Wikipedia talk:Manual of Style (dates and numbers)/Archive B11

Convenient links

Anyone can add relevant links to this section as a convenient editing resource. Greg L (talk) 04:59, 14 April 2008 (UTC)[reply]

General IT prefix discussion

The IEC prefixes were approved in January 1999. After nine years, virtually nobody uses them. Esperanto is in wider use. When Steve Jobs introduced the MacBook Air (skinny notebook) at Macworld he did not use the term gibibyte once. The news reports give the RAM size as 2 gigabyte, 2GB or 2Gbyte. The Manual of Style should reflect what the outside world is doing. The computer industry and the publishing world have ignored the IEC prefixes. -- SWTPC6800 (talk) 06:09, 17 January 2008 (UTC)[reply]

Do you have any opinion on the topic that is being discussed here? — Aluvus t/c 13:02, 17 January 2008 (UTC)[reply]

Yes I do. A few peoples here are trying to get the Manual of Style to adopt an obscure method of measuring computer storage. This edit war has been going on for several years. You are arguing over the rules of the edit war. The real question is should the Manual of Style follow a standard that had not reached 1% adoption in the real world after 9 years. -- SWTPC6800 (talk) 14:52, 17 January 2008 (UTC)[reply]

Just let me extract the interesting parts of your text: "few people", "an obscure method of measuring computer storage", "war", "1% adoption", "real world". Hopefully people will realize how pointless this discussion is. Don't waste your time, don't let them trick you. As long as this topic is under tight control of certain individuals, you can't win. Again don't waste your time with facts. They shall be ignored, absolutely, without remorse. --217.87.122.179 (talk) 05:57, 2 February 2008 (UTC)[reply]

You are wrong and do not attempt to misrepresent other editors with your incorrect anonymous rants about "certain individuals". Fnag aton 10:03, 2 February 2008 (UTC)[reply]

Whether manufacturers are using these units is irrelevant. Fact is that the units are used inconsistently. Sometimes they have a binary meaning, sometimes a decimal one. Many of the readers wil not be aware which meaning is applicable in a specific mention of the unit. Therefore it is the task of an encyclopedia to make sure the reader is able to draw the correct conclusion. This can be achieved by always adding a conversion to (or a confirmation of) a standardised and well documented unit. We achieved consensus to do it that way a while ago. According to the guidelines units from a primary source should come first. So the usage would be:

with decimal meaning: 64 MB (61 MiB)
with binary meaning: 64 MB (=MiB)

−Woodstone (talk) 19:42, 17 January 2008 (UTC)[reply]

Fact, KB, MB, GB are defined by standards organisation to be power of two. Fact, some manufacturers use other meansing. Lets say for the sake of argument some manufacturers started using KiB but in a decimal sense, that would be inconsistent use, so what then? It's not *always* needed though, just disambiguate (if you need to at all) the first occurrence in an article. Fnag aton 22:18, 17 January 2008 (UTC)[reply]

Regardless of how inconsistent sources are, a conversion to standard units would always tell the user unambiguously what the correct meaning is. I agree that if the same number is used several times, one conversion might be enough. But it is just a welcome service to the reader to convert every number at least once. −Woodstone (talk) 22:31, 17 January 2008 (UTC)[reply]

I'm just going to quote "Fact is that the units are used inconsistently" and "Regardless of how inconsistent sources are" then grin for a while because I find it funny. ;) Seriously though the JEDEC, who are the standard organisation for memory and who have majority consensus, do define kilobyte etc in their standard. So which standard body is the better one to choose, the one that has a tiny 0.5% use (no consensus) in the real world or the one that has the huge majority consensus? And then if there is any use that differs from the JEDEC standard then make a point of disambiguating it with exact numbers of bytes. Fnag aton 22:56, 17 January 2008 (UTC)[reply]

I see your point about fun: it is funny: the more inconsistent usage is, the more there is a need for conversions. You may know exactly in which cases a particular interpretation is standardised by JEDEC, most people have never even heard about JEDEC. Why not help them by addding a conversion? −Woodstone (talk) 23:03, 17 January 2008 (UTC)[reply]

Because according to the JEDEC it's not inconsistent, it is powers of two in size. Following on from the above I could point to companies using KiB but in the decimal sense. Then I could say "inconsistent" and then you'd have to drop pushing for KiB to be used, to ape the arguments used by some IEC supporters for example. Pushing for a particular style of prefix isn't actually the point, as we'll see in a second. What happened is that some other "standards organisation" came along and invented a new term, but it's not used except by a microscopic minority. However the question isn't about what standards organisation is best or whatever, no matter how tempting that is it doesn't actually tackle the real issue and it just causes people to sit behind their preferred camp. Remember you cannot say IEC is consistent since it has been shown IEC is inconsistent with the consensus in the real world. Also remember you cannot say IEC is not ambiguous because of the companies that use KiB in the decimal sense and also because JEDEC define it to be not ambiguous. The question is, why advocate something that only has a tiny 0.5% support, i.e. doesn't have consensus? Fnag aton 23:09, 17 January 2008 (UTC)[reply]

Well, a kilobyte appears to be 2 raised to the power 10 (=1,024) in most systems and one megabyte can be either 1,000 kilobytes or 1,024 kilobytes; it gets a bit difficult to determine how many bytes there are in a particular wiggit. Some system is need to sort it out. I've lost track of the arguments, what's your proposal to sort it out - I don't think context is a valid approach.Pyrotec (talk) 23:45, 17 January 2008 (UTC)[reply]

The most common use being power of two. The only non-ambiguous way which also doesn't use neogolisms, i.e. is consistent with consensus, would be the following: Follow JEDEC or be consistent with the sources relevant to the article. If something uses JEDEC specified prefixes but in a non-standard sense then use the units found in the source but disambiguate by stating the exact number of bytes in brackets. Otherwise (and rarely) if some other units are used (like IEC) in the article due to being consistent with sources then disambiguate with JEDEC. Fnag aton 00:00, 18 January 2008 (UTC)[reply]

Or we forget it all for now and just make sure the guideline stays as it is in its stable state. Fnag aton 00:02, 18 January 2008 (UTC)[reply]

There is some confusion on binary versus decimal units of computer storage. However the computer industry and the technical press do not think is significant enough to change to a new units system. Fnagaton is very generous to say that the IEC prefixes have 0.5% acceptance. The industry treats kibi and gibi as something the IEC made up one day. It is not Wikipedia's mission to promote a failing standardization effort. If the IEC binary prefixes gain significant support Wikipedia could consider using them. -- SWTPC6800 (talk) 01:55, 18 January 2008 (UTC)[reply]

Now that's one horrible misuse of a quote. Did you think it applies because the headline looks fitting? I suggest you actually read this policy. It's also not legit (and I don't need any policy to back this up) to use arbitrary made up numbers like "0.5%" in a discussion. --217.87.122.179 (talk) 22:03, 2 February 2008 (UTC)[reply]

0.5% is not arbitrary and is not "made up".

Historical use search terms	Results
kilobyte -wikipedia	1,940,000
megabyte -wikipedia	6,190,000
gigabyte -wikipedia	3,640,000

Total: 11,770,000

IEC Search terms	Results
kibibyte -wikipedia	28,800
mebibyte -wikipedia	17,100
gibibyte -wikipedia	19,000

Total: 64,900

Consensus for historical use: 99.449%

Fnag aton 22:51, 2 February 2008 (UTC)[reply]

It's good that you explain how this figure was determined. This shows that the value is less than useless. Many results may to a certain extent show that something is widely known. The opposite is not true. Lack of results does not prove anything. Not to mention that this method is still arbitrary because it excludes the short forms which are far more common in my experience. Of course it's obvious that MiB has more meanings than Mebibyte and MB has also many meanings in different areas. Last but not least, this method excludes not only results from wikipedia or citing wikipedia, it excludes any kind of result which mentions or links wikipedia which may have nothing to do with this topic. --217.87.122.179 (talk) 00:15, 3 February 2008 (UTC)[reply]

The preponderance of evidence shows that real world consensus is against your point of view. What evidence have you given in reply? Nothing, no evidence, you've just written waffle about "less than useless". The numbers are facts that do not support your point of view so you are wrong again because when you claim "it's less than useless" does not make it so. Wikipedia is excluded for the very good reason that a while ago someone made many hundreds of changes to use kibibyte etc and that means including Wikipedia would contaminate real world results. Also Wikiepedia is excluded from the results to show real world use. You also showed a search earlier on that did "-wikipedia", unless you now want to retract your earlier post? Trying to do a search for the much shorter versions like MiB ("Men in Black" for example) is also much more likely to pickup use of those three letters that has nothing to do with computers so your point is not well thought out because your proposal would have less much meaningful results. Fnag aton 01:07, 3 February 2008 (UTC)[reply]

It is Wikipedia's mission to provide clear and accurate information to the general readership. If unit like MB is met, it is never obvious what quantity this represents. Usage depends on the context (e.g. disk, memory chip, data tranfer). Many people do not know which is used when, and even less what JEDEC is and if it applies to the particular occurrence of the unit. The solution is giving a conversion to a world standard. Even if that standard not often used, it is still well documented and unambiguous—just what is needed to eliminate uncertainty. −Woodstone (talk) 09:33, 18 January 2008 (UTC)[reply]

Woodstone, as explained above what you see as "never obvious" is a red herring because what you call "the world standard" is not actually a "world standard" and it is not unambiguous. The real question is this, why advocate something that only has a tiny 0.5% support, i.e. doesn't have consensus? Fnag aton 12:04, 18 January 2008 (UTC)[reply]

No, the real question is: how can we inform the reader clearly and unambiguously about the meaning of quanties given. Just using MB does not do that, because its meaning is context dependent. So what device are you proposing to achieve the goal of being informative. −Woodstone (talk) 12:41, 18 January 2008 (UTC)[reply]

Like I said above, express the exact number of bytes as disambiguation in brackets. For example 2KB (2¹¹ bytes) is the only unambiguous way of stating the number of bytes without using neologisms and it also shows that in this case it is using the binary context. Fnag aton 13:27, 18 January 2008 (UTC)[reply]

No, "2 KB (2^11 bytes)" is not unambiguous at all because there is always the possibility of mistakes. In this case, I'd assume the editor forgot the "i" and typed KB instead of KiB. Or worse, maybe someone added (2^11 bytes) because he assumed 2 KB was meant to mean 2048 when in reality it's really 2000. You might think it's "obvious" from the context, but context can only provide a rule of thumb. In many cases, the editor was just careless and typed "KB" instead of "kB". Many people don't know that SI prefixes are case-sensitive and that 'K' is incorrect as abbrevation for kilo (1000, one thousand). Thus, a chain of errors and "well-meant" edits can cause the following: 2000 bytes -> 2kB -> 2KB -> 2048 bytes. Same applies to "bits". Last but no least, this kind of hint reinforces the idea that KB absolutely means 1024. Sometimes less is more. If you think writing "(2^11 bytes)" is useful at all, why don't you drop the "2 KB" completely? It's not common practice in Wikipedia to write the same twice in different words. --217.87.122.179 (talk) 21:13, 2 February 2008 (UTC)[reply]

You are incorrect because "2 KB (2^11 bytes)" is not ambiguous and because it expresses exactly the number of bytes that are used. You are also incorrect because your "what if there is a mistake" scenarios are also irrelevant red herrings because as already stated in the guideline "...one must be certain... before disambiguation" and trying to throw out using a particular unit just because someone might edit an article incorrectly is no valid reason at all. Taking your point of view that would mean nobody changing anything because there might be a mistake somewhere. Thus your "many people" point is also irrelevant because if someone is not certain then they shouldn't be editing on a topic they are not certain about. Also changing KB to KiB when in the uncertain "many people" scenario you use is also wrong because the person is not sure what they are doing and could change something incorrectly. Dropping the "2 KB" completely is also not correct since as I have posted before consistency with the terms used in the reliable sources relevant to the subject is important. You are also wrong because disambiguation, "writing the same twice in different words", is very common. Fnag aton 21:24, 2 February 2008 (UTC)[reply]

(unindent) That makes sense. Perhaps more recognisable would be to use only multiples of 3 for decimal and of 10 for binary powers:

with decimal meaning: 64 MB (64×10⁶ bytes)
with binary meaning: 64 MB (64×2²⁰ bytes)

−Woodstone (talk) 16:15, 18 January 2008 (UTC)[reply]

I really like the way this is heading, if it can be made to work. If there were such a guideline, I wonder whether it would actually be followed though. I think it is worth trying. Thunderbird2 (talk) 16:30, 18 January 2008 (UTC)[reply]

Yes using 2 and 10 notation would make it completely unambiguous and it also gives a very good hint as to what format is used for binary or more rarely decimal. Also it lets articles use the type of units that are used in the sources. Which is a bonus since maintaining consistency with sources as well as following the real world consensus is a really important issue for me and I suspect many others think the same. Of course as with disambiguation it need not be everywhere, just so that the article makes sense. Fnag aton 17:00, 18 January 2008 (UTC)[reply]

The suggestion appears to have some merit. Note: powers of three in decimal, e.g. 10^3x, means that everything is rounded in thousands; and powers of ten in binary, e.g. 2^10x, means that everything is rounded in kilobytes (=1,024).Pyrotec (talk) 17:11, 18 January 2008 (UTC)[reply]

I'm not sure how many readers wil be comfortable with this notation. In any case, I think what is being suggested is that for the binary meaning of the prefixes, it should be written as a power of 2, where the exponent is a multiple of 10. For the decimal meaning of the prefix, it should be written as a power of ten, where the exponent is a multiple of 3. --Gerry Ashton (talk) 17:28, 18 January 2008 (UTC)[reply]

I'm not sure whether Gerry Ashton's comment relates to my comment above, or an early one. My comment related to a question by Fnagaton which has since been edited, so it reads differently now & is no longer a question. My interpretation of Woodstone's comment, was a sequence of thousands & kilobytes, e.g 0.02*10³, 0.2*10³, 2.0*10³, 0.02*10⁶, 0.2*10⁶, 2*10⁶, etc, and a similar sequence for kilobytes (sorry I'm too lazy to do the binary sequence in kilobyte sequences). But if someone wants to see it?Pyrotec (talk) 17:43, 18 January 2008 (UTC)[reply]

Yes, of course. Very simple: keep the number as it is (no rounding needed) and convert, depending on context:

KB to ×2¹⁰ bytes or ×10³ bytes
MB to ×2²⁰ bytes or ×10⁶ bytes
GB to ×2³⁰ bytes or ×10⁹ bytes (etc)

Furthermore, it is not very important if all editors follow up on the guidelines. There will always be volunteers that will add proper conversion. Having it in the MOS will hopelfully prevent reversions. We still have to find a way out of the occasional MB = 1024,000 bytes. −Woodstone (talk) 18:03, 18 January 2008 (UTC)[reply]

How about: "This memory stick from company X is labeled as 1MB (1024×10³bytes)" Fnag aton 18:09, 18 January 2008 (UTC)[reply]

How about: "This memory stick from company X is labelled as 1MB (2¹⁰×10³bytes)" (not "*" as above). Rich Farmbrough, 14:03 1 February 2008 (GMT).

or "This memory stick from company X is labelled as 1 MB (1.024 million bytes)" Thunderbird2 (talk) 14:10, 1 February 2008 (UTC)[reply]

No. These are the worst options. In one regard it fails as soon as you get to a billion and especially beyond because US and European billion differ. The next higher magnitude which will be common in 2-4 years (tera respectively tebi) has no layman equivalent. If you write "1 MB (2^30 bytes)" any sensitive reader would assume that 1 MB is always exactly 2^30 bytes. There is no reason to assume a unit would differ depending on context. The solution is very simple, use units correctly or don't use them at all. If a unit has supposedly more than one meaning, it is by definition not a U-N-I-T. unit comes from unity. No unity, no unit. Kindergarten logic suffices. --217.87.122.179 (talk) 06:07, 2 February 2008 (UTC)[reply]

You are wrong 217.87.122.179 because the units are de facto standards used in the real world and Wikipedia is descriptive not prescriptive. To use neologisms that are very rarely used in the real world (<1%) only confuses the matter even further. Fnag aton 09:53, 2 February 2008 (UTC)[reply]

He is right in so far that the IT industry’s adoption of metric prefixes, beginning 40 years or so ago, in a different than standardised sense is the root of the problem. Only that made ambiguity possible. Separate binary prefixes should have been developed back then, but they weren’t, leaving us with the mess.

