greco–Persian Wars than in the Wars of the Roses?

Another example of #1 that is possibly less clear would be List of wealthiest religious organizations. It uses more modern sources, whose methodologies won't differ so wildly, but the sources are still unrelated. Is it synth to use different articles, from different publishers, to tell the reader that the Catholic Church in Germany is wealthier than the Catholic Church in France?

For an example of #2, List of massacres in France. It uses a variety of different sources to determine whether an event should be called a massacre; is it synth to use different sources to tell the reader that Siege of Avaricum, Massacre at Béziers, St. Bartholomew's Day massacre, and Marseille bar massacre are all comparable and classified under the same definition of "massacre"?

BilledMammal (talk) 02:04, 20 March 2023 (UTC)

There's generally no problem with using multiple sources to build out a list, as long as all the sources are generally reliable. Eg: there's no issue with using an academic journal to list the death toll in one war, a book for another, and a modern-day newspaper article for yet another, as long as all three are generally reliable. I think there can be an issue with the last example, where the definition of the list includes terms that can be taken subjectively, and that's where there must be clear reliability on the sources. Masem (t) 02:09, 20 March 2023 (UTC)

For #1, what if the academic journal, the book, and the modern-day newspaper all use different methodologies? If there is a widely agreed upon methodology to calculate a number then I would consider that similar to a list of French writers - an objective list that a reliable source would assemble, if they had an interest in doing so - but if there is not, if different reliable sources have different methods of coming to their own conclusion, then I am concerned that we are producing a list that no reliable source would ever assemble and that makes statements that no reliable source would ever make.

The same goes for #2; when no reliable source has placed two items in the same subjective categorization, is it appropriate for us to do so? I see that List of video games considered the best takes a novel approach to this; they require that six reliable sources consider a game to be the "best/greatest of all time" - in other words, it appears to require that there is a consensus among reliable sources that the included game is very good. It might be a good idea to apply this requirement to all subjective categorization lists; require that for a topic to be included in any such list there must be a consensus among reliable sources that the topic belongs in such a list. BilledMammal (talk) 01:46, 21 March 2023 (UTC)

If there is a list definition that is decidedly more subjective or requires more than simple factual statements, it does seem reasonable to ask list editors to require multiple sources, so that one source doesn't create UNDUE inclusion. Masem (t) 02:16, 21 March 2023 (UTC)

I've opened a discussion on that; Wikipedia talk:Stand-alone lists#Creating minimum inclusion criteria for lists involving subjective categorization.

I'll think more on #1, as I'm still concerned that we're engaged in WP:SYNTH and in the process making incorrect statements that no reliable source would make. BilledMammal (talk) 02:40, 21 March 2023 (UTC)

I think your description of these as containing "subjective information" does not represent your examples. Subjective means that your view depends on your personal experiences/beliefs/values. For example: Is it good or bad to have teenagers wear school uniforms? One person will say "It's good, because then rich kids aren't showing off so much, which made me feel like we were all equal in the classroom." Another person will say "It's bad, because my school uniforms were always ugly and I wanted to be able to express my individuality." Neither of them are wrong; it depends on what "the subject" thinks.

In the lists you've mentioned, we're not talking about subjective information. We're talking about different sources counting to the best of their objective abilities. We're not going to get differences based on personal experience or identity; we're going to get differences based on newly discovered information or specific limitations (e.g., only battlefield casualties vs population-wide excess mortality attributable to the war). Some of this can be handled by providing a variety of estimates (high, middle, low numbers) or by adding a note ("called the best, but only on his mom's Facebook page").