You are right that Wikipedia is descriptive – in intention at least, by its importance and influence nowadays, being part of the real world, it is defacto quite prescriptive! –, but you are also wrong, because its style guide, unlike encyclopaedic information, has to be prescriptive to some degree. MOS may very well choose to adopt a rule by reason that would not have been derived from observing common usage. IT prefixes used to be such a case, where MOS editors came to the conclusion that it’s better to diverge from common and source usage, adopting an international public standard instead to make the project less ambiguous and more homogenous. Some months ago this changed (after epic-length, tiresome discussions), because some article authors, like you, didn’t like to adapt their habits. The descriptivism myth evolved.

You are also wrong in that you didn’t respond to any of the points the IP user raised; just called him wrong, repeating your weak arguments once again. He does have at least one very valid argument: “There is no reason to assume a unit would differ depending on context.” — Christoph Päper 14:07, 2 February 2008 (UTC)[reply]

The IP user and you are wrong because it is fallacious to try to claim something is "different than standardised sense" just because a so called standards body decides to produce something contrary to what is the de facto standard. You are also wrong because the MOS is not prescriptive beyond what is actually common sense. You are also wrong because I did respond to the "points" the IP user raised, you will note this is the case since I wrote "because the units are..." giving a perfectly valid and correct reason. What you claim is the IP user's "valid argument" is actually shown to be a red herring by the very fact that the units have well known use. You are also wrong because the arguments put forward by me are stronger than what you have put forward and just because you disagree doesn't mean you are correct. Actually I'm giving you the benefit of the doubt here because you have put forward no such counter argument, instead you have attempted to question my motives and claimed a valid logical reason is somehow "repeating your weak arguments" without giving any supporting evidence. You are also wrong in your summary of the history of this topic and I demand that you retract your misrepresentation about what you think my motives are. Fnag aton 18:39, 2 February 2008 (UTC)[reply]

You are reacting on trigger words again instead of reading carefully and responding adequately. First came decimal metric prefixes (which I wrote about), then came bits and bytes (and octets and words), then came binary “adaptation” of SI prefixes in IT – with the side effect of capital k which I welcome –, then came disambiguity, then came IEC prefixes, then came Wikipedia. Where am I wrong here?

There is no such thing as common sense, never use it as an argument. The MOS is not prescriptive in the sense that it would say people what to do, but it’s prescriptive in the sense that it says what articles should look like in the end. (And yes, it’s often watered down in that regard, which I consider a failure.)

The IP users said, a unit with multiple meanings was not a unit by definition. You say that (he’s wrong and) the units under discussion were being used with multiple meanings, trying to prove your but actually also proving his point. His definition can’t be wrong, it just may not match yours.

The (here logical, not empirical) expectations of the readership are quite the opposite of a “red herring” argument in a discussion about our style guide! If the prefixes, when applied to bits and bytes, always were used in a binary sense (and decimal everywhere else), your point might stand, but they ain’t, e.g. in rates (kbit/s etc.).

I tried to limit my response (after I couldn’t suppress it completely), because I think everything has been said on the matter, although people still come to different conclusions, because their views about the role of Wikipedia and its MOS differ. (You try, intensional or not, to devalue the other position by calling it prescriptive and neological, which aren’t bad things per se, neither is de facto.) If I agreed with your presuppositions I would probably come to the same conclusion, because I don’t contest most of the facts you bring up again and again, thinking or at least implying that this was what needs to be discussed when actually we disagree on a higher level, which makes your points irrelevant – weak probably was an inappropriate or misleading term.

You’re in no position to demand anything from me (nor anyone else here), but, please, feel free to correct or at least discredit my summary, my view of the process.

Gee, I wish there was anything besides carnival outside so I hadn’t felt the temper to engage in this again. — Christoph Päper 01:24, 3 February 2008 (UTC)[reply]

You are incorrect because you are at fault here for not "reading carefully and responding adequately" (as you put it). This is because you misrepresent what I write and you attempted to misrepresent my motives in your summary of the history of this topic when you wrote "because some article authors, like you, didn’t like to adapt their habits". You are utterly wrong to try to misrepresent what you think are my motives. You are also wrong when you say there is no such thing as common sense because common sense is a large part of building consensus. Next we get onto you misrepresenting what I write. This is because I point out his "argument" (from 06:07, 2 February 2008 (UTC)) is a red herring and yet you insist on trying to rewrite my point to mean something else entirely. Just so you are perfectly clear, what you wrote is rubbish because I never wrote what you claimed I wrote. You are using a straw man, so your "point" is irrelevant and false. Fnag aton 02:01, 3 February 2008 (UTC)[reply]

I’m sorry it took me so long to respond, but I had much more important, non-WP things to do lately. Actually, I’m just taking a break from them.

So you think I misunderstood you or your motives. Maybe I did, big deal, happens all the time. You wouldn’t even have noticed if I hadn’t paraphrased what I understood and remembered (“misrepresent” you call it), just like I noticed your misreading. It doesn’t help, though, that you don’t explain where and how I misunderstood exactly. I assume it’s only about the changing habits part.

Nobody denies that kilo, mega and giga (at least) with bit or byte are often being used and understood in a binary sense instead of their classic, metric decimal one. Nobody denies that this occasionally results in ambiguity (in any imaginable way). Nobody is happy with the situation, although many take it as a given. Nobody intends a unit (or prefix) to have more than one meaning, although in practice (i.e. your de facto) occasionally one does. The IP user raises this point to a defining quality of a unit, you pragmatically don’t and call this (intentionally) irrelevant to the discussion (“red herring”). Am I right so far?

My main argument, which you ignored, still stands: We have different ideas of the purpose and foundations of this style manual. Therefore we cannot come to the same conclusion! So every argument is moot until we agree on the basic principles.

Whenever you claim the MoS was descriptive not prescriptive you are right and wrong at the same time, because a descriptive observation once published becomes a prescriptive rule in the understanding of many. This is how grammar and orthography “rules” came to be (for most natural languages).

Anyhow, a suggestion for compromise for the time being, although I would much prefer consensus in the long run:

SI prefixes are decimal and do not need disambiguation in general, but when applied to bits or bytes (incl. words and octets) without composition with any other units (as in rates, e.g. kbit/s) their meaning has to be disambiguated (one possibility is the replacement by ~~IEEE~~IEC prefixes, which are always binary), except for RAM chips [and?] when used with a preferred number based on a power of 2 where they take a binary default meaning and only need disambiguation when having a decimal meaning. — Christoph Päper 19:36, 21 February 2008 (UTC)[reply]

I can see you are continuing to misrepresent me because I did not ignore your argument, your "argument" is fallacious for the reasons I have already given above and previously on this topic. Your compromise is not good because it does not represent real world consensus. Your "compromise" chooses IEEE which you prefer, which is not a compromise at all, it is pushing your own point of view. For example it is disingenuous to say "SI prefixes are decimal" and not mention the fact that the JEDEC defines K = 2¹⁰, M = 2²⁰ and G = 2³⁰ when being used for semiconductor storage capacity and also because recent legal rulings have stated that despite what SI/IEEE/IEC claim the de facto standard is different. RAM chips commonly use the units KB, MB, GB in the binary sense, for example and this is defined in the JEDEC standard. If you really want to get into the whole "orthography" argument then you're going to be refuting your own point of view because orthography is to use correct spelling according to what is considered to be accepted (i.e. generally approved) and established use. In this case orthography easily refutes using the -bi prefixes because it is obvious they are not accepted and established use. Fnag aton 08:42, 22 February 2008 (UTC)[reply]

Gosh, either my English is worse than I thought or productive discussion with you (on this subject) is close to impossible.

I don’t care whether we use IEC prefixes for disambiguation. I just want to limit ambiguity inside and among WP articles. (Ambiguity outside WP is outside the scope of this style manual.) To achieve this we can either

adopt something with mutual exclusive meanings (e.g. SI and IEC prefixes, despite ambiguous usage outside WP),
disambiguate (SI prefixes) inline each and every time or
specify in MoS where we assume which default meaning (for SI prefixes), that wouldn’t have to be disambiguated, and where it’s too dubious to choose a default.

The current guideline does nothing to achieve the goal, uses only partially the second option. My suggestion for compromise does the third, although I very much prefer the first. Maybe I set the requirements for a binary meaning of SI prefixes higher than you would like, but with preferred numbers in the field of semiconductor storage capacity most cases would still be covered to your satisfaction. I see no good solution for file sizes, though, because files can be stored on different media (binary or decimal) and can be transmitted (decimal). — Christoph Päper 17:10, 22 February 2008 (UTC)[reply]

I note your continued misrepresentation and use of weak personal attacks, this shows that you are not interested in valid debate. You are also wrong because the current guideline fixes what you think is ambiguous by encouraging the exact number of bytes to be specified in the disambiguation or in footnotes. For example "256 KB (256×2¹⁰bytes)".

The first option "adopt something with mutual exclusive meanings" ignores real world consensus and makes articles inconsistent with their sources and actually doesn't solve the problem of using -bi units that can also be ambiguous.

The second option "disambiguate (SI prefixes) inline each and every time" is not logical since the purpose of disambiguation is not to include brackets (or similar inline text) after every number in an article, it is usually the case that only the first occurrence of any such number needs disambiguation.

The third option "specify in MoS where we assume..." is only going to get my support if it follows the real world consensus, i.e. not use -bi. Otherwise it is just going to be pushing point of view against consensus.

The best option, which you don't list, is to use the terms found in the majority of reliable sources relevant to an article. This means internal consistency for articles over and above consistency for the whole of Wikipedia. Fnag aton 17:33, 22 February 2008 (UTC)[reply]

Like I said, the current guideline, if anything, uses the second option. It’s the worst! It being basically optional doesn‘t make it any better. 256 KB would default to binary meaning by my proposed compromise (in the context of semiconductor storage at least) and thus would only have to be diambiguated if it had a decimal meaning instead.

I didn’t deny that the first option would only be internally consistent. Unlike you I don’t consider this a huge problem. I tried to point that out earlier. IEC prefixes are always binary and thus unambiguous, even if you should find someone using them wrongly.[1]

Read it as “disambiguate (SI prefixes) in each and every article” if you must.

Real world consensus is that IEC prefixes are one way to achieve disambiguity where needed; the real world isn’t just very often unambiguous. For symbols the little i is probably the least cumbersome, least space-taking and – like it or not – most established solution. If you don’t like the terms “kibibyte” etc. – I don’t – you may use “binary kilobyte” or something like that where needed.

What you call a best (or fourth) option is not solving the problem at all, because it would mostly result in SI prefixes being used with varying meaning. Every reader would have to make a (hopefully educated) guess. My proposal was to provide a rule of thumb for that guess in the MoS. I already tried (without success) to discuss our different views of the scope for consistency. — Christoph Päper 10:11, 11 March 2008 (UTC)[reply]

To tie this thread with the new thread this comment demonstrates why Crissov is still wrong for the same reasons as above. Fnag aton 19:13, 24 March 2008 (UTC)[reply]

The IP user raises only one valid point, that even more confusion will arise when larger memory becomes available. In respect of (computer) memories, in the real world units do change depending on context - read the discussions above: a million can mean 1024 x 1000 and 1024 x 1024 depending whether is a megabyte of storage on a hard drive, memory stick, etc - go and look at amazon.com. The IP user appears to be confused, it is not misuse of units in wikipedia that causes problems it is inconsistent use of units in the computer industry that causes the problems. Wikipedia MOS is attempting to add clarity to the confusion that already exits.Pyrotec (talk) 19:52, 2 February 2008 (UTC)[reply]

I've changed my view upon reflection. The computer industry uses megabytes, Gigabytes, etc, the difference in UK and US definitions of a billion is irrelevant as it is unlikely that billion will be used as a description of the number of bytes.Pyrotec (talk) 20:20, 2 February 2008 (UTC)[reply]

Indeed, that's what I meant when I said earlier the point put forward "is actually shown to be a red herring". It's like saying "the sky is blue and that proves that I'm right about cell meiosis". Fnag aton 20:35, 2 February 2008 (UTC)[reply]

That difference is irrelevant everywhere, except perhaps nursing homes in the UK. Tony (talk) 23:28, 2 February 2008 (UTC)[reply]

Maybe you don't know that "billion" is not only an English word. Billion is still frequently mistranslated as "billion" when it actually means 10^9 in one (European) language and 10^12 in the other like French or German. The UK might have adopted the US meaning by now, that's not true for any other country. You see it's almost the same issue as Megabyte vs. Megabyte. One is 10^6 and the other is 2^20. Keep in mind that this neither the American nor the British Wikipedia, it is an international effort. As there is no official authority for the English language, really nobody can decide which meaning of billion is correct but it is trivial to avoid these few well-known sources of confusion. --217.87.122.179 (talk) 00:00, 3 February 2008 (UTC)[reply]

No, you didn't mean that. If you read carefully, the term billion was only a part of argument. Calling it a "red herring" makes clear that you assume bad faith, especially your "sky is blue..." blah blah shows that you are interested in facts or any kind of useful discussion. There actually over 120000 Google hits for '"billion bytes" -wikipedia'. I don't know which formula Pyrotec used to determine his assumed probability but I'd think we can all agree that this isn't the right place for speculation about the future. For the record, a billion bytes is in US-American English equivalent to a gigabyte, that's why it's already in use. I was talking about Terabyte (tera means 10^12) which is less common for now and which has no well-known -illion equivalent. So it'd be called 1000 billion or a million million but nobody knows what's the marketing industry is going to establish. Nonetheless "billion bytes" is actually used despite the assumed low probability. —Preceding unsigned comment added by 217.87.122.179 (talk) 20:56, 2 February 2008 (UTC)[reply]

Yes I did mean that so do not presume to tell me what you think I really meant because when you do so you are misrepresenting me and what I wrote. Also your incorrect accusation of "bad faith" shows you have not presented a valid argument because the term "red herring" is actually referring to the logical fallacy, which your "argument" is actually using. This is also pointed out by Pyrotec as being irrelevant. Fnag aton 21:08, 2 February 2008 (UTC)[reply]

Okay, then I'm relieved that I could help in making clear what you actually meant because it wasn't obvious at least to me. I don't quite agree with your presumed definition of red herring but discussing this would be another one and off-topic anyway. --217.87.122.179 (talk) 21:58, 2 February 2008 (UTC)[reply]

Red herring fallacy also known as irrelevant thesis or conclusion. You'll see what I wrote is correct. Fnag aton 22:43, 2 February 2008 (UTC)[reply]

Not taking either side, but the Google test is worthless. As is citing voluntary standards (as are the IEEE's, IEC's, and JEDEC's) as examples of "policy". Voluntary standards are just that, and it's hardly surprising you can find standards which reflect common usage. Consider:

mile -wikipedia - 24,300,000

kilometer -wikipedia - 1,520,000

inch -wikipedia - 26,000,000

centimeter -wikipedia - 6,800,000

pound -sterling -wikipedia - 11,200,000

kilogram -wikipedia - 8,240,000

acre -wikipedia - 3,420,000

square kilometer -wikipedia - 1,660,000

Can these results be used to argue that preference should always be given to imperial units of measure, since apparently more English-speaking people will be familiar with them? Can they be used to decide on a case-by-case basis which units are preferable? Can you say that common use is universally more important than standardization in every case or vice versa? No matter how you concoct the Google Test argument, it is always flawed.

Experience with Wikipedia should tell you that no amount of dialog on this will ever settle the issue wholly in favor of or against IEC prefixes. There's zero editorial direction on this site, and unfortunately the MOS as a whole is more or less a farce (used merely when it is convenient in an argument), so fights like this will keep breaking out. I strongly suggest you look for a middle ground where the use of IEC prefixes is accepted in some contexts and the common prefixes are accepted in others. Pages like the one on floppy disks really do need some concise method to deal with the prefix ambiguity, and the IEC prefixes are one such way.