Overall, I offer this advice: Whatever you do, try not to break pages like List of alternative rock artists and List of anti-war songs. Musical genres are blurry (which is not quite the same thing as being subjective), and whether a song like "Turn! Turn! Turn!" is sufficiently anti-war to be included could be debated. WhatamIdoing (talk) 03:42, 21 March 2023 (UTC)

"Subjective" might have been the wrong word, but the differences aren't just due to newly discovered information or specific limitations (e.g., only battlefield casualties vs population-wide excess mortality attributable to the war), but also due to different methodologies - this is particularly clear in List of wealthiest religious organizations where the wealth of the Sree Venkateswara Swamy Temple is determined through public declarations by the temple, the wealth of the Catholic Church in Australia is determined by assessing property values in New South Wales and extrapolating the value across the country, the wealth of The Church of Jesus Christ of Latter-day Saints is determined by a whistleblower, the wealth of the Seventh-day Adventists is determined through an undisclosed method of estimation, etc.

It is this difference in methodologies that I am concerned causes WP:SYNTH issues, because no reliable source has said that the Sree Venkateswara Swamy Temple is one and a half times as wealthy as the Catholic Church in Australia, and I don't believe any reliable source would unless they have used the same methodology to estimate their wealth.

I don't know enough about pages like those to comment, and will leave them to more educated editors. BilledMammal (talk) 03:54, 21 March 2023 (UTC)

This last comment is very helpful for me, because I don't think "subjective" is in any way the "right word" here. I would observe the following:

(1) a list (or in-text comparison) that brings together sources that use multiple, different methodologies would generally fall afoul of WP:SYNTH;

(2) A list (or in-text comparison) that brings together multiple sources using essentially similar methodologies would in general not represent SYNTH;

(3) reading sources to determine how they report their own methodologies is an appropriate role for editors as they collaborate to establish a page-level consensus (which, in this context, could be "the sources are essentially coherent", "the sources are essentially incompatibile", "a subset of sources are coherent" or "the domain is too fundamentally contested to assess the available sources", among other possibilities).

These observations are entirely orthogonal to whether or not "subjective" criteria are used - public opinion polling, for example, can reach findings about "subjective" beliefs that are in themselves somewhat robust, based on consistent methodologies. Meanwhile, competing and incompatible estimates can be developed - using differing methodologies - of phenomena that in themselves are not "subjective" at all, such as the mass of astronomical objects. Newimpartial (talk) 18:34, 21 March 2023 (UTC)

You've raised a concern about making unsourced statements, and I don't think that's what's happening here, at least not in the list that I looked at (List of wealthiest religious organizations). That list is a collection of reliably sourced information that could exist anywhere in the encyclopedia, and doesn't draw any conclusions other than the data provided. For example, if a reader can make a conclusion from the list that church A is twice as wealthy as church B, they can make the same conclusion from the individual articles as well. So are those wealth estimations reliably sourced enough to be included in their respective articles? If so, then I don't see much of a problem including them in other articles, including lists. If they're not reliable, then inclusion is problematic everywhere, not just in that list. Orange Suede Sofa (talk) 23:04, 21 March 2023 (UTC)

The difference I see is that with the inclusion of the estimates in the individual articles we are not inviting and encourage comparison; we are not saying that these figures are comparable. When we put them in a list that is what we are saying. BilledMammal (talk) 12:42, 22 March 2023 (UTC)

This type of list is common, and what should be done in the lede or pre-list is to explain, to the reader, how entries are included and in such a case, while there is no common scale or calculation used, the RSes that support the number are given within the list table. If there is one most authorative source but known to have gaps, the list can explain that most entries are to that authorative source while other entries are slotted appropriately using data from other RSes. Some brief explanation of how the list is assembled, set at the reader level. One example of such is the prose right before the table on List of biggest box-office bombs. Masem (t) 12:47, 22 March 2023 (UTC)

I think that differing methodologies count as "specific limitations". The way to present that is to add explanatory notes. It's fine to for someone to see "Alice Church 100 (extrapolated from real estate prices)" next to "Bob Church 150 (per disgruntled ex-employee)". What you don't want to see is "Alice Church 100 – Bob Church 150", with no hint that these estimates are not comparable. WhatamIdoing (talk) 02:34, 23 March 2023 (UTC)

Most lists are technically WP:OR taking the policy literally and by itself. . The creator has created a topic which (particularly for compound criteria lists) is per se not covered by RS's. Then, the applicability of the title of the list to the entry has no straightforward sourcing. But the overall Wikipedia system does not treat them as OR, so they aren't. North8000 (talk) 19:08, 21 March 2023 (UTC)

• As is often true on WP… there is no “one-size-fits-all” answer to the question. Such lists certainly CAN be Original Research… but they are not always Original Research. You have to look at the specific list (and its sources) to determine whether it is OR or not. Blueboar (talk) 19:39, 21 March 2023 (UTC)

• If no reliable source has ever included such a list, it is OR to generate the list. A justification that might work is if the entries are all bluelinks and your list is a navigation aid. —SmokeyJoe (talk) 23:09, 21 March 2023 (UTC)

• I agree with Blueboar that there's no single answer to this question. In relation to List of wealthiest religious organizations, there are issues that should be addressed, including the methodology used to assess wealth and the date on which the wealth was assessed. My concern is not so much OR as whether the list is accurate.
  In practice, we have sometimes distinguished between list articles and categories, with a less strict approach to the latter. Whether this is justified is another matter: is Category:Anti-war songs any different from List of anti-war songs? Peter coxhead (talk) 11:30, 22 March 2023 (UTC)

• We expect people to base Wikipedia articles on multiple unrelated sources, and we usually require it. Where all the sources are related, that's potentially a POV problem. In fact, writing proper Wikipedia content involves properly educating yourself about the topic by reading a variety of good sources in a critical and reflective way. We research the topic. The reason it's not original research because this process is entirely derivative of published work by others.
  This train of thought gives me a set of principles for the construction of lists. To me, it's clear that you can base lists, or any other content, on multiple unrelated sources and in fact wherever possible you should. There's a specific challenge with comparing numbers because we want to know those numbers are properly comparable -- were they calculated using the same methodology? Did the studies cover the same period? The same geographical area? The same population? Where you don't have that information, but you still want to construct a list, I think you need to disclose all the potential problems and inconsistencies as clearly as possible.—S Marshall T/C 10:09, 23 April 2023 (UTC)

  One way to construct lists based on multiple sources can be seen at List of countries by GDP (nominal) per capita; rather than merging the figures together and suggesting that the value for Cuba from the UN is directly comparable to the value for Montenegro for the IMF it makes it clear that the sources are different. I think this is what you are suggesting; when the sources aren't equivalent we make it clear to the reader that they are not. I think it would be worth adding a paragraph to Wikipedia:Stand-alone lists#Content policies requiring this?

  However, I feel this is only applicable to numerical lists; for categorical lists where the criteria for inclusion is debated by reliable sources I still believe we should follow WP:DUE; further discussion can be seen here. BilledMammal (talk) 14:15, 23 April 2023 (UTC)

  Categorical lists are for helping encyclopaedia users find content. They have their own rules at WP:CLN, but the key point is that they're doing the job of our encyclopaedia's index. You say "the criteria for inclusion", and I don't think that's the right way to think about them at all, because these subjective criteria that we're talking about will very rarely be mutually exclusive. So you allow all the different lists with different inclusion criteria. To take a trivial example: I could make a List of Star Wars movies with nine entries, and I could make another Complete list of Star Wars movies with however-many entries, and each would meet some users' needs, so both should exist.—S Marshall T/C 22:58, 23 April 2023 (UTC)

Social media account

If a person has a well-known social media account, does it count as "original research" to refeer to that account about a claim regarding themselves? Felixsj (talk) 14:27, 24 April 2023 (UTC)

• It depends on what you want to say. Are you talking about a direct quote? Blueboar (talk) 14:59, 24 April 2023 (UTC)

Translation (?) of song lyrics

Please see this discussion and chime in if you please! Do we need a clearer guideline re: song lyrics? --SergeWoodzing (talk) 20:11, 1 May 2023 (UTC)

Recent addition - maps, charts, etc.