Basically, the way the MOS currently reads is the only way it can read. In reality there's no such thing as standardization on Wikipedia, so that argument won't work. The current reading explains the history of the debate, lays out the facts (common use versus ambiguity) and doesn't take any strong position. The only change that might be useful is provision for changing prefixes to IEC variants where appropriate and discussed beforehand. This excludes contentious en masse changes, but still acknowledges that there are some circumstances in which utilization of IEC binary prefixes might be useful. (maybe also remove the mention of the JEDEC, which is utterly comical since their standards have nothing at all to do with standard prefixes for unit measures)

You will never convince each other. You will not be able to defeat the common usage argument. You will not be able to defeat the ambiguity argument. You will never end this debate by brow beating the same tired and cyclical arguments into one another. I humbly suggest just leaving the dead horse alone and letting the editorial ebb and flow take its course. Otherwise you're just wasting your own time in neverending trite rhetoric. -- 74.160.99.252 (talk) 20:27, 11 February 2008 (UTC)[reply]

Unfortunately the IP user from Germany (not you 74.160.99.252) but the other user from 217.87.122.179 above (and their other IPs in that ISP range) decided that instead of trying dialog they would vandalise ANI, several talk pages including mine and an admin. Eventually leading to several semi-protects and range blocks. The ambiguity argument has been refuted since it relies on the false premise "nobody uses -bi in other ways except binary" because it has been shown that -bi has been used in the decimal sense. Fnag aton 20:39, 11 February 2008 (UTC)[reply]

I'm pretty sure 74.160.99.252 is the same person you're talking about. --85.25.12.31 (talk) 21:28, 11 February 2008 (UTC)[reply]

I'd like to add a note concerning the argument about base 2 centric prefixes somehow being out of agreement with SI base 10 prefixes. It is my understanding that the prefixes are not strictly measures of absolute value, but rather shorthand for differeing degrees of magnitude. As such, it makes perfect sense that in a base 10 system, the exact values of the prefixes are different from a base 2 system. Suggesting otherwise is akin to saying that a statement "...the majority of GDP..." should always have the same meaning in direct dollar amount, regardless of whether you are talking about Alaska or New York. Additionally, outside of marketing materials, I have never seen the base 10 interpretation of prefixes used in referring to computer hardware or internals.

That said, I do see a need for disambiguation for laymen. However, some non-standard (in common useage) alternate form of the abbreviations, such as MiB, does not seem a good idea, as it does not present the best interpretation for translation to real world experience. An encyclopedia is supposed to inform within context of reality, not impose some non-standard environment that makes integration of the information in to the life experience of the reader more difficult. In addition, where the prefixes are used ambiguously (eg GB = 1000X1024KB etc), specificity is of less importance. In such instances (the main one I could see being lists of products with marketed disk sizes etc), if absolutely necessary, a disambiguation such as "GigaDisk Pro 750GB (actual size XXXGB)" would seem the best solution.--207.14.29.3 (talk) 21:56, 8 April 2008 (UTC)[reply]

I think including powers of 10 or of 2 is going to add to confusion rather than reduce it. The clearest scheme I can see looks like this:-

1KB = 1,024 bytes
1MB = 1,024×1,024 bytes or 1MB = 1,000×1,024 bytes or 1MB = 1,000,000 bytes
1GB = 1,024×1,024×1,024 bytes or 1GB = 1,000,000,000 bytes etc.

This can be put into a footnote (or follow in parentheses) after the first usage of the unit to mean this amount. Sheffield Steel^talk_stalk 22:56, 8 April 2008 (UTC)[reply]

Exactly. A solution that is immediately more understandable to the average reader than using the IEC prefixes. Fnag aton 16:36, 9 April 2008 (UTC)[reply]

Binary prefixes redux

This is a reboot. MOS* editors are collectively very tired of the overblown invective in this argument "series"; consider it "canceled". Think of this as getting to do ONE wrap-up episode (or even a Serenity-style feature film, as it were). But please keep your positions grounded in WP policy, MOS, logic and civility bounds. We all realize this is an important issue for many participants, but endless fighting serves no purpose.

Informal mediation

Given that I effectively have no position on this issue whatsoever, other than (and do I not take this trivially) service to our readers, I volunteer to mediate this dispute, in all honesty and fairness, with the caveat that I have a life and cannot respond in realtime. Given that this dispute has gone on for years, and that that I have successfully mediated WP disputes in the past, I don't see that as a real issue. Step up. — SMcCandlish [talk] [cont] ‹(-¿-)› 12:27, 6 April 2008 (UTC)[reply]

And yes, I am being way WP:BOLD in archiving the "debate" (read: flamewar), yet quite short of WP:IAR.

I can't remember and don't care who said what, but the notion that dispute resolution has to go through some kind of official channels is incorrect. There are all sorts of mediation venues leading up to the official ones. Consider me the first line in this forum. The Mediation Cabal is the next step, and they aren't "official" either. If I fail, call in Kim Bruning, who is probably even more suited to the task, given that he mediated trans-continentally with me and [X] via Skype for several hours over WP:ATT issues back in the day, at a point where it was really needed.

Anyway, youse guys need to STOP. Take a break of a day or two away from this, agree on a time to resume and be willing to agree on willingness to develop a compromise position.

PS: What drove me to this to bold intervention is that your (plural) arguments were getting so long that you were crashing my browser, I kid you not.

You are already an involved editor and therefore not suitable to mediate. Fnag aton 12:31, 6 April 2008 (UTC)[reply]

Fnagaton, I honestly do not care at all how this debate plays out. I care about things like whether terminal punctuation goes inside or outside of quotation marks. And I care about end-user (i.e. non-editor encyclopedia reader) experience. The rest is "noise" to me. I have both criticized and agreed with you in the past; I think neither situation is relevant to my ability to mediate here. As far as I can recall, my stance on the position at hand is that the current wording a) reflects a reasonably neutral wording and is WP:POLICY consistent; and b) can't be deleted without a strong show of consensus. If you think this disqualifies me from mediation, feel free to say so. — SMcCandlish [talk] [cont] ‹(-¿-)› 12:59, 6 April 2008 (UTC)[reply]

In the past I have witnessed SMcCandlish showing scrupulous fairness to other editors, whether or not he agreed with their views. I welcome his offer to help. Thunderbird2 (talk) 12:41, 6 April 2008 (UTC)[reply]

But that doesn't change the fact that he is an involved editor and is therefore not suitable to mediate on this issue. Fnag aton 12:45, 6 April 2008 (UTC)[reply]

I am involved, yes, but only (so far as recall) in advancing the idea[l]s of reader usability and WP policy adherence. I have no position to advance other than consensus really has to change, in a specific direction, in order to change. I offer (again not in realtime, but checking in often) to informally mediate in this particular dispute, because I honestly couldn't care less about what unit symbols are preferred as long as the readers are served adequately, and I have some WP mediation experience. I don't think anyone else here can say the same. I'm a first-line chance at resolution. Next is the MC, after that formal mediation. Let's not "go there". It's tiresome (been there...). — SMcCandlish [talk] [cont] ‹(-¿-)› 12:59, 6 April 2008 (UTC)[reply]

Give me a try. When I go into mediator mode that's all I do; my own opinions are shunted aside. I've done this pretty effectively. Ask WP:SNOOKER if there are any doubts. — SMcCandlish [talk] [cont] ‹(-¿-)› 13:22, 6 April 2008 (UTC)[reply]

I would certainly recuse myself and refer this to a higher-up form of mediation if my neutrality were to be questioned during mediation.

Disclaimers: I feel neither positive nor negative toward units like Mib. They make sense to me, but their lack of public adoption within my [sub]culture gives me pause. I feel the same way about kilograms. I.e., I do not oppose the metric system in any way, but recognize that it does not make sense to some people and needs conversion. If this is seen as a radical position, then I may not be the appropriate moderator. — SMcCandlish [talk] [cont] ‹(-¿-)› 13:38, 6 April 2008 (UTC)[reply]

I am slightly disquieted by this form of mediation, but I am prepared to go along with it based on the following factors; if I'm wrong about them, then I guess I'm not prepared to do so, probably.

There's no commitment to "abide by" the conclusion of the moderator, as that isn't what moderation is about
If there's a view (of more than one or two editors) that SMcCandlish has become biased, recusal will follow.
There won't be an unhealthy fixation with any previous behaviour or proposals or polls, only an honest and civil attempt to move forward.

I think those will be necessary for any fruitful discussion. SamBC(talk) 14:24, 6 April 2008 (UTC)[reply]

Precisely. The only point of mediation like this is to get people to listen to each other. As far as I know, the only binding mediation of any kind on WP is the ArbCom, and that's like going to court. Very much like going to court. Ick. I have fortunately never been involved in any way on any side of an ArbCom fight, and hope never to be. — SMcCandlish [talk] [cont] ‹(-¿-)› 00:55, 7 April 2008 (UTC)[reply]

Just my two cents. SMcCandlish: I think what you are doing here (archiving, organizing, fresh start) is a good thing. I can also see that you have (wisely or fortunately) stayed out of the issue on binary prefixes. I would say that if you can not get Fnagaton to agree to mediation by you, that pretty much settles the issue that it will have to be someone else. I am heading into an FDA animal study in a bit over a week and will have to have very limited involvement from hereon. Fnagaton: I’ve read SMcCandlish’s writings above as to his values and method of operation and it is unclear to me why you think he is unsuitable to help bring peace and organization to all of this. I don’t think he is proposing binding arbitration—just mediation (what he is already sort of doing here). I see that he professes to have a deep interest in the reader’s experience on Wikipedia. What’s wrong with his help? Greg L (my talk) 17:53, 6 April 2008 (UTC)[reply]

P.S.: I also believe that the “{{disputed}}” tag preceding the current policy on binary prefixes is insuficient. Current policy was improperly rammed through without a proper consensus (see Archive 22). And there clearly isn’t a current consensus that it should be retained—far from it. There exists no rightful argument that the current policy is now sort of “grandfathered in” and enjoys the protection of a properly framed consensus to overturn it—not when it was rammed through the way it was. I think the proper thing to do is make MOS silent on the issue of binary prefixes, other than a simple statement that a new policy is being worked on. Now that’s a fresh start. Greg L (my talk) 18:09, 6 April 2008 (UTC)[reply]
- Greg, following on from what you say I'll be wiling to go along with SMcCandlish doing mediation and see how it goes. Fnag aton 18:17, 6 April 2008 (UTC)[reply]

I think SMcCandlish has the right idea here, and I hope with his help we can find some solutions. Seraphimblade ^{Talk to me} 19:10, 6 April 2008 (UTC)[reply]

Very well. Lead on, SMcCandlish. Fnagaton: As I stated above, I will be relatively inactive here for a while. When you wrote “…and see how it goes”, I ask that your litmus test not be whether or not you are getting everything you want; only that progress is being made. Your end objective is the same as mine. The current policy produced only endless discord among the editors, even well-read readers now encounter terminology on Wikipedia they’ve never seen before, and—worst of all—terminology like “256 MB” and “2 GB” have different meanings in different Wikipedia articles. We differ possibly only in the manner in which we might pursue our end goal. SMcCandlish can not possibly make any progress without buy-in from you. I ask that you afford him the greatest degree of “assumption of good faith,” and that you make his job as easy as possible. Greg L (my talk) 20:06, 6 April 2008 (UTC)[reply]

“Terminology like “256 MB” and “2 GB” have different meanings in different Wikipedia articles” – whatever the outcome of this discussion, this will remain, unless we adopted decimal-only meaning for SI prefixes and something else for binary. Heck, if Fnagaton gets his way they’ll even mean different things in one article. You really can’t have all at once, familiarity, readability and (inline) disambiguity.

Btw., I have a (currently busy) real life, too, so I’ll stay out of this from now on. I’ve said all there is to say at least once. I hope the best, fear the worst. — Christoph Päper 21:08, 6 April 2008 (UTC)[reply]

Issue summary discussion

A summary for this issue is:

There are two definitions by different standards organisations, these definitions are for the prefixes used in computing where sizes can either use decimal or binary powers of two numbers. About ten years ago some standards bodies (for example IEC/IEEE) proposed that new prefixes are used, these would be KiB/MiB/GiB etc and kibibyte/mebibyte/gibibyte etc. These new "IEC prefixes" have very limited use in the real world sources we use for articles, this is a fact and is not disputed as can be seen in the straw polls you archived.

The common use of the terms KB/MB/GB etc and kilobyte/megabyte/gigabyte etc can represent different values depending on their context. This use is defined by the JEDEC standards body and the JEDEC specifically deal with standards relating to computer memory. These older common use prefixes are still widely used today by most computing literature, manufacturers and even the IEEE use these terms in their publications. The IEEE also have a standard IEEE 100 and in this standard these common use prefixes are defined with both values. These are also facts that are not disputed. Fnag aton 14:41, 6 April 2008 (UTC)[reply]

The problem starts when some editors start to insert references to IEC prefixes into articles where the sources relevant to an article do not use IEC prefixes. Common reasons given for such changes are:

1) Just because I prefer IEC prefixes.

2) IEC prefixes are defined by the standards bodies therefore they should be used.

3) Readers could be confused about the number of bytes so we must disambiguate.

Refuting those arguments:

1) Individual preferred style is not a valid reason to promote one particular style over another. The overriding concern for editors writing Wikipedia articles is to be unbiased and to reflect the use of prefixes found in sources relevant to the article. Putting it another way, consistency with relevant reliable sources is more important than individual style preference.

2) Wikipedia wisely follows the example of the real world and ignoring the standards organisations when needed, see the BIPM issue.[2]. For example Encyclopedias like Britannica do not use IEC prefixes in their articles related to computers. (Go to http://www.britannica.com/ and search for kibibyte, mebibyte or gibibyte, the search finds no such terms in the encyclopedia.)

3) a) Disambiguation is a valid method but disambiguation should also be free from personal preference and free from promoting one particular style over another.

3) b) Disambiguation should also use terms that help the reader to understand and because the IEC prefixes are virtually unknown this does not significantly help the reader. Also the problem arises that decimal numbers cannot be accurately expressed using binary powers of two is used. For example a hard drive of 100GB, which might actually be 100,000,000,000 bytes, expressed as some multiple of a power of two is 93.1322574615478515625×2³⁰ bytes. Of course for brevity this needs to be shortened to be 93.1×2³⁰ which then results in 99965363814.4 bytes this leaves an accuracy difference of 34636185.6 bytes. The reader is left with the impression that either this difference doesn't matter or that fractional bytes can exist in computing hardware and both are not true in computing articles.

3) c) Some editors will try to claim that actually the exact number of bytes doesn't matter and what matters instead is that the difference between decimal and binary prefixes is highlighted, this is fine but it is not true that the best way to accomplish this is to use the virtually unused IEC prefixes. That said, if the IEC prefixes were widely used and understood then of course it would serve the interests of the reader to disambiguate with them, however they are not widely used or understood so it does not serve the interests of the reader to use them when better methods (3d below) already exist. The claim is then sometimes made that IEC prefixes can be wikilinked, but again by introducing virtually unused terminology does not help to explain to the reader why this change was made and gives the false impression that MB can be equated with MiB. (Just look at the disagree comments for [3] and [4].

3) d) What would help the reader, i.e. be better for the reader, is a short footnote explaining that there exist decimal and binary systems and giving links to articles that explain the issue much more fully without needing to promote IEC prefixes in the article itself, for example linking to Binary prefix. The fair and balanced solution is therefore to use some other form of precisely disambiguating decimal or powers of two numbers that doesn't push/promote any prefix style. Being fair, balanced and unbiased is something that should be true for Wikipedia articles. So this means adopting a system like "256 KB (256×2¹⁰bytes)".

3) e) Some people say that removing IEC prefixes from the guideline unfair or unbalanced deprecation of IEC prefixes, this is fallacious because the definition of deprecation is "to express earnest disapproval of". Removing advocation of IEC prefixes does not specifically express earnest disapproval of only IEC prefixes. The argument is therefore fallacious because in the current text specifically mentioning IEC prefixes is to advocate IEC prefixes and that is pushing a biased POV. Putting it another way, specifically mentioning IEC prefixes is actually unfair and unbalanced advocating of IEC prefixes. Pushing a biased POV towards either of the prefix style should be unacceptable in the guideline, especially considering that we should have the reader's best interests at heart. So, removing a push for a biased POV from the guideline is not deprecation, it is actually removing bias.

Given that supporting one style of prefixes over another is not welcomed, especially when that style is obviously in the minority. Given that some people also don't like it when a particular prefix style is deprecated, note this means specifically saying "do not use these terms" and does not mean removing advocation of such terms. Given that MOSNUM also has the wise guideline about using prefixes/units employed in the current scientific literature. The only logical and fair conclusion is to remove reference to any hint of support or deprecation for any of the standards organisation styles at all and make it clear that the units must defer to the sources for the article which will then logically defer to what is the real world consensus. In doing so all the text about "no consensus for using IEC prefixes" is removed as is the text explaining the history of the prefixes and as is the text that says "Use of IEC prefixes is also acceptable for disambiguation". This is because the respective articles (Binary prefixes etc) already contain the history of these prefixes and already explain why these differences between decimal and powers of two systems happen. Also gone would be the text about not changing from one style to the other and keeping with the original style, this is important because it allows articles to naturally change over time to reflect what is the consensus in the real world if the sources for that article considerably change from using KB to KiB for example. I don't see that happening any time soon but this is a compromise to those that feel it just might. This would also try to remove what is seen as a Wikipedia guideline pushing any one standard over another and to remove any bias coming from Wikipedia. The guideline would then becomes a series of examples of what to do for disambiguation, wikilinking and footnotes.