• So… apparently there was an RFC (which I missed) concerning maps, charts and similar sources… and based on that RFC a new paragraph has been added. Having now read through it, I don’t object… BUT… I do have a concern:

This policy clearly tells editors NOT to engage in analyze or interpret sources themselves… however the new paragraph says that “routine interpretation” of maps, charts, etc is OK. That is problematic. It sounds like we are carving out an exception to NOR for certain types of sources (and I don’t think that was the intent).

Could we either clarify what “routine interpretation” consists of, or use a different word than “interpret” to convey what we are trying to say. Blueboar (talk) 11:30, 17 May 2023 (UTC)

For the record, you did know about it. There's a comment signed by you on April 19th at the RFC at Wikipedia:Requests for comment/Using maps as sources. Do you really want to re-open this? It was debated ad-nauseum by dozens of editors (including you) over the span of several weeks. The debate is over 400k of text and got quite spirited. That included arguments over the specific word intepret, with your exact argument being made and responded to. I don't think anybody got the exact outcome they wanted, but such is the joy of "governing by consensus". Dave (talk) 14:36, 17 May 2023 (UTC)

Um… having checked my contributions for that date (and double checked the RFC) I don’t see a comment by me. Perhaps you have me confused with someone else? Blueboar (talk) 17:08, 17 May 2023 (UTC)

My apologies, March 19th. My bad. Copy-paste from the RFC: Questions - are maps considered primary sources (with the restrictions and cautions of such), secondary sources or tertiary sources? Or some mix of all three? Does it depend on the specific map? Does it depend on the specific information WE are attempting to cite to the map? I’m not sure we can make blanket statements here. There is a LOT of nuance and grey zone when it comes to maps. Blueboar (talk) 13:03, 19 March 2023 (UTC) Dave (talk) 17:23, 17 May 2023 (UTC)

Huh! I actually remember writing that… I guess I did notice the existence of the RFC at one point!… anyway… my concern with the language still stands. I find the words “routine interpretation” confusing. We either need to use another phrasing, or we need further explanation. Blueboar (talk) 19:34, 17 May 2023 (UTC)

@Moabdave: Can you clarify if it was the intent to carve out an exception to NOR for certain types of sources? If not, then Blueboar's point has merit and perhaps we can tweak the wording. BilledMammal (talk) 14:44, 17 May 2023 (UTC)

I repeat, the appropriateness of the word interpret was debated ad-nauseum. I stand by my comments made at the RFC, and you can read the argument I made as to why I think interpret is an appropriate word to use by searching my comments. I don't see how it would help to repeat the same arguments here, just because you want to accuse people who voted against you of bad conduct to try to re-open the litigation. Dave (talk) 15:29, 17 May 2023 (UTC)

Regarding the personal attacks, see your talk page. Regarding the rest, it was a yes or no question; an attempt to determine if it would be possible to find different wording that would meet your intent while addressing Blueboar's concerns. BilledMammal (talk) 15:37, 17 May 2023 (UTC)

The wording was taken right from the closing note of the RFC. I suggest talking to the closing admin if you don't like the verbiage. We've already been through the discussion so there's no need to start it over just because you don't like the outcome. –Fredddie™ 16:40, 17 May 2023 (UTC)

• Simple solution. "Do not reinterpret sources". - Floydian ^τ _¢ 15:27, 17 May 2023 (UTC)
• Would a phrasing like "Routine reading of maps, charts, etc."; the intent is, if a map shows, say, a river called "Nile River", then reporting that fact is not WP:OR. I think the problem is the contradictory use of "interpret" here; a slightly different wording may help alleviate that and keep the spirit of the policy intact. --Jayron32 15:45, 17 May 2023 (UTC)

  Well I was hoping to avoid another 400k of text re-hashing the arguments made at the RFC. But I guess that's futile. My counter point to that WP:OR does not forbid interpretation, but rather puts limits on how to interpret. By other policies we are required to summarize sources. One must interpret a source to be able to summarize it. In fact, summarization is a form of interpretation. If we can't interpret, we can't summarize only regurgitate. Dave (talk) 15:58, 17 May 2023 (UTC)