Now we get onto the topic of the "third hybrid proposal" text.[5]. As can be seen from the support comments these mostly follow a similar vein of using technical reasoned argument against using IEC prefixes based on the Wikipedia principle that consistency with reliable sources is important. On the oppose comments we have, 1) "votes are evil", 2) "the proposal is messy" along with personal style, 3) we need to discuss, 4) a claim that it deprecates which is not true and a claim that it doesn't address "ambiguity" when actually the proposal specifically addresses this, 5) another fallacious deprecation argument, 6) another fallacious "out right ban" and misrepresentation about the motives of myself and Greg, 7) fallacious deprecation, 8) the "standard body defines this" argument which is refuted above, 9) point of view, 10) fallacious argument claiming there has been no valid argument and a comment about "ILIKEIT votes".

Looking at the oppose votes it is clear to me that most of them use falacious reasons that are already refuted and can be disregarded when considering this issue on aspects that are relevant to putting the interests of the reader first within the scope of existing Wikipedia guidelines. Fnag aton 14:41, 6 April 2008 (UTC)[reply]

I believe Fnagaton misstates the facts as follows:

"JEDEC is the leading developer of standards for the solid-state industry" and as such has no dealings with computer memory - semiconductor chips and modules, yes, but computer memory no!
Historically mega (M) and giga (G) were well established as 1,000,000 and 1,000,000,000 long before they acquired a binary meaning. Furthermore, MB (Mbyte or megabyte) and GB (Gbyte or gigabyte) were also well established as 1,000,000 bytes and 1,000,000,000 bytes long before they acquired a binary meaning.

It is particularly disturbing that some anonymous editor reverted this comment with the justification, "misrepresentation." The above statements are factual. The solid state industry is a supplier to the computer industry. The MB history is well documented in the time line. You may disagree with my interpretation of these facts, you may even call it POV, but it is improper to supress this, especially on a talk page. Tom94022 (talk) 23:08, 6 April 2008 (UTC)[reply]

Fnagaton is completely correct. The JEDEC does set standards for computer memory and it is the industry leader. Your statement is incorrect. DavidPaulHamilton (talk) 23:51, 6 April 2008 (UTC)[reply]

JEDEC memory standards proves they set computer memory standards. I propose Tom's incorrect statement and these following comments are moved out of this section to keep the thread clear of rubbish.DavidPaulHamilton (talk) 00:30, 7 April 2008 (UTC)[reply]

Except that flash RAM devices, which certainly come under the heading of "computer memory" and "semiconductor devices," are quoted in decimal MB or GB (as are hard drives). JEDEC member companies' practice therefore seem to be internally inconsistent. IMO this means we should adopt a more useful and internally-consistent standard. Jeh (talk) 03:02, 7 April 2008 (UTC)[reply]

That is inaccurate. The chips used for flash memory are defined in their technical data sheets to use binary MB. We aim higher by not using unknown confusing IEC prefixes.DavidPaulHamilton (talk) 06:44, 7 April 2008 (UTC)[reply]

I guess I wasn't clear enough. I said "flash RAM devices", meaning the likes of CompactFlash, SecureDigital, etc., devices. How the internal chips are spec'd means notbing to the buyer. What is relevant here is that when one buys a "4 GB" flash device, it will be a bit more than 4×10⁹... but most OSs then display that capacity by using a power of two divisor and an SI-style prefix, so one sees a number like "3.72 GB". If that doesn't demonstrate that use of a single prefix with two different meanings is confusing, or why we must disambiguate these numbers in some way, I don't know what will. Jeh (talk) 07:26, 7 April 2008 (UTC)[reply]

You know as well as I do that when memory chips which are defined as binary powers of two by the JEDEC standards are repackaged into other offline storage devices they are are marketed as drive storage and therefore not "computer memory" as such. You might not know this but also when the powers of two memory chips are used in something like a USB stick some part of the memory can be used to store extra information not related to the storage of files, for example recovery software, ISO images of CDs so the device can initially appear as a CD device to install drivers, encryption key storage, etc. This still means my statement is correct because the computer memory part still follows the JEDEC standard. Tom94022 is also wrong because semiconductor chips are the same as computer memory (RAM) in the context of using the phrase with the JEDEC, as such the statement "JEDEC specifically deal with standards relating to computer memory" is completely accurate. Cites from manufacturers and technical sites that also show why you are wrong: "All Kingston memory is JEDEC compliant, an important specification used by leading semiconductor manufacturers. JEDEC sets the standards for semiconductor engineering and is well respected throughout the industry. "[6] "JEDEC is a standards body for the global semiconductor industry. ... JEDEC sets the basic specifications to which all RAM [computer memory] must adhere at a minimum.". "JEDEC is famous for its computer memory (RAM) standards, for example, DDR SDRAM and DDR2 SDRAM. "[7] If you need more links Google: "computer memory JEDEC". Or you could just look at the article JEDEC memory standards and JEDEC which says "JEDEC memory standards for computer memory (RAM)". It cannot get any clearer than that, really. Fnag aton 08:47, 7 April 2008 (UTC)[reply]

Fnagaton, nobody is disputing the claim that "JEDEC deals with standards relating to computer memory", so quoting all kinds of things and giving links to "prove" this is off point. Jeh (talk) 07:00, 11 April 2008 (UTC)[reply]

Also, don't insert your comments half way inbetween my post, it makes things harder to read. Fnag aton 08:47, 7 April 2008 (UTC)[reply]

Just as you earlier asserted that you would edit as often as you like, despite complaints of edit conflicts... I will post as I please. Jeh (talk) 07:00, 11 April 2008 (UTC)[reply]

A hard disk drive industry standards group, IDEMA, has defined Gbyte as 1,000,194,048 bytes in order to avoid confusion in the reporting of capacity of ATA and SATA drives^[1]. With this definition all HDDs with the same labeled capacity (to several digits after the decimal point) will report the same number of 512 byte blocks. IDEMA is close to but not a decimal prefix definition. IDEMA is no more or no less an authority to the computer industry than JEDEC. I do not dispute that JEDEC sets standards for memory modules but I do dispute that in resolving this Wikepedia issue they have any more significance than IDEMA, or IEC or the HDD manufacturers or the DVD manufacturers. To that extent Fnagaton's summary, emphasizing JEDEC is his POV and needs clarification. Tom94022 (talk) 19:09, 8 April 2008 (UTC)[reply]

Mentioning IDEMA is a red herring since that is not relevant to and does not refute my statement that "the JEDEC specifically deal with standards relating to computer memory". The fact that the IEC prefixes are virtually unused by the real world shows that the IEC has to be given less weight when considering what scheme to use. To do anything else is illogical and is pushing a POV. Fnag aton 19:28, 8 April 2008 (UTC)[reply]

Tom94022: It’s not complex. The manufacturers of USB flash drives are no different from anyone else in the transistorized computer memory market and follow the convention used by the rest of the computer industry. A “2 GB” flash drive has 2 × 2³⁰ bytes of memory inside. Because of that fact, they advertise using binary values to denote device capacity. However, the memory inside the USB drive is not made available to the computer like ordinary RAM; the drives are treated like hard drives and the internal memory must be formatted. To ensure the drives can be recognized by both Macs and Windows machines, manufacturers usually format using FAT32. To quote Edge Tech Corporation: “ The usable size of the DiskGO™ should be within about 10-12% of the device's stated size.” The actual thumb drive memory available to your computer after formatting will be displayed to you per the convention used by your operating system. Your computer likely uses the same convention my Mac does, which right now displays this regarding my 512 MB USB flash drive: “Capacity: 483.5 MB (506,970,112 Bytes)”. Greg L (my talk) 21:03, 8 April 2008 (UTC)[reply]

Greg L: I'm afraid you're mistaken, and Edge Tech is just using weasel words to avoid really explaining things... which, as we've seen, causes confusion more often than not. Reality is that a flash drive sold as "2 GB" means that it has a bit more than 2,000,000,000 bytes available to the user, just like a "300 GB" hard drive has a bit more than 300,... well, you get the idea. The difference between what your OS reports and (in this case) 512 MiB is not due to formatting. No way does formatting on a 512 MB drive use up 30 MB, which is the difference between 512×2²⁰ and your reported 506×10⁶. And on the "8 GB" flash drive I have here, Windows reports it as "8,003,842,048 bytes 7.45 GB". 8 GiB would be 8.6 GB; you don't really think FAT32 format uses up 600 MB, do you? Try it on a hard drive and see. Incidently, Lexar, among others, honestly prints "1 GB means 1 billion bytes, 1 MB means 1 million bytes" on their flash drive packaging. Ladies and gentleman, this is a perfect example of why the poor SI prefixes need to be rescued and reclaimed for decimal purposes only: Using them for both decimal and binary is confusing. Look, it even confused GregL, and he understands what is going on! Jeh (talk) 07:00, 11 April 2008 (UTC)[reply]

Fine, Jeh. Let’s assume for the moment that what you say about flash drives is true. Your point speaks to the issue of the actual capacity of thumb drives, not the units of measure we use to describe capacity. That’s what we’re all here for to settle. Encyclopedia Britannica and all computer magazines would simply report the actual, formatted capacity of my “512 MB” flash drive the very same way yours and my and everyone elses’ computers report available capacity to us: “Capacity: 483.5 MB (506,970,112 Bytes)”. Why can’t Wikipedia go with the flow on this? It’s been nine years since the IEC proposal and it still hasn’t caught on with the computer industry. Is what we’re trying to fix really all that broken if everyone else is managing to deal with the units? Why don’t we simply agree to disambiguate on Wikipedia just like our computers are currently doing it for us? Greg L (my talk) 15:04, 11 April 2008 (UTC)[reply]

Yes, my point speaks to the units, or rather to the prefix. Look at the package that 506,970,112 byte memory stick came in. It says "512 MB". So they are using the decimal prefix, not binary, even though it's clearly a semiconductor device. The reason disambiguation to exact numbers of bytes is not sufficient is simple: It begs a question and will leave many readers confused: "Wait, why is 483.5 MB equal to 506 million? I thought mega meant million?" Yes, I know, "MB = 2²⁰ bytes" is common in the computer industry, but it is not always so within that industry and "mega-" never means anything but "million" anywhere else. In no other field of measurement is the unit of measure (bytes vs., say, meters or Hertz) allowed to change the meaning of the prefix ("mega-")! Why "consistency within and following the precedent of the computer industry" is lauded as a banner to march under, while the fact that the computer industry's use of SI prefixes for RAM is inconsistent with every other use in the entire world since SI was conceived is pooh-poohed as unimportant even though it demonstrably creates confusion, is something your side has yet to explain satisfactorily. Jeh (talk) 18:29, 11 April 2008 (UTC)[reply]

Your memory stick examples fail to take into account that not all the semiconductor memory is used for storing files. Using numbers will confuse readers less than trying to introduce yet more prefixes which are virtually unused in the real world. As for "something your side has yet to explain satisfactorily" you are wrong to claim that because it has already been explained many times already. Also you are wrong to try to claim that is the issue when actually it is not. The actual issue is for you to explain why you want to depart from the real world consensus. i.e. Forcing virtually unused prefixes onto people that are not familiar to average readers is something that you have not adequately explained. Fnag aton 19:17, 11 April 2008 (UTC)[reply]

1. Nor is all of the capacity of a hard drive used for storing files. There are spare tracks or sectors, there is space "lost" to the low-level format, etc., etc. 2. I disagree that expressing the exact number of bytes is the best way to disambiguate. Many people "fuzz out" upon seeing a large number; that is why most operating systems do NOT give you the actual number in bytes unless you "ask really hard" (e.g. file or drive Properties in Windows). I personally would prefer that Explorer list every last file size to the exact byte, but I'm a distinct minority. It is also not immediately obvious, for example, whether a number like 536,870,912 is exactly a power of 2 or just close to one, yet this is a useful distinction in many contexts (though not usually for file size). An expression such as "512 MB (2²⁹)" or "512 MB (512×2²⁰) conveys that distinction, but is in my opinion likewise even more unfamiliar and/or offputting to many readers than a Really Big Number. 3. I claimed it has not been explained satisfactorily, and that remains my opinion, therefore I am not wrong: "Satisfactory" is a value judgement and mine is as valid for me as yours is for you. 4. "Real world consensus" is that "mega-" is 1,000,000; it's the computer industry that's leading us astray. 5. Yes, the prefixes are initially unfamiliar, but that can be handled easily with a Wikilink and/or a footnote. And once explained, I think—again, my opinion, so you are of course free to disagree but please don't presume to tell me "you're wrong"—they're a heck of a lot easier to remember than "536,870,912". Jeh (talk) 19:40, 11 April 2008 (UTC)[reply]

1) I know because I wrote that before when I explained the first time why you are wrong. 2) You may disagree but the fact is that numbers are completely unambiguous. Also, using numbers would confuse the average reader less than trying to introduce them to IEC prefixes. Using IEC prefixes does not explain the differences in harddrive capacity. 3) Then don't try to pretend your opinion passes as fact. Also don't use unhelpful language. You should have realised by now that I will use your own quotes and language back at you to refute what you write. 4) Wrong. Real world consensus clearly shows that both decimal and binary versions of the two letter prefixes KB/MB/GB are very common. 5) The numbers can easily be wikilinked or added to footnotes that explain the situation to the average reader and that is more useful than trying to introduce the virtually unused IEC prefixes to satisfy the personal opinion of a some editors. Lastly, you are wrong as I have demonstrated when I again used your own quotes and language against you. If you don't want that to happen then you're going to have to avoid posting things that are obviously wrong. Fnag aton 20:00, 11 April 2008 (UTC)[reply]

1) The fact is that flash memory devices are branded using decimal prefixes, like hard drives. But these devices are often sold as "flash memory" (hence "computer memory"). And actual computer internal memory devices like DIMMs are of course branded using binary divisors, but decimal prefixes. I understand fully the actual distinction in product type. My point is simply that the usage adds to the confusion. The consumer thinks he has a handle on the concept that "in computer memory, MB means a weird number somewhat larger than 1,000,000" but all of a sudden here is a "memory" device that breaks the rule. The knowledgeable consumer is probably aware that inside that device, at some level, there IS a memory chip with a capacity (for some definition of "capacity") that can be expressed as an exact power of two. Nevertheless the package quotes the device using a decimal prefix and decimal divisor. Confusion. Please don't tell me that said confusion doesn't happen; even GregL here was misled. Jeh (talk) 21:21, 11 April 2008 (UTC)[reply]

2) Numbers are completely unambiguous, yes, but that doesn't mean they're always the best way to explain something. Why do you think scientific notation was invented? Or any prefixes at all, for that matter? I think IEC prefixes are trivial to learn and then, in some cases, are more useful and more quickly expressive of the desired meaning than any other method. No, IEC prefixes by themselves don't explain the difference in hard drive capacity, but why would you use one there? Nobody is saying "use IEC prefixes everywhere"! But consistent use of BOTH SI and IEC prefixes, each as appropriate, will help. Jeh (talk) 21:21, 11 April 2008 (UTC)[reply]

3) Characterizing my post as "unhelpful" does not make it so. I'm not trying to pretend that my opinions are fact, I'm just pointing out that rebuttal of an opinion that's expressing a value judgment (as opposed to a rebuttal of a claim of fact) as "wrong" is inherently fallacious, no matter how much "factual evidence" is provided. You see, most of the points of contention here are not over matters of fact, they are value judgements. Example: It is my opinion that the initial unfamiliarity of the IEC prefixes is outweighed by their usefulness. Obviously you disagree; you think the unfamiliarity is one of several overriding concerns. That's a value judgement for each of us. You value familiarity more; I value more the concision and preciseness that's achieved after the unfamiliarity has been overcome. Fine. But endless quotes and links "proving" that e.g. the IEC prefixes are unfamiliar, or that JEDEC sets standards that are used by RAM makers, etc., etc., are not going to sway either of our value judgements because the points proven by those methods are not being questioned in the first place and are not the points of contention here. And any further claims based on such a "proof" are fallacious as well, no matter how many times repeated. (And furthermore looks simply like "not addressing the real issue".) Now if there was a valid survey that found that, say, IEC prefixes were difficult for most people to learn and understand, or that most people surveyed found them unhelpful and confusing even after learning them, then you can of course point to that and say something like "your value judgment is not supported by this study..." But so far no such studies have been done AFAIK, and all the Google hit counts in the world aren't a substitute for that information. Jeh (talk) 21:21, 11 April 2008 (UTC)[reply]

4) No, "Right". The computer industry notwithstanding, there are far more uses and much longer history of mega = 1,000,000 in the real world than mega=2²⁰. Even within the computer industry there are a significant number of uses of mega=1,000,000. And there are NO other fields in which the unit of measurement is permitted to change the meaning of the prefix! Therefore while real world usage within the computer industry, and only for certain types of products, does support "mega=2²⁰", this is only a portion of the computer industry (e.g. please don't try to tell me that JEDEC sets standards for hard drive makers) and the computer industry is only a portion of the world. It is not valid to draw a line around a small part of the world and say "see, this subset of the world supports my view, therefore I'm right." Now I know someone is going to say "but we're just talking about the computer industry." Fine, but that industry is demonstrably inconsistent with itself. So why is it a good example to follow? Jeh (talk) 21:21, 11 April 2008 (UTC)[reply]

5) Yes, but the IEC prefixes can be easily wikilinked also. And once explained they are more easily read, far less visually obtrusive, and more informative than either scientific notation or an exact number (my opinion). Jeh (talk) 21:21, 11 April 2008 (UTC)[reply]

1) Not all of them. The prefixes are not decimal for memory chips, the are defined by the JEDEC to be binary. Using IEC prefixes adds to the confusion because they are virtually unused and unknown by the average reader. It is better to explain the real world situation to the reader so that they are able to read real world sources without trying to force them to understand virtually unused IEC prefixes. Using IE prefixes in the article only adds to the confusion. Any point you can make about adding wikilinks to make the IEC prefixes understood can also be applied to disambiguation of the sources used in the article and doing that is better, for the reasons already given. 2) Advocating IEC prefixes is pushing a point of view and it is not the editor's job to use prefixes that are unused in article sources when there exist other better ways to disambiguate. 3) Yes your language is unhelpful. I used your own language back at you to refute your own fallacious statement, it's irnoic you say what I wrote is fallacious when I was actually copying what you wrote. You just refuted yourself. You are also misrepresenting my position, do not do that in the future. Since you misrepresent my actual position the rest of your statement is fallacious. My "value judgement" as you put it is not the issue here because my position is that we have to follow wikipedia policy, as I already explained before. My personal feelings do not affect the logical argument that I make. You on the other hand do admit your position comes from your "value judgement". You are also forgetting one thing, value judgements are biased. We have WP:NPOV as a policy. This means your "value judgement" has no place in articles. 4. You are still wrong and restating your personal point of view, which is contrary to the real world consensus. I'll say it again, real world consensus clearly shows that both decimal and binary versions of the two letter prefixes KB/MB/GB are very common. You are incorrectly trying to conflate K and KB. The two are not exactly the same, the letter K may be the same letter but when it is used with the letter B or "bytes" then the meaning subtly changes. You need to read IEEE 100. The use of these prefixes is enough for that standard to specifically include them as both formats. This neatly proves you are wrong. 5) They could be wikilinked but that adds virtually unused prefixes and confusion to articles which is not in the best interests of the average reader. You still have not given an adequate logical argument for including prefixes contrart to real world consensus. Fnag aton 21:56, 11 April 2008 (UTC)[reply]

Again you start out by "arguing" a point not in contention. The NPOV rules cover main article namespace. By contrast, a great many WP policies must necessarily include "points of view." Many of the existing topics covered by WP:MOSNUM are already value judgements: each of several possible decisions may have some advantage, but at some point it has to be agreed upon that one advantage is more compelling than another. That's a value judgement. You're concluding that consistency with the inconsistent use in the real world is more important than adopting a concise, unambiguous, and easily explained notation; that's a value judgement too... in my opinion. "I'll say it again," real world consensus has led us to a very confusing situation and therefore something better should be chosen. Jeh (talk) 22:47, 11 April 2008 (UTC)[reply]

No you are wrong, it is the issue because what you are proposing affects articles. Your point of view is contrary to Wikipedia policy. Your proposal is not precise because I have already shown using IEC prefixes results in ambiguous quantities of bytes being represented. You cannot allow your value judgements or point of view to affect articles or policies. The "something better" is not to force using IEC prefixes because that is against Wikipedia policies and is therefore not in the interests of the reader. Answer this, how is it being consistent with reliable sources and following WP:NPOV when you introduce prefixes to articles that are not used by those sources? Fnag aton 23:07, 11 April 2008 (UTC)[reply]

(unindented)

Really Jeh, you need to come up with a better argument than WP:IAR. The entire content of Jimbo’s statement is “If a rule prevents you from improving or maintaining Wikipedia, ignore it.” Hell, before even knowing Jimbo had that concept officially memorialized as an article, I’ve long believed in that concept and practiced it because it makes too much sense. That’s my whole world-view: screw rules if they make you do stupid things. Check out Kilogram. Note what’s at the bottom of it. Seen a Glossary before on Wikipedia? Well I haven’t. And I’ve had someone complain about it—once. But all the other editors who worked on it with me never complained—not even an arch-enemy über-prick who got blocked for a week—because it was just so nice to have one in that article and it serves a valuable purpose. In fact no one even mentioned it during that major editing storm. Sometimes you need to make your own rules to improve Wikipedia. This is the sort of thing Jimbo is addressing with WP:IAR; don’t let stupid rules get in the way of making Wikipedia a better place. Now…
Clearly this is a very different animal we’re dealing with. There never was a proper consensus to begin using the IEC prefixes in the first place (just a majority vote after a couple of days of debate on Talk:MOSNUM ), there’s been nothing but continuous vitriolic discord since, and support has waned since it was first adopted. Perhaps support for it is eroding because it is becoming increasingly clear that even with Wikipedia’s “'Follow us, we’ll lead the way'!!!” prohibited soapbox advocacy, the IEC prefixes still haven’t caught on with the rest of the world. We’ve got to get past this notion that we can hide behind a rule that says there are no rules as a justification to break every damn fundamental principal Wikipedia has (as well as the basic tenets of good technical writing). The reality of it all comes down to the fact that the continued use of essentially unknown units of measure just can not be justified given that no one else in the computer industry, general-interest computer publication world, and encyclopedia world, perceives that there is even a problem that requires fixing. This “problem” with ambiguous meanings of terms like “megabyte” has been blown so completely out of proportion, it is effectively an imaginary one of our own making. I think it’s that, and the fact that there is so much prior work that has to be undone, the prospect of fixing it is just too distasteful to contemplate. There are so many other ways to make it perfectly clear how many bytes are being referred to without using the IEC prefixes. Greg L (my talk) 23:49, 11 April 2008 (UTC)[reply]

Greg L, who are you talking to? Which comment does your response refer to? --217.87.103.237 (talk) 00:27, 12 April 2008 (UTC)[reply]

Oops. Sorry. I was addressing Jeh. Now fixed. Thanks. Greg L (my talk) 01:10, 12 April 2008 (UTC)[reply]

For the record, I wrote to JEDEC for a clarification on JESD100, and they state that it is not at all a requirement, but just a list of definitions. They say that these definitions are only included to document common usage, and to document the fact that they are deprecated. "The standard meanings of the terms kilo, mega, giga, etc are defined in the SI system of measurement by international agreement to be powers of 10".

They do say that one could hypothetically self-impose an obligation to these definitions, by saying "in this document, we use the definitions as per JESD100", but 1. this would be entirely voluntary, and is not necessary for compliance with JEDEC and 2. the manufacturer could use either the traditional prefixes or the IEC prefixes in this situation, since both are defined in the document. — Omegatron (talk) 00:18, 10 April 2008 (UTC)[reply]

For the record I wrote to the JEDEC as well and your statement is incorrect. The definitions they show in their standard are not deprecated as you claim. Also you misrepresent the facts because to claim compliance with the JEDEC standards KB/MB/GB have to be used in powers of two sizes. The fact is the JEDEC standard says "No claims to be in conformance with this standard may be made unless all requirements stated in the standard are met.". That disproves what you just claimed. IEC prefixes are not defined as part of the standard either, they are mentioned as a footnote. Fnag aton 09:12, 10 April 2008 (UTC)[reply]

I will also bring up your recent edits here including this one, since you chose to to make a few chnages to binary prefix related articles and people here need to be made aware of the changes you have tried to make. This is because the text reads as "JEDEC Solid State Technology Association, the semiconductor engineering standardization body of the Electronic Industries Alliance (EIA) in Standard 100B.01[6][42] continues to include definitions in the binary sense K, M and G as prefixes to units of semiconductor memory, though these definitions are “only included to reflect common usage” and are deprecated. All standards published by JEDEC are still using the common usage, including end-user packaging recommendations for memory chips.". This bit highlighted in bold is completely wrong, the JEDEC have not deprecated K/M/G at all. Your edits are not representative of the facts related to this subject. Fnag aton 15:08, 10 April 2008 (UTC)[reply]

As you know, all JEDEC memory standards use the terms MB and Mb for megabyte and megabit. Here is a JEDEC standard published in January 2008. "PC3-6400/PC3-8500/PC3-10600/PC3-12800 DDR3 Unbuffered SO-DIMM Reference Design Specification" Look on page 20. The memory modules in your computer are based on JEDEC standards like this one.-- SWTPC6800 (talk) 01:27, 10 April 2008 (UTC)[reply]

Yes SWTPC6800 the JEDEC define and use the prefixer so it is incorrect for anyone to say they are deprecated.DavidPaulHamilton (talk) 21:11, 10 April 2008 (UTC)[reply]

There is no disagreement that MB stands for megabyte. Both, IEC and JEDEC define MB as short for megabyte. --217.87.124.227 (talk) 21:33, 10 April 2008 (UTC)[reply]

You have got to be kidding me. I actually wrote to JEDEC about this and got a response, and their position is pretty impossible to misunderstand. See Template_talk:Quantities_of_bytes#JEDEC. Please stop going around every page on Wikipedia and misrepresenting this as an "official definition". They define both the traditional convention and the IEC convention in the same document, which is not a requirement.

As you know, all JEDEC memory standards use the terms MB and Mb for megabyte and megabit.'

And that's a perfectly valid argument.

Just please stop misrepresenting this document as "a standards organization officially endorses this notation", because it's obvious that they don't. As far as I can tell, they are officially agnostic on the subject. Manufacturers are required to follow a number of standards to be JEDEC compliant, but usage of a certain unit convention is not part of those requirements. One of the JEDEC members I talked to specifically said that a manufacturer using IEC prefixes while declaring themselves in conformance with this document would be valid. — Omegatron (talk) 00:29, 11 April 2008 (UTC)[reply]

The KB/MB/GB prefixes are in a document that has the text "Terms, Definitions, and Letter Symbols for Microcomputers". Therefore it is correct to say the terms are defined by the JEDEC. To try to claim those prefixes are "not endorsed" is point of view and is contrary to the facts. Omegatron is not a reliable source. Since Omegatron's claims are not from a reliable source then this means what he writes cannot be verified. Therefore trying to use his claims in articles (or templates used in articles) is in violation of Wikipedia:No original research. Use of his claims in articles (or templates that are included in articles) is also in violation of Wikipedia:Neutral point of view. Since his claims do not meet the criteria for inclusion in Wikipedia articles (or templates that are included in articles) then they should be disregarded. A case in point is Omegatron's recent edit warring to include material that is unsourced and entirely his own point of view, the reason he gave for one of the edits "i've talked to JEDEC directly about this and the interpretation here is misleading" violates these Wikipedia policies. Just to make it absolutely clear Omegatron, Wikipedia does not publish original thought: all material in Wikipedia must be attributable to a reliable, published source.. If you want to put in articles that "the JEDEC do not endorse KB/MB/GB" then find reliable sources that say so. Also if you want to put in articles "the JEDEC have deprecated KB/MB/GB" then you must also supply reliable sources that state that. Fnag aton 08:50, 11 April 2008 (UTC)[reply]

Only consensus opinion should appear on MOSNUM

All: It is wholly inappropriate for a policy that was improperly adopted in the first place (without a proper consensus) to proudly masquerade on MOSNUM as a legitimate guideline of any sort (see Archive 22). The proper thing to do is for MOSNUM to remain silent on the issue until a consensus of any sort is arrived at. No matter how small the point of agreement is on the issue of binary prefixes, anything that appears on MOSNUM should be arrived at via a Wikipedia-style consensus. I have made it so (∆ here) and I ask all other editors, if they feel strongly about MOSNUM’s silence on the issue, to join in here and find some common ground with each other. Greg L (my talk) 23:03, 6 April 2008 (UTC)[reply]

Greg_L, I feel that it is only according to your and a few others' assertion that the current wording was not properly adopted. Your standard for "consensus" seems to be a unanimous vote, but that is not at all a requirement under WP:CON. Since you have raised the question, this is one of the points I listed (now archived) upon which I would like to see a mediator's decision -- did the current wording represent, at one time, a consensus? In the meantime, there is certainly no consensus that previous consensus was NOT achieved, therefore YOUR recent changes to WP:MOSNUM are out of bounds by your own standards -- and IMO, are bordering on disruptive behavior. Disputes on talk pages should not be brought to the subject main page, particularly pages concerning WP policy and guidelines. Jeh (talk) 00:21, 7 April 2008 (UTC)[reply]

The solution according to the xkcd standards committee Gimmetrow 00:32, 7 April 2008 (UTC)[reply]

I wholeheartedly agree with the xkcd standards committee’s views on the binary prefixes. Thanks Gimmetrow. I’ll let SMcCandlish deal with 1) whether 20:7 was ever a proper “consensus” as the term is typically understood, and 2) whether it is wise/warranted to have such a disputed and poorly supported policy in MOSNUM (13:10 to replace the current guideline lock stock & barrel with one that would prohibit their routine use and deprecate many existing uses) while we’re trying to agree on anything here; I think its existence hinders progress here. But I’ll leave it alone for the moment. Greg L (my talk) 01:39, 7 April 2008 (UTC)[reply]

Greg, please read WP:CON. The way I read it, consensus on WP is arrived at when the "edit loop" ends. It seems to me that the edit loop, and even the discussions on the talk page, ceased for many months after the current wording (or something very very close to it) was established. Therefore (again, in my opinion) it does constitute consensus -- or did at one time, as of course, consensus can change. And maybe it has. But that possibility is not justification for removing long-standing text from an existing article. Jeh (talk) 01:50, 7 April 2008 (UTC)[reply]

Regarding the consensus about the 20:7 vote: I'll quote Omegatron "There was no consensus in Archive 22..."[8]. I remember Bladestorm mentioning that he was being told there was consensus for the proposed guideline text and that being given as the reason to only use IEC prefixes, which was against the spirit of the guideline because it was only recommended and not mandatory. Just look at the changes made by Sarenne before she was banned, she used the old guideline text as justification for only using IEC prefixes and forcing them into hundreds of articles. So yes, people in the past have misrepresented the consensus issue with the guideline text that was being used and misrepresenting that to force using IEC prefixes. Fnag aton 16:16, 7 April 2008 (UTC)[reply]

You've got it half right:

There was no consensus to use IEC prefixes throughout the project
There was no guideline that said to use IEC prefixes throughout the project

There was a strong majority in favor of them, so the guideline said they were recommended in cases of dispute, but to my knowledge it has never required their use, since even a strong majority wanting to mandate their use does not equal a consensus per Wikipedia policy. — Omegatron (talk) 01:16, 10 April 2008 (UTC)[reply]

No, I got it completely right, you will note this is because I said "people in the past have misrepresented the consensus issue with the guideline text that was being used and misrepresenting that to force using IEC prefixes". Those people that claimed the policy (User:Sarenne) required IEC prefixes to be used are those who got it wrong. Fnag aton 09:07, 11 April 2008 (UTC)[reply]

Goals

Okay, so I would love it if each interested party would in a RfC or XfD style bullet-comment briefly (1-4 sentences?) express what their personal goals are with regard to this debate and its outcome. I think this would help us all understand each others' positions better. Like, is it important to you that both styles be usable, but always be given a conversion for max. reader understanding? Is it crucial that the KiB-style units be promoted and the KB ones be deprecated, for standards reasons? Or the KiB ones be avoided because their general-public buy-in is low? Or whatever. Without getting into why the other side(s) are "wrong". I think we can probably all agree that the meta-goal is of course consensus, and that further one is surely an increase in mutual editor respect, leading to peace and calm and all 'at. I have to go run a pool tournament right now, but will read the summary discussion issues above as soon as I get back. PS: I'll go first; I'm not a party, but I think you should all know where I come from on the topic. — SMcCandlish [talk] [cont] ‹(-¿-)› 01:07, 7 April 2008 (UTC)[reply]

My preferred outcome here is that, whatever the resulting consensus is, the average reader will not be confused or mislead in any way. By nature I tend to lean towards preferring standardized ways of doing things mostly (I'm a fanatic about Web standards), but I don't in real life use the metric system at all (except where it has been forced on me, e.g. soda pop being available in liter but not quart bottles). This makes me either flexible or a hypocrite. :-) Secondarily, I want MOS to be consistent and guiding, and accepted by the community. — SMcCandlish [talk] [cont] ‹(-¿-)› 01:07, 7 April 2008 (UTC)[reply]