  Jayron's suggestion would resolve my concern. Blueboar (talk) 17:13, 17 May 2023 (UTC)

  Not really; we're quite allowed to just make Wikipedia better without needing to ask permission, and that includes improving policy pages where the wording is confusing. --Jayron32 18:30, 17 May 2023 (UTC)

  Summarizing and interpreting are two different things. If I wanted to explain to someone how to create a neutral and accurate summary, I would tell them to avoid interpreting. An interpretation is something that could reasonably be disputed. A good summary should never in dispute, even if you disagree with what is being summarized. Shooterwalker (talk) 03:21, 18 May 2023 (UTC)

  The policy now, by RFC, allows you to interpret a map: meaning you can, in a routine and uncontroversial way, read the scales and labels and keys of a map, and figure out what it actually means about spatial truth in the world, and then write that in your own words in Wikipedia. Andre🚐 03:34, 18 May 2023 (UTC)

  The text added is Routine interpretation of such media is not original research provided that there is consensus among editors that the techniques used are correctly applied and a meaningful reflection of the sources , not read the scales and labels and keys of a map, and figure out what it actually means about spatial truth in the world. If you want that specific wording you'll need another RFC. Given the way 2a and 2b both failed, I doubt that is the communities interpretation of the wording. Bit such things can be worked out at WP:NORN on a case by case basis. -- LCU ActivelyDisinterested ∆transmissions∆ °co-ords° 23:53, 18 May 2023 (UTC)

  I'm sorry, I didn't put my text in quotes or mean it verbatim. "interpretation" of "such media," to me, means exactly what I just said. But I am indeed making the argument that we should not mess with the RFC closure without a new RFC. Andre🚐 00:11, 19 May 2023 (UTC)

A strictest enforcement of the most literal interpretation of wP:NOR would require the deletion of the majority of Wikipedia. Things like the RFC result give guidance on the practical interpretation of how we operate. As a side note, the location of the RFC can make it easy to miss. When an RFC is put elsewhere that is about a specific policy page, we probably need multiple prominent notices on the subject policy page. North8000 (talk) 17:29, 17 May 2023 (UTC)

• Having just added this wording as a result of the RFC, if there's a need to change the wording, it would need a strong enough consensus to override the consensus that was obtained in the RFC. And it's not a great look to say "huh, I didn't know about the RFC" when you did know about it and even participated. Andre🚐 03:29, 18 May 2023 (UTC)

  Meh… My participation consisted of one post - written two months ago - that asked a few questions… but yeah… I freely admit that I was mistaken when I said I was unaware of the RFC. A more correct comment would be that I had long since forgotten that the RFC existed. My bad, but not a big deal.

Anyway… moving forward… what do people think about Jayron’s suggestion of simply changing one word: from “routine interpretation” to “routine reading”? This would absolutely resolve my concerns. Do we really need a follow up RFC to change one word? I am willing to go that route, but it shouldn’t be necessary. Blueboar (talk) 13:26, 18 May 2023 (UTC)

I think it is a good change; it matches what I understand the intent of the proposal to be (ensuring that editors are permitted to summarize maps) and the beliefs behind many of the support !votes; for example, Jc3s5h said I believe the kind of interpretation intended by the proposed addition is routine map reading, and fits into the kind of statement that can be verified by any education with access to the map. BilledMammal (talk) 13:33, 18 May 2023 (UTC)

I still believe what I wrote earlier, but wish I had written it "...can be verified by any educated reader with access to the map." I concur with the change. Jc3s5h (talk) 13:38, 18 May 2023 (UTC)

That verbiage was the result of a well advertised RFC that was open for 2 months where dozens of editors participated. You want to override it with an unadvertised discussion between a half dozen of people who didn't like the outcome? In other words RFC's are now meaningless and governing by consensus is replaced by governing by which side refuses to concede. I'm not 100% happy with the results of the RFC either, but I accept it. Dave (talk) 15:46, 18 May 2023 (UTC)