Well, I guess I'll start here then. My main goal is to avoid confusion, and, as a formal reference work, to see the terms Wikipedia uses be unambiguous, technically correct, and clear. I believe that this can be best achieved by using the binary prefixes to represent binary values, as these values are clear and unambiguous, and to use decimal prefixes to represent only decimal values. These uses are clearly defined by standards bodies, so veriication of the terms' values is not problematic. However, the only way ambiguity can be avoided is for a clear practice to be adopted sitewide. Seraphimblade ^{Talk to me} 08:09, 7 April 2008 (UTC)[reply]

Rationale: The IEC prefixes are not clear to the average reader because they are virtually unknown. It is not the place of Wikipedia to advocate a style of prefix especially when it is unused by the vast majority of the computer industry. The vast majority of formal references works by manufacturers do not use IEC prefixes, the Encyclopedia Britannica doesn't even mention IEC prefixes, so Wikipedia should not promote their use either. Common use, real world consensus and the sources we use for articles defines what is correct for Wikipedia to use, not a standards organisation that proposed a standard which is very rarely used. Goals: My goals are those of the Wikipedia policies and guidelines. Consistency with relevant reliable sources used in an article. Unbiased articles that represent the real world consensus. Making the contents of articles clear to readers where needed. This can be accomplished without needing to push virtually unknown and unused IEC prefixes onto our readers. A good test is to look at what Wikipedia uses itself for representing powers of two sizes. As can be seen from this link to a search [9] the sizes of the pages use KB. As can be seen from the history page for the Cat article it shows "16:29, 6 April 2008 Ramdrake (Talk | contribs) (88,610 bytes)" and this displays in the search results as "87 KB (13627 words)". Since 88,610/1024 = 86.533203125 = 87 KB (rounded up) it is therefore a fact that Wikipedia search uses KB in the powers of two sense and doesn't need to disambiguate using IEC prefixes. Fnag aton 09:29, 7 April 2008 (UTC)[reply]

Wikipedia should use correct prefixes not wrong ones just because they are still in wide use. This is an encyclopdia that should focus on the truth and not on public opinion. Please also remember the confusion of somewhat inexperied users buying a 320GB harddisk and expecting to get 320 GiB but only receiving 320GB (312.5 GiB). To avoid confusion or reverts on articles with binary prefixes they probably should have a hidden notice on top of the article like "This article use Binary prefixes" and/or this info should go to a Notes section just above references.--Denniss (talk) 13:02, 7 April 2008 (UTC)[reply]

My goal is to minimise ambiguity to the greatest extent possible without compromising the fundamental objective of an encyclopaedia as I see it, which is to provide informative and readable articles.

I see two main disambiguation approaches. These are a) quote exact number of bytes; b) use IEC prefixes. What I would like the guideline to do is spell out both approaches, and encourage the editors of individual articles to consider which of the two approaches works best for that particular article. I believe this flexibility is needed due to the wide variety of articles that we need to cater for. One size doesn't fit all.

Unfamiliar terms should be linked on first use.

Thunderbird2 (talk) 18:43, 7 April 2008 (UTC)[reply]

I'm still working on how to phrase my goals, but I wanted to say that I can't help noticing that people are using this section to, yet again, voice their opinions rather than state their goals. This makes it seem worryingly like a number of editors' goals are to win the debate... SamBC(talk) 18:52, 7 April 2008 (UTC)[reply]

Goals Produce a guideline that's in keeping with the spirit of wikipedia. Wikipedia is not the place to right great wrongs. We report the truth as found in reliable sources, neutrally and giving due weight to majority and minority positions. Position Regardless of what might be argued as "correct", we should use the units that our sources use - and that our readers understand - in the context of the article in question. If those units are ambiguous (I bought a hard drive in the '90s whose manufacturer defined 1MB as 1,000 KB, or 1,024,000 bytes) then we must make clear what the units mean - whether that is by wikilinking, footnoting, etc. I do not mind. The one thing I do not think I could agree to would be a guideline encouraging use of KiB etc in articles where the sources do not use those units. Sheffield Steel^talk_stalk 18:59, 7 April 2008 (UTC)[reply]

I agree with everything Fnagaton and Sheffield Steel wrote above, and which SWTPC6800, DavidPaulHamilton, and agr wrote below regarding goals. The Goal should be that Wikipedia always embraces the time-accepted objective of all technical writing: to communicate clearly, comfortably, and with minimal confusion to any given target readership (which can vary depending on the technical level and subject) using level-appropriate terminology and units of measure that the readership is accustomed to encountering on that subject.
As that applies to this debate, wherever a unit of measure—such as kilobyte—has an ambiguous meaning, editors should not vary from widely adopted, familiar industry practices and should disambiguate where appropriate using terminology and techniques that are themselves familiar and well recognized by that readership. In short, Wikipedia should use terminology that 1) a typical, well-read reader who will be visiting that article already recognizes in order that they can quickly and easily learn more about the subject, and 2) a relatively novice readership should recognize in order to be properly primed to absorb what they will likely encounter elsewhere on the subject.
An important component of achieving this objective is to have consistent practices across all of Wikipedia’s articles so a unit of measure has consistent meaning and connotation no matter which Wikipedia article a reader visits. Further, Wikipedia’s articles must be consistent with the way Wikipedia itself calculates file sizes (which currently eschews the IEC prefixes and observes the convention of 1024 bytes per “kilobyte”). To eventually achieve that end, editors should use only agreed-upon editing practices that afford the greatest possible courtesy and patience to all the volunteer editors who have labored on many of these articles and have helped make Wikipedia what it is.
My stated goal could be achieved with this MOSNUM policy:

Units of measure: In any given article, editors should use the units of measure and methods of disambiguation commonly used in that discipline to communicate to a given readership.

Greg L (my talk) 20:27, 7 April 2008 (UTC)[reply]

I have a decade of experience on standards committees and know that the successful standards are quickly adopted by industry and users. If the potential users of the new standard don't think the improvement outweighs the cost of change, the standard is ignored. My goal would be for Wikipedia to use the same terminology that the leaders of the computer industry use. Intel, Samsung, Apple, Microsoft, PC World, IEEE Spectrum and everybody else has not seen the benefit of switching to the IEC binary prefixes. The MOS states we should use the "units employed in the current scientific literature". After a decade, the IEC prefixes are virtually unused outside of a few standards groups. -- SWTPC6800 (talk) 20:41, 7 April 2008 (UTC)[reply]

Unfamiliar terms like IEC prefixes should not be used when the sources do not use them.DavidPaulHamilton (talk) 14:02, 8 April 2008 (UTC)[reply]

My goal is to do what is best for our readers. Wikipedia is written for a general audience. The one thing universally agreed upon in the polls above is that our audience is not familiar with the IEC prefixes. They have not been adopted by print encyclopedias, newspapers, trade magazines or the major computer and operating system vendors. And given the time that has passed since the IEC prefixes were introduced, this failure to adopt can be considered a rejection of their use in literature aimed at the general public. Wikipedia is a tertiary source. We should follow the practice other print publications, not attempt to lead them. There are other ways to resolve the ambiguities in industry use of the terms megabyte, etc., and we should use them to the extent we have a sourced basis for doing so. (Sourcing can be an issue, particularly when dealing with articles on computer history.) The spirit, if not the letter, of WP:SOAP, WP:NEO, and WP:V should apply here.--agr (talk) 14:35, 8 April 2008 (UTC)[reply]

My primary goal at this point is to end this stupid debate. It's been going on for years now and generated a lot of incivility and wikidrama with no resolution whatsoever. Previously, my goals included minimizing confusion, eliminating ambiguity, and maximizing reliability of Wikipedia as a reference work. After several decades, the idiosyncratic redefinition of SI units are virtually unused outside of a few sub-fields of computing, are not used consistently even within those fields, and are virtually unknown to the average reader. Since Wikipedia is a general reference work spanning many fields, and not a computer science textbook, we should use the standardized units used in the real world. — Omegatron (talk) 01:28, 10 April 2008 (UTC)[reply]

My goal is to do what is best for our readers. Wikipedia is written for a general audience most of whom do not understand IEC prefixes, binary or conventional, but to the extent they do, they are far more aware of decimal meaning of conventional IEC prefixes that they are of EITHER their alternate binary meanings OR the IEC binary prefixes. Most of the computer industry uses the conventional IEC prefixes with either meaning, frequently on the same line or page and without disambiguation (see, e.g. this HP webpage) or for that matter any advertisement for a laptop, desktop or server. WP should do better and it seems to me that the use of IEC Binary Prefixes is one part of the solution. Again to the general audience, which is less confusing, 2 GB (2 x 2³⁰ bytes) or 2 GB (2 GiB)? I suggest that the general audience will not understand either but is far better served by a wikilink to GiB wherein the distinction can be explained in great detail. BTW, 1,???,???,??? bytes is not much better (I can never remember the specific number) That is why I oppose any effort to deprecate binary IEC prefixes Tom94022 (talk) 01:29, 11 April 2008 (UTC)[reply]

My prime goal in working on any wikipedia article, or any other tutorial writing that I do for that matter, is always to leave the reader with more information and less doubt or confusion than before reading my words. I believe that confusion between e.g. "MB = 1,048,576" and "MB = 1,000,000" is rampant in the real world—as most eloquently stated by the IBM employee quoted by Omegatron above—and that this is something we can and should help reduce and explain while we're writing about, for example, computer architectures and address spaces and hard drives. My goal here is a policy that will at least allow, if not promote, the use of IEC prefixes where appropriate (i.e. where a power of two divisor was used to obtain the quoted significand), while providing enough explanation or wikilinks so that the reader unfamiliar with these prefixes will still come away with greater understanding and less confusion than before. I further believe that the goal of "greater understanding, less confusion" justifiably supersedes the principle of following established usage in the rest of the world, particularly since established usage in the rest of the world has caused the problem! Precedent can serve to guide us, but the mistakes of the past should never be assumed to bind us. Just like all WP rules and guidelines, for that matter: WP:IAR n.b.: This section according to SMcCandlish is not for arguments, so I will move any responses (even agreements) to this "goal" to another section. And Omegatron, I suggest you do the same with the responses to your "goal" above. Jeh (talk) 07:29, 11 April 2008 (UTC)[reply]

Stupid debate

Moved a series of responses to Omegatron goal statement above. If anyone has a better section name please feel free to change this one. -- SWTPC6800 (talk) 21:28, 12 April 2008 (UTC)[reply]

If you really think "this stupid debate" then you don't have to keep on writing here. I think it is unhelpful for you to state the debate is stupid. The international standard units for computer memory are those defined by the JEDEC and those are KB/MB/GB using powers of two. By the way, your three stated goals are better accomplished by not using IEC prefixes in the vast majority of cases, I've already shown why with my previous posts. Fnag aton 10:02, 10 April 2008 (UTC)[reply]
It's stupid because it's gone on for three? years with no resolution and everyone repeating the same arguments over and over again: The best way to accomplish my stated goals would be to consistently use SI prefixes for decimal units and IEC prefixes for binary units. It's simple, unambiguous, standardized, and wouldn't confuse readers like the KB = 1024 convention does. — Omegatron (talk) 23:46, 10 April 2008 (UTC)[reply]

The honest truth? I’m well read on computer industry issues. My “Mac news & gossip” tab on my Safari browser has 37 web sites that I open all at once and skim through. I subscribe to Mac World. I’ve pored over thousands of data sheets on memory chips when I helped design industrial control equipment (some of which I have patents on). And I had simply never run across terminology like “kibibytes” before landing here on Wikipedia. Never. You don’t see that kind of stuff in PC World either. Nor in Encyclopedia Britannica. The first time I came across "MiB” on Wikipedia, the thought that flashed through my mind was “Oh God! The people who go to Star Trek conventions wearing Spock ears have hijacked Wikipedia.” That’s not intended as a personal attack whatsoever. It’s the simple truth; an observation intended to help yank some authors here back to reality! Do you really think I’m the only one who has had that reaction to seeing Wikipedia’s computer-related articles? No other general-interest computer magazine uses the IEC prefixes, Omegatron. None. No other professional print encyclopedia uses them. And your position seems to be that the big sucking wound of ambiguousness inherent in the conventional prefixes is so bad that Wikipedia can be all alone leading the charge on this one. I just can simply not fathom why it is so difficult to get this point across to you: Encyclopedias use the units of measure a well-read reader is accustomed to seeing (100% of the time in this case) in that discipline and simply do not adopt some organization’s proposed new units before the rest of the world does. It is just preposterous that it’s taking so much effort to get sanity to prevail here on Wikipedia. My spell check recognizes an extraordinary amount of computer technology terms and proper nouns. But I simply refuse to teach it “KiB” and “kibibyte”. {Ignore}… {Ignore}… (that one too)… {Ignore}… *(sigh)*… {Ignore}. Now that’s stupid. Greg L (my talk) 00:29, 11 April 2008 (UTC)[reply]

I don't know what exactly qualifies as "general-interest computer magazine" but the German magazine PC Games Hardware has switched to the Ki/Mi/Gi prefixes last year. So there are professional, commercial magazines using the new IEC prefixes. It's not "none". --217.87.103.237 (talk) 22:53, 11 April 2008 (UTC)[reply]

And they delimit numbers way differently in Germany than in the U.S. But this is the en:Wikipedia. Greg L (my talk) 03:45, 12 April 2008 (UTC)[reply]

What has delimiting of numbers to do with anything? And no it's not entirely different, you just have to swap comma and period. Just for your information, "en" stands for English, not for US-American. It refers to the language, not nationalities. --217.87.103.237 (talk) 03:57, 12 April 2008 (UTC)[reply]

It means the en.Wikipedia doesn’t use the German language and other German conventions and it doesn’t matter what a German magazine does. It matters only what it takes to communicate most effectively to an English-speaking readership. Greg L (my talk) 04:28, 12 April 2008 (UTC)[reply]

You already know the IEC 60027-2 is an internal standard which has nothing to do with German at all. The example was just to show that your claim "there are no magazines using this standard" is incorrect. --217.87.103.237 (talk) 04:49, 12 April 2008 (UTC)[reply]

Greg, you are right about the Star Trek crowd, but I don't think it is the Spock fans, it is the ones that have learn to speak Klingon that are foisting kibibytes on us. -- SWTPC6800 (talk) 02:19, 11 April 2008 (UTC)[reply]

Would you care to explain why you keep on adding these kind of comments? Do you seriously think you can insult anyone but yourself with such statements? --217.87.103.237 (talk) 21:49, 11 April 2008 (UTC)[reply]

Greg L, what's it with you and Star Trek? It's not the first time you bring this up but I really hope it's the last. You're not doing yourself or your position the slightest favour by trying to "argue" this way. --217.87.103.237 (talk) 21:53, 11 April 2008 (UTC)[reply]

Oh… come on… Lighten up a little. kibbles ’n bits. Greg L (my talk) 03:37, 12 April 2008 (UTC)[reply]

There's some tiny flaw in your joke because "kibble" has as much to do with kibibits or kibibytes as this this. --217.87.103.237 (talk) 04:12, 12 April 2008 (UTC)[reply]

We already understand your position; you don't need to keep repeating it. The point has already been gotten. Everything you're saying has been said over and over again over several years, even before you registered an account. We understand that you don't think there's a problem with ambiguity, that you find the IEC prefixes alien, unnecessary and stupid. The software that you use and literature that you read doesn't use them. You think that people who use SI prefixes for decimal quantities in software, marketing, computing magazines, and computer science papers are only doing it to mislead, or because they're unaware of the powers-of-two convention. You suspect that hard drive manufacturers switched from binary prefixes to proper SI prefixes at some recent point in history purely because it made their drives look bigger than the numbers reported by Windows. We understand that you think the only way someone could possibly support international standards is if they're irrational pedantic nerds. As an IBM employee puts it:

"Megabytes have always been base 2, and always been written as MB. Everyone knows that 1MB is 1024KB, unless you’re talking about DVDs, or reading manufacturer specs for a hard drive, and that’s just the hard drive manufacturers being stupid. Everyone knows that ‘K’ on a computer means 1024; except for speeds, where it means 1000, except for file download speeds where it means 1024, except when it’s the speed of your modem, when it’s 1000. Everyone knows that. What, are you stupid?"