I agree with Dave, the change suggestion is out of order and I oppose it on principle as it would violate the consensus obtained by the RFC. Andre🚐 15:53, 18 May 2023 (UTC)

The question remains if the specific verbiage was approved, or merely a concept that could be expressed by many different possible expressions was approved. I see people asserting that the specific wording is sacrosanct and cannot be changed because everyone voted on it intending it to be exactly as written, but that wasn't how I read the RFC. The RFC reads to me like its to confirm that simply reading and reporting what a map says doesn't violate WP:NOR policy. There's many different ways the same concept can be expressed, and if one of them causes confusion, its better to use words that don't cause confusion. If we can make the wording less confusing and still maintain the meaning of the RFC, why not? --Jayron32 15:58, 18 May 2023 (UTC)

For the record - no, I DON’T want to override the RFC. I want to clarify it by changing one single word. Blueboar (talk) 16:01, 18 May 2023 (UTC)

Two rival wordings were proposed, one with and one without the phrase "routine interpretation". That debate happened at the RFC. Dave (talk) 17:50, 18 May 2023 (UTC)

I am the only editor here who opposed proposal one (although I supported the similar alternative one); everyone else either supported it or didn't !vote. I don't think it is fair to characterize this as people who didn't like the outcome. BilledMammal (talk) 16:06, 18 May 2023 (UTC)

Considering I supported the original proposal, I find Dave's characterization of me as being on the "side refuses to concede". Insofar as I was on a "side", my side already won. Still, I don't think in terms of winning, never have, never will, and Dave's accusation against me is demonstrably false, and quite frankly, insulting. --Jayron32 16:16, 18 May 2023 (UTC)

Charitably I think he was referring to Blueboar. But in terms of the wording and the process, since this wording comes pretty closely out of the RFC close, a new RFC should be started to change the wording unless it's an uncontroversial change. Andre🚐 16:18, 18 May 2023 (UTC)

Well, given that my sole contribution to the RFC was a single post to ask some questions (which I promptly forgot I had even asked)… I don’t think you can accuse me of taking a “side” either… but I’m not upset about it. I try to look forwards, not backwards. Blueboar (talk) 17:23, 18 May 2023 (UTC)

No insult was intended. An alternative proposal was created which removed the phrase "routine interpretation". This proposal was called proposal 1a in the RFC. Again, all this was already litigated in the RFC. The consensus can be judged by comparing the comments from proposal 1 to proposal 1a. This discussion is arguably to nullify the results on proposal 1 and re-instate proposal 1a of the RFC (or perhaps retroactively create a proposal 1b now that it had been declared that prop1 passed, not prop 1a). So I don't think a characterization of "not accepting the results" is inaccurate. Dave (talk) 16:29, 18 May 2023 (UTC)

I'm still bothered by the whole WP:NOTBURO ickiness this "WE MUST OBEY THE RFC AT ALL COSTS" gives me. We must do what is best for the encyclopedia, and if the existing wording is a problem, we shouldn't feel the need to not fix it because there was a vote, which may or may not have even considered the issue being raised today. --Jayron32 16:42, 18 May 2023 (UTC)

I understand your point. I just disagree with it. Literally everything being said here, including the debate over the phrase "routine interpretation" was brought up at the RFC and voted upon. The dislike of the phrase "routine interpretation" was one of the primary reasons for creating proposal 1a. That debate occurred, and the results are in. Dave (talk) 17:10, 18 May 2023 (UTC)

I think that's also an uncharitable straw man. We don't have to obey an RFC at all costs but you need a better reason to IAR than just, "the wording is unclear." I don't think everyone agrees it's so unclear. Andre🚐 17:58, 18 May 2023 (UTC)

Not everyone, in a unanimous sense, but most people in this discussion apparently do. --Jayron32 18:08, 18 May 2023 (UTC)

Comment - what strikes me about this situation is that, if an RfC reached consensus for language that is in some important respect ambiguous, then any attempt later to "clarify" that language cannot take its mandate from the result of that RfC. "Interpretation" is a word that is itself susceptible to multiple interpretations, and to select one more specific synonym on the basis that one or two RfC participants actually meant to restrict the meaning of "interpret" in that way does not seem in line with the RfC process or result.