We completely understand your position, and we continue to disagree with it. You should make an honest effort to understand our position instead of engaging in personal attacks and trying to wage a "campaign" against something that dozens of other people think is a good idea. We are not, in fact, mentally deficient morons. — Omegatron (talk) 00:59, 11 April 2008 (UTC)[reply]

Nope. That’s not an accurate description of my position. So we clearly could benefit from the efforts of SMcCandlish to mediate here. But I do agree with you that this debate has been raging for years; from the very beginning since you first rammed it through without a proper consensus (as you yourself stipulated). The current policy has produced nothing but endless bickering and has proven to be an utter failure and must change. Or… are you thinking it has somehow been a raging success and now serves as a paradigm of a good Wikipedia policy? Greg L (my talk) 01:48, 11 April 2008 (UTC)[reply]

Sigh. I thought we were going to avoid counter-arguments, at least, in this "goals" section. :( Jeh (talk) 07:04, 11 April 2008 (UTC)[reply]

Search result methodology

[Refactored out of the goals section which was for statements not argument.]

FYI, search results are cached, so the page size reported there may not correspond to the current version of the page. Gimmetrow 16:12, 7 April 2008 (UTC)[reply]

I conducted multiple tests with lots of different articles to check. There just isn't space to include all of the tests carried out and the single example proves the point succinctly. Fnag aton 16:23, 7 April 2008 (UTC)[reply]

You do understand the search results *are cached* - that means 87 KB there may or may not have anything to do with the 6 April version. For instance, a search for my talk page currently lists it as 49 KB (7420) words, with last size in history as 49736 bytes, and it reports 48 KB when edited. This is just a FYI about your methodology. Gimmetrow 19:02, 7 April 2008 (UTC)[reply]

Yes I know. That is why I conducted lots of tests to make sure. The time and date of the "cached revision" is listed on the search page and I was careful to choose the exact revision in the article history corresponding to the exact time and date listed in the search results. The article I cited belongs to those search results that are updated quite often, i.e. the caching time is quite small. This means the 87 KB I cited is directly related to the size of the article history I cited. The conclusion is therefore the same, that is Wikipedia uses KB and kilobyte in the powers of two sense and doesn't use IEC prefixes in the example I gave. Fnag aton 10:51, 8 April 2008 (UTC)[reply]

Oh. I thought the date was just the date of last edit and wasn't connected to the search size. I've tested it to make sure, and you're right. Gimmetrow 04:02, 10 April 2008 (UTC)[reply]

Well isn’t that an inconvenient truth?? Indeed, Fnagaton, Wikipedia’s itself follows the industry-wide practice where “KB” in the context of file size equals 1024 bytes. I’ve long been keeping careful track of certain article’s file sizes and have long known this obvious fact. Yet I’ve never made the logical connection between that reality and this debate. It’s sort of a ‘Well… Duh!’ point that should have been brought up here earlier. I’ve got a special test page to exercise the {{delimitnum}} magic word and the template by the same name here at User:Greg L/Delimitnum sandbox. It has been stable for quite some time and I’ve been keeping track of its size. It measures precisely 402,845 bytes in size. And what happens when you click the edit this page tab on it? Anyone want to venture a guess what Wikipedia itself says is the article’s size? Anyone?
*(sound of crickets chirping)*
Does it say it is “393 kibibytes” or “403 kilobytes”? No. It uses the convention which causes the least confusion among readers; the same convention used by the rest of the computer industry and general-interest computer magazines: “This page is 393 kilobytes long.”
That’s also why this table showing the total HTML download burden of four, large Wikipedia articles on the nuclides uses “KB” in the conventional, 1024-byte manner: so Wikipedia doesn’t baffle the reader with absolutely schizophrenic behavior. Greg L (my talk) 17:28, 7 April 2008 (UTC)[reply]

That doesn't mean anything. The user interface itself has even been edit warred due to this dispute. The outcome of this guideline determines what happens to those system messages, not the other way around. — Omegatron (talk) 01:37, 10 April 2008 (UTC)[reply]

The three stages of great ideas:

Ridicule the idea as preposterous.
Dismiss the idea as being obvious.
Claim that you thought of it in the first place.

I believe your post above is a variation of step #2 above; sort of “pay no attention to that man behind the curtain!” Greg L (my talk) 03:30, 10 April 2008 (UTC)[reply]

The long page warning does not need to be unambiguous. It doesn't matter if the page “393 kilobytes” or “403 kilobytes" long. "400 kilobytes" is good enough. Most people know that kilobyte is some kind of file size measurement. When someone is editing the article on Ima Hogg they don't need to learn about an obscure computer science measurement that is unused in the real world. -- SWTPC6800 (talk) 02:29, 10 April 2008 (UTC)[reply]

Agreed that it doesn't need to be exact and isn't important enough to fight about. — Omegatron (talk) 01:07, 11 April 2008 (UTC)[reply]

Questions to explore

Ambiguity and consistency, flexibility

The first question that comes to my mind out of the "Goals" discussion above is how to resolve the tension between the view that WP must be consistent, across the board, in how to handle these units, in order to avoid ambiguity and user confusion, versus the countervailing opinion that "one size does not fit all" when it comes to article writing here. Neither of these viewpoints are nutty – MOS effectively enforces consistency in some cases (e.g. always use double-quotes for a quotation and single-quotes for a quotation within a quotation), and by consensus avoids forcing consistency in other areas (UK vs. US spelling, for example). I have an opinion on this, but will reserve it since I'm trying to moderate, so I need to act like a judge not a jury member. PS: Good job everyone on being civil so far. Very refreshing. :-) Anyway, can we come to a consensus that consistency in this case is more important than flexibility or vice versa? — SMcCandlish [talk] [cont] ‹(-¿-)› 23:39, 9 April 2008 (UTC)[reply]

Regardless of the style we settle on, consistency is of critical importance here. In other areas, such as the American-British example, it's not so critical that all articles be consistent—"color" and "colour" mean exactly the same thing. On the other hand, we must settle here on terms which have a single, unambiguous meaning, and then use those terms consistently across the board. Inconsistency would only increase the ambiguity problem already present, and would be the worst of all possible outcomes. Seraphimblade ^{Talk to me} 00:12, 10 April 2008 (UTC)[reply]

Consistency with what? Consistency with sources? Yes. Consistency with Wikipedia? Yes. Consistency with the IEC? No. DavidPaulHamilton (talk) 00:17, 10 April 2008 (UTC)[reply]

The practices of other encyclopedias should serve as a guide here. Encyclopedia Britannica doesn’t use “megabyte” differently depending on which article you read. The reasons for using terminology consistently from article to article within an encyclopedia are too obvious to belabor here. When you (SMcCandlish) ask whether consistency is “more important,” I submit that we may have to go more fundamental than that and ask whether or not our top objective is to appease factioned camps of editors in order to avoid editing conflicts (which doesn’t seem to solve anything in the long run), or if the objective is to have the best possible encyclopedia for the visiting reader. If it’s the latter, then consistency in this case is more important than flexibility. Greg L (my talk) 00:18, 10 April 2008 (UTC)[reply]

Yes, consistency across the entire project is important, which is why we need to follow the international standards. :) See the problem here? — Omegatron (talk) 01:07, 10 April 2008 (UTC)[reply]

JEDEC define international standards for computer memory and their standards define KB/MB/GB in powers of two sizes. Fnag aton 09:37, 10 April 2008 (UTC)[reply]

I would prefer to see consistency between articles, but see that as unattainable in the short term. It is even more important (and attainable) to have

consistency within articles
unambiguous use of prefixes (of whichever variety) within articles

Thunderbird2 (talk) 06:08, 10 April 2008 (UTC)[reply]

Those strike me as two bullet points worthy of a clear show of consensus (or not). I.e., if we can agree on those (or some alternative) then we have a basis from which to move forward collectively, whatever our disagreements. — SMcCandlish [talk] [cont] ‹(-¿-)› 09:35, 10 April 2008 (UTC)[reply]

Yes, I think it's safe to say that everyone wants unambiguity and consistency within articles. It's consistency between articles, and the method used to achieve this consistency, that are the problems. — Omegatron (talk) 01:09, 11 April 2008 (UTC)[reply]

Using IEC prefixes for disambiguation is ambiguous. Here is an example: Using the MiB prefix express 10,000,000 bytes using one decimal place (9.5 MiB). Then express the quantity of 9,970,000 bytes using the MiB prefix with one decimal place (9.5 MiB). Then express the quantity of 10,001,000 bytes using the MiB prefix with one decimal place (9.5 MiB). Then express the quantity of 10,010,000 bytes using the MiB prefix with one decimal place (9.5 MiB). As you can see the different quantities of bytes are shown using the same 9.5 MiB. This makes using MiB ambiguous unless you start using lots of decimal places of accuracy. Using lots of decimal places does not make it easier for the average reader to understand. What does make it easier for the average reader to understand and what does make prefixes (using any system) cited from sources completely unambiguous is disambiguate (using parenthesis or footnotes) by explicitly stating the exact number of bytes being used in the context of the artcle and the sources relevant to the article. Fnag aton 09:40, 10 April 2008 (UTC)[reply]

Clearly, this means that we should give up on the idea of using units altogether. — Omegatron (talk) 01:10, 11 April 2008 (UTC)[reply]

Give up on the idea of using IEC prefixes to disambiguate, yes. Using the units found in the reliable sources relevant to each article, we still have to do that because that is part of writing the articles. So since disambiguation using IEC prefixes is a bad idea we can state the number of bytes instead. Fnag aton 14:05, 11 April 2008 (UTC)[reply]

Again you are stating your opinion and value judgement ("...is a bad idea") as if it were established fact. You are of course free to do that... Jeh (talk) 22:47, 11 April 2008 (UTC)[reply]

Ambiguity and understandability

The second most obvious point to discuss is whether one, the other, both, or neither unit symbols are useful in Wikipedia. If I am reading the debate correctly, the major viewpoints are:

The KB/MB/GB-style symbols are perfectly clear despite having multiple meanings, because the meaning is clear from the context. Sometimes 1 KB is 1000 bytes ~~(and is equivalent to 1 KiB)~~ (which is equivalent to 0.9766 KiB) and ~~sometimes~~ usually it is 1024 bytes, but we know which is what by whether we are talking about storage or RAM. (The counter-argument is that our readers cannot always be expected to understand the difference.) I think we have people agreeeing with some errors. Isn’t the above correct with my underlined corrections? Greg L (my talk) 18:51, 10 April 2008 (UTC)[reply]
Stick with KB/MB/GB and just translate for the reader, e.g.:
With decimal meaning: 64 MB (64×10⁶ bytes)

With binary meaning: 64 MB (64×2²⁰ bytes)

or something like this, and avoid the KiB/MiB/GiB style. (Counter: This is awkward and could potentially mislead readers who see the first into thinking that MB always means 10^6 bytes, or those who see the latter into thinking it always means 2^20 bytes.)
The KB/MB/GB-style symbols are essentially "polluted" for WP purposes, and should not be used at all. Where appropriate KiB/MiB/GiB should be used because they are definable by a standards-issuing body, and otherwise specific byte amounts should be used. (The counter-argument is that KiB-style units are not well-accepted, and our readership won't, on average, understand them.)
The KB/MB/GB-style symbols should stand for 1024-based, not 1000-based units, period, and units that could be expressed in KiB/MiB/GiB-style units should be given in long form. The fact that the latter are nominally standardized is irrelevant, because the standard has not been widely adopted. (Counter: Many readers will still think of them in 1000-based terms and get confused; also, we do not ignore sourceable standards on the basis of their adoption level today; it might be very different tomorrow, and at least it satisfies WP:V.)
They're both humped; specify byte numbers in every case instead of using abbreviatorial symbols, even if this is awkward. (Counter: It is awkward to do this, and we can expect later editors to change "1,700,000 bytes" to "1.7 MB" or ~~"1.7 MiB"~~ "1.6 MiB" with either the result of "MB" ambiguity or "MiB" unfamiliarity; so we need to pick one of the above options, else we'll have to editorially correct them to full byte values day after day).

Are these summaries inaccurate in any way? Are any missing? — SMcCandlish [talk] [cont] ‹(-¿-)› 10:34, 10 April 2008 (UTC)[reply]

Yes, that's about it. We all know the computer harddrive industry mostly uses KB as 1000. We all know that the computer memory industry mostly uses KB as 1024. We know the computer industry as a whole very rarely uses IEC prefixes. Wikipedia has to rely on sources for its articles. So we have to decide on what gives most benefit to the reader. One thing that is universally understood and are universally unambiguous are exact numbers. When we say, "there are 30 apples" we know exactly how many apples there are. When we say "there is a box of apples" this box might contain 30 apples, but it could also be 32 or 28 apples. If a "standards body" comes along and then says "there is this standard box and it shall be used for all apples" yet the fruit growers don't use the standard box, the market stalls don't use the standard box and the general public don't know about the standard box, then obviously the "standard box" is not a de-facto standard. Fnag aton 10:45, 10 April 2008 (UTC)[reply]

Thinking some more. I don't think anybody is saying "never use IEC prefixes" because they do have their place in articles where the sources mostly use IEC prefixes or the topic is specifically about the IEC prefixes. Also I don't think anybody is trying to say we should only use KB as 1024 bytes and disregard the 1000 byte version because as you say there are plenty of sources that show use with either way. What is clear from the majority of comments in the above sections is that using IEC prefixes is not preferable for the majority of topics, so we need to agree on something else like using the exact number of bytes for disambiguation. (To be prefix agnostic.) Also I have to point out a small error in your summary: "Sometimes 1 KB is 1000 bytes (and is equivalent to 1 KiB)" this is because 1 KiB is not 1000 bytes, it is 1024 bytes. ;) But I do think this demonstrates the sort of general confusion the IEC prefixes cause by being virtually unknown. *Even bigger grin*. Fnag aton 13:38, 10 April 2008 (UTC)[reply]

I noticed this immediately too. He even did it twice, also in the example of 1,700,000 bytes. This doesn't seem to be explainable by a typo. Apparently he got the meanings of KiB/MiB/GiB backwards. Considering that the relevant articles or tables can hardly be misunderstood, it says very little about the IEC-defined prefixes. I hope SMcCandlish can clarify this. --217.87.124.227 (talk) 18:03, 10 April 2008 (UTC)[reply]

I would include a coupe of additional options. (I've numbered them in the hope SMcCandlish will go back and number his options).

6. Use KiB/MiB/Gib when the meaning is clearly binary and use KB/MB/GB when it is clearly decimal or when we aren't sure of the exact meaning but that is what the source uses. In oddball situations that aren't covered, give the actual number. (I think this is the natural option if we go with IEC prefixes).

7. Write an article that explains the whole story and link to it from all articles that deal with computer memory. There are various place such a link might go, including:
- After the first occurrence of a prefix with binary meaning
- After each occurrence of a prefix with binary meaning
- In a Notes section
- In See also
- In a head note template (This article deals with computer memory. See XXX for an explanation of units used.)
- or some combination.

(This is my preferred option. It informs readers who are interested, without getting in the way of others.)--agr (talk) 14:43, 10 April 2008 (UTC)[reply]

Which we could certainly do, agr, if we can all agree that the shortcomings in the conventional options for disambiguating are so severe and so compelling, that our use of protologisms justifies violating the spirit of WP:SOAP, WP:NEO, MOSNUM:Which system to use, and WP:V so that Wikipedia can be justified in using terminology that no other general-interest computer magazine in the observable universe has seen fit to use.
I submit further, we should all have a show of hands as to who else here thinks we Wikipedia contributing editors are somehow more *enlightened* and somehow know better than the editors at all the general-interest computer magazines and all the professional print encyclopedias like Encyclopedia Britannica and World Book. I know this may seem combative. But there’s no ducking it; this is precisely what is underlying this debate. So let’s see an honest show of hands.

No, I am not more enlightened than the editors at all the general-interest computer magazines and professional print encyclopedias.

Greg L (my talk) 23:19, 10 April 2008 (UTC)[reply]
SWTPC6800 (talk) 00:18, 11 April 2008 (UTC)[reply]
Fnag aton 11:42, 11 April 2008 (UTC)[reply]
Thunderbird2 (talk) 14:21, 11 April 2008 (UTC)[reply]

As a matter of fact, yes, I am more enlightened than all the editors at all the general-interest computer magazines and professional print encyclopedias.