(This spoken by someone who was uninvolved in that RfC, and whose views on "interpretation" are far too complicated to factor into this particular discussion. As a more general observation, I have found many editors to use a definition of "interpret" that means something like, "read something into a source that I don't", which differs from the way the term is used in the top shelf of the human sciences.) Newimpartial (talk) 16:19, 18 May 2023 (UTC)

• I think this hinges on whether the closer of the RFC found consensus for specific language or consensus for a broader concept when closing. So… I have asked the closer to comment here. Blueboar (talk) 17:15, 18 May 2023 (UTC)
  • The consensus was for the specific language as that was the question of the RFC, however I didn't read that consensus as mandating "interpretation" in the Wikipedia legalese sense which matches the second definition here, as much as the first definition which is plain explaining what the source says. I don't believe it would require a new RFC to adjust that language to something that doesn't have a specific on-wiki meaning or is causing confusion, but I also don't see a need to get hung up on the word interpretation. One word can have two meanings where one doesn't communicate the right thing and the other is okay. What I'm saying is feel free to change it to read or something if you think that will lessen confusion. ScottishFinnishRadish (talk) 18:02, 18 May 2023 (UTC)

    Thanks… that helps. Blueboar (talk) 19:24, 18 May 2023 (UTC)

    This is something I've tried to get across to no avil. Interpret has two distinct meanings. You interpret glyphs on a page when you read a book. You interpret something when you read a news story and put your own broken-telephone spin on it while recanting. - Floydian ^τ _¢ 01:11, 19 May 2023 (UTC)

    Thank you for that clarification; since you did not see a consensus for interpretation in the Wikipedia legalese sense some clarification is required, as otherwise at least some editors will interpret it in that sense. I've implemented Jayron's suggestion for this. BilledMammal (talk) 02:52, 19 May 2023 (UTC)

From a procedural side IMO having the RFC result say to put that text in doesn't necessarily include "don't change a single word for a long time unless there is a new similar scope RFC. So IMO that would not categorically preclude minor changes even shortly afterwards without a similar RFC. But IMO the discussed word change is a major change, not a minor one. "Interpret" is a farther reaching word than "read" and is at the core of what the meaning of the decided RFC wording says. Also, the RFC was very recent. IMO it at this point it would take another RFC to make that change. Sincerely, North8000 (talk) 18:55, 18 May 2023 (UTC)

• I'll note that the single most common objection during the RFC was in regards to this exact issue, so I don't think it would necessarily be contradicting the RFC to change the wording slightly to accommodate that fear. I don't think we need to re-litigate the whole discussion over a nitpick. _{Obviously it being part of my !vote on the matter biases me slightly here, but still} --Licks-rocks (talk) 21:55, 18 May 2023 (UTC)

Dynamic maps

I have to strongly object to further modifications of the text here. Neither Proposal 2a or 2b had consensus, but that does not mean that there was consensus to disallow the use of dynamic maps or referring to the satellite layer (though admittedly, by my own reading of 2b, there was enough negative commentary that disallowing the satellite layer would be a reasonable finding). --Rschen7754 03:06, 19 May 2023 (UTC)

The issue is that there was a consensus not to allow them; given that the current paragraph could be interpreted as allowing them, we need to clarify that it doesn't. To do this I use the word "excludes", not "disallow" - I think this fairly represents that consensus.

If you disagree, how do you suggest representing it? BilledMammal (talk) 03:10, 19 May 2023 (UTC)

The issue is that there was a consensus not to allow them --Rschen7754 03:14, 19 May 2023 (UTC)

(Ec) So are we gonna allow a single user to dictate one of the most important policies on Wikipedia? I for one suggest representing it exactly as the discussion panned out. No consensus = no change. If you disagree, how do you suggest strong-arming your position? - Floydian ^τ _¢ 03:17, 19 May 2023 (UTC)