[Your name here]

GregL (I think this was from GregL), this doesn't just seem combative. You're linking a position on a specific issue with a claim to general "enlightenment." "Oh, so you think you're better than everyone else." And no matter that you claim "this is precisely what is underlying this debate," that's a logical fallacy, a loaded question that's on all fours with "When did you stop beating your wife?" and I won't play by those rules. Suppose I had put up a proposal such as "we should continue to use ambiguous and demonstrably confusing prefixes in the computer field just because most of the world has been doing it wrong for decades"? Would you vote "no" on that? What would you think about the person who proposed such a "poll option"? Frankly, I think that your needing to cast your opponents as "claiming to be more enlightened" says something about your position: If you can't make your arguments without such tactics you must not think you have much of a position. (Any responses along the lines of "oh, so you DO think you're more enlightened" are just more of the same, attributing a position not expressed. So save it, please.) Jeh (talk) 17:51, 11 April 2008 (UTC)[reply]

I don’t think there is any weaseling out of the reality that this is precisely the effect of promoting the IEC prefixes. Just because the truth is unpleasant doesn’t mean it can’t be said, Jeh. Real, paid, professional editors at computer magazines and encyclopedias (people who have advanced degrees in journalism and make their careers in technical writing) are all following the common sense principal—one that is embodied in Wikipedia’s own policy—of using the units of measure in the current literature. Yet that is all simply being flouted in this case. It seems the proponents of routinely using the IEC prefixes just somehow know better than the pros and think it is best when Wikipedia is off doing things its own way. That comes across to me as if some of the editors here clearly have extraordinarily high self-esteem but I am convinced it is not proper technical writing. Not in the least. Thus, the poll above; to help you to see this. If you dispute this, that’s OK, but please don’t tell me how I may think or express my thoughts. If you want to counter with your own poll, be my guest. Greg L (my talk) 19:31, 11 April 2008 (UTC)[reply]

Greg, the rules used by "real, paid professional editors at computer magazines and encyclopedias" do not necessarily apply here. Wikipedia is already "off doing things its own way", far different from any print medium I can think of. This is a new medium. With things like Wikilinks (and carefully worded article titles to which they link) the essence of a new prefix can be explained to the curious reader without so much as a mouse click. This is an ability the print editors can never have, so their decisions are not necessarily the best for us. More: I really don't think you'd find all those "paid, professional editors" agreeing that the continued confusion over MB and GB is a good thing. But because they can't stop to explain any change, they have to "go with the flow." I have no doubt that if we could go back and time and suggest to the first person who was about to use KB=1024 "how about using KiB instead", the suggestion would be immediately adopted IF we brought back evidence of the scope of the confusion that exists today. In short: What you're arguing for, in my view, is that the mistakes of the past should continue to be a straitjacket, even in the face of changing circumstances and vastly improved presentation abilities. No, I don't think I'm "more enlightened." Nor do I think I think I'm more enlightened, nor is this my motivation here. I just think Wikipedia is capable of a lot more than, and so should not be constrained by the limitations of, piles of dead trees. As for Wikipedia’s own policy, I once again refer you to WP:IAR. Jeh (talk) 20:09, 11 April 2008 (UTC)[reply]

With things like Wikilinks (and carefully worded article titles to which they link) the essence of the differences in decimal and binary meaning of the commonly used prefixes can be explained to the curious reader without so much as a mouse click and without trying to push a point of view to use IEC prefixes. What you are arguing for is to dismiss real world consensus on this issue and to push your point of view that IEC prefixes should be used, that is not acceptable. Fnag aton 20:17, 11 April 2008 (UTC)[reply]

But after explaining the IEC prefix, its meanng is then known. Whereas with an ambiguously used SI prefix, every usage must be disambiguated and every disambiguation must be checked by the reader (whether it's in the text, or in a footnote, or a popup, or a wikilinked page) to discern the true meaning. That's why the IEC prefix, once explained, is a better solution. "Real world consensus" is what has gotten us into this mess, so its value should be questioned in this case. Jeh (talk) 22:47, 11 April 2008 (UTC)[reply]

Jeh, this is good progress because we’re sorting through the issues. Your argument seems to be that the unique nature and hyperlink power of Wikipedia allows easily linking of “GiB” to an article explaining the term. That still doesn’t satisfactorily address the issue that we’re 1) “fixing” a problem that every other computer magazine, encyclopedia, and general-interest web site (which, I might add, could all also use links in order to explain “GiB”, but don’t) doesn’t perceive as being a problem; and 2) we’re asking readers to learn and remember when they won’t encounter or need it elsewhere. This amounts to WP:SOAP and still violates the spirit of MOSNUM:Which system to use. Regarding Jimbo’s WP:IAR, I agree with broader principal he’s trying to rally us all towards: boldness! Indeed, Wikipedia isn’t a bureaucracy and does have some flexibility to vary from established policy in order to make a square peg fit into a round hole. But diverging so greatly on such a notable, fundamental issue to fix a problem that no one else perceives is a real problem just isn’t justified here. Greg L (my talk) 20:27, 11 April 2008 (UTC)[reply]

I just don't see that this is a huge divergence. It's supported by a standard that has been endorsed by multiple international bodies. It's one extra letter added to existing text, rather than links or footnotes or whatever that "MB here means 1,000,000 bytes" and "MB there means 1,048,576" bytes" scattered all over a page, so I think it is more concise, more quickly understood, and a better solution. Yeah, that's a value judgement. So is every other WP policy. Jeh (talk) 22:47, 11 April 2008 (UTC)[reply]

Jeh, And… now we’ve gone full circle. Yes, the IEC proposal is supported by the IEC and whoever else. We all know that to be true. But we all also know the IEC prefixes are not recognized by the typical reader—a fact you agreed to but declared that the existence of this reality was “not an overriding concern here.” Given that you agreed that the IEC prefixes are not recognized by the typical reader, it logically follows that you realize that this is due to the fact that the IEC prefixes have not seen real-world adoption. So you apparently see no problem with getting on a soapbox here to use and “teach” these units to Wikipedia’s readers even though they will encounter them only here on Wikipedia. You don’t see anything wrong with that; is that correct? Because Wikipedia has convenient Wikilinking tools, it’s OK for us to use unfamiliar terms and units of measure here. Is that right? Greg L (my talk) 02:51, 12 April 2008 (UTC)[reply]

Not answering on behalf of Jeh, but that isn't soapboxing at all. What exactly should apply? I guess you're referring to the word "advocacy". Using an international standard isn't advocacy, neither is reducing confusion/inconsistency in articles advocacy. Feel free to browse through all Wikipedia policies and guidelines but we've been through most of them and so far none was useful which isn't surprising considering that they mostly represent common-sense which all of us are already capable of, if we just try. --217.87.103.237 (talk) 03:14, 12 April 2008 (UTC)[reply]

This is pointless. It’s time for SMcCandlish to step in. This position still amounts to Wikipedia going off doing its own thing, embracing an “international standard” that no other print magazine, on-line magazine, or encyclopedia sees fit to recognize (for damned good reasons), to fix a problem with ambiguity that no one else seems to have a problem with. The end result here on Wikipedia is to attempt to teach a unit of measure to readers that they will only encounter here—on only some articles because Wikipedia isn’t even internally consistent! Why do we have this current state of affairs? Because a handful of well-intentioned editors on Wikipedia, assisted by a single rogue editor (later banned for life) who multiplied the bad decision nearly overnight across a hundred articles, backed for years by an administrator who flouted the rules of consensus and helped block reversion of the rogue editor’s handiwork, somehow just seem to know better than the majority of the other Wikipedia editors and the rest of the English-speaking print and Web world on this one. And this makes sense? We’ll see how SMcCandlish manages to bring this to a sensible conclusion. Greg L (my talk) 04:05, 12 April 2008 (UTC)[reply]

I've just mentioned the de:PC Games Hardware magazine. It's an on-line and print magazine (for damned good reasons). Nobody has a problem with the ambiguity? Then tell me why are so many people screaming bloody murder if their HDD seems to smaller than claimed and even make up urban legends like the HDD vendor conspiracy? Why has the IEC come up with 60027-2? Why has IEEE adopted it? Because nobody has problem with the ambiguity? By your arguments we could or even must remove everything from Wikipedia that is not widely known and publish false information instead. Or even better: Remove everything that a couple of editors claim to be widely unknown. By the way, "unknown" and "unused" isn't the same. Millions of people do know the prefixes whether they like it or not, whether they use them or not. --217.87.103.237 (talk) 04:42, 12 April 2008 (UTC)[reply]

No, my position is that for any given target readership, Wikipedia should use the units of measure used in current literature on that subject. (Hey, I rather like that; I think I’ll start another poll.) I’m not suggesting that Wikipedia “remove everything from Wikipedia that is not widely known and publish false information instead.” That is a metric ton of weapons-grade bullonium. The only logical counter to my here-stated position is that the rest of the English-language publishing world (PC World, Mac World, Encyclopedia Britannica etc.) and the Web-based equivalent publications, and the manufacturers of computer equipment, all have it all wrong but you guys are right. Well, just pardon me all over for not recognizing that I had landed in the forum of the technical writing gods. Why don’t we just go off and do things our way shall we? Greg L (my talk) 20:59, 12 April 2008 (UTC)[reply]

There is a risk of underestimating the difficulty of the disambiguation task. In my view there is no one solution of those described above that solves all of the problems. Further, the actual problems encountered vary from article to article. Because of this, MOSNUM should prescribe not one disambiguation method but a range of acceptable ones (to be decided). The advantages and disadvantages of each approach should be spelled out to help the editors make a good decision for each article. The hope is that over time it will become clear which disambiguation methods are adopted in practice, and at that point perhaps true consensus can be reached. But if we are over-prescriptive and enforce an impractical solution now, there is a danger the guideline will just be ignored. To come to my point, I would like to see one more option added to the list, along the lines of:

permit two or more different (specified) disambiguation styles; describe the associated pros & cons with each; let WP evolve until a clear preference emerges.

Thunderbird2 (talk) 18:24, 10 April 2008 (UTC)[reply]

Picking up on what T-bird touched upon above (“MOSNUM should prescribe not one disambiguation method but a range of acceptable ones”), please note that in my hybrid proposal, I used some example footnotes to illustrate ways to disambiguate. There were two tables in the proposal that showed five different ways of expressing equivalencies; all of these would be perfectly acceptable ways to disambiguate. Further, disambigutions don’t have to be in the form of footnotes; they can be as in-line parentheticals or whatever any given editor thinks is suitable given the context and where he or she is going.
SMcCandlish: I believe the root of the problem between the editors is centered around your third bullet point above but should be rephrased to focus on the crux of the dispute: Whether it is good practice for an encyclopedia to use units of measure that the average reader is unfamiliar with. Not a single author disagreed with the fact that the IEC prefixes are not widely recognized by the typical Wikipedia reader. Most of the editors who have weighed in on this issue lately don’t believe it is the proper thing to do. The counter argument is that the prefixes hijacked from the SI are used ambiguously in the real world and Wikipedia should recognize and use the IEC prefixes even though this has the effect of “leading the way” given that they are only rarely if ever used elsewhere in places the typical reader would visit. Greg L (my talk) 19:19, 10 April 2008 (UTC)[reply]

It's just a matter of perspective, as I've tried to point out elsewhere. To people who are only used to seeing the Windows power-of-two convention, the IEC prefixes are obscure, "virtually unused", and confusing. To people who are used to working with SI units, the powers-of-two convention is obscure, virtually unused, and confusing. Both are equally convinced that their usage is in the majority, and both are correct, since they're approaching it from different perspectives. This isn't "pro-IEC" vs "anti-IEC". It's "pro-SI" vs "anti-IEC". — Omegatron (talk) 01:16, 11 April 2008 (UTC)[reply]

Viewpoints:

1. It's not "usually" 1024. 1000 for hard drives, DVDs, network speeds, processor speeds, some software, etc. 1024 for memory and some software.

3. Does anyone really think that "MB" shouldn't be used at all?

4. I don't understand this one. KB should always be used for 1024 and not 1000, but when we mean 1024 it should be written out explicitly? — Omegatron (talk) 01:22, 11 April 2008 (UTC)[reply]

Can we keep this simple and not unnecessarily drag “processor speeds” (GHz) and “network speeds” (megabaud) into this discussion? There’s no need for us to start enumerating a big hairy list of all the weird exceptions to the general rule that we’re really talking about here. Why does this have to be so complex? We’re talking about prefixed forms of the byte; that’s all. I’d be most pleased if we could just focus on the meat & potatoes of this issue: whether or not we can be like the rest of the damn industry and say “The XYZ computer now comes standard with 2 GB of memory.” This is the way the computer manufacturer advertises their computer in PC World. This is the way PC World writes reviews about the XYZ computer. This is what is on the box that contains the XYZ computer. This is what is in the owners manual to the XYZ computer. This is what is on XYZcomputer.com’s Web site. This is what the readers understand and expect. Can we all just simply agree that notwithstanding the well known shortcomings in the way the conventional prefixes are sometimes used in the industry, Wikipedia can conform to the way the real world works and, where necessary, disambiguate with any of a handful of available methods that don’t entail using and promoting unfamiliar units of measure no one else is using? Why is it that the only people who perceive deal-breaker inadequacies with the conventional prefixes are a minority group of editors on Wikipedia? Everyone else in the real world seems to get along with them just fine. Greg L (my talk) 05:10, 11 April 2008 (UTC)[reply]

Greg L, in my opinion, the "big hairy list of all the weird exceptions" shows that what some perceive as the rule is actually the exception to the rule - and a very unnecessary exception, again in my opinion. --217.87.103.237 (talk) 22:37, 11 April 2008 (UTC)[reply]

Greg, please listen to our arguments. You can perhaps avoid GHz but you cannot avoid MB/s. This unit is conventionally decimal but now in need of disambiguation because of the semiconductor industry’s binary use of the megabyte. Why should we wish to emulate that ambiguous practice? The readers expect MB, GB etc only because that’s what they’re used to reading, not because they understand what these units mean. It is the job of an encyclopaedia to explain that. Let me propose an objective that I hope we can all rally around:

The role of MOSNUM should be to encourage those editors who can work out the intended meaning of MB each time it is used to disambiguate that meaning, and to do so in such a way that the article remains readable.

Thunderbird2 (talk) 15:55, 11 April 2008 (UTC)[reply]

He is listening to your arguments, it is just that in the examples given using IEC prefixes would not help the reader as much as stating the exact number of bytes. Using exact numbers is readable and instantly understood because exact numbers do not rely on any other confusing and virtually unknown prefixes. Fnag aton 16:12, 11 April 2008 (UTC)[reply]

Answer this question for me Thunderbird2. Which option from the two below do you find easier to understand the exact number of apples?

1) 100 apples.
2) 3.3 standard boxes of apples.

Fnag aton 16:31, 11 April 2008 (UTC)[reply]

Gosh I wish all questions were this easy to answer. 1) 100 apples :-)Thunderbird2 (talk) 16:38, 11 April 2008 (UTC)[reply]

The question is rather pointless without context because even "100 apples" doesn't necessarily mean exactly 100 apples. Let's assume 100 apples yield about 5 liter of apple juice. In this context you can be almost sure it's neither 5.000 liter nor 100.000 apples for the simple fact that they won't have exactly the same size but it's a valid approximation. How about this instead:

1) 100 bytes
2) 0.1 kB
3) 0.1 KiB

and then

1) 131072 bytes
2) 131 kB
3) 128 KiB

the plot thickens:

1) 214748364800 bytes
2) 215 GB
3) 200 GiB

A little more explicit, few articles talk about "100" bytes but far larger values. Up to 3-4 digits use of prefix isn't always good style because we can easily handle those numbers. Just measure the time it takes you to figure out how to pronounce a number. You really have to count the number of digits if it's more than a handful, don't you? It doesn't even matter whether you're pronouncing it or reading it silently to yourself. You don't grasp a number by reading a chain of digits, do you? --217.87.103.237 (talk) 22:16, 11 April 2008 (UTC)[reply]

OK next question, which one is easier to understand the exact number of apples now?

1) 100 apples.
2) 3.3 [[Standard boxes of apples|SBA]].

The nowiki is intentional, please imagine what the link would actually look like. ;) Fnag aton 17:18, 11 April 2008 (UTC)[reply]

You know that doesn't change my answer. I understand your point, but I don't see its relevance here. Where is the ambiguity in "100 apples" that is somehow resolved by introducing the SBA? And since when has the SBA been adopted as a standard unit by the International Agricultural Commission? Thunderbird2 (talk) 17:30, 11 April 2008 (UTC)[reply]

I'm getting to my point. ;) OK, imagine the SBA as described in my posts above (it's 30 apples) is defined by the SI/IEC. However nobody uses the SBA except Fred from the SI/IEC fruit standards division and a couple of his friends. Does that change your answer from number 1?Fnag aton 17:37, 11 April 2008 (UTC)[reply]

No. There's no need for the new unit because 100 apples is unambiguous, and even introduces an undesirable approximation because 3.3 SBA = 99 apples. Thunderbird2 (talk) 17:45, 11 April 2008 (UTC)[reply]

Next question. The old lady buys three trays of apples for the children of the village from the shop. The shop gets its apples from the distributor, also selling apples in trays, who gets the apples from the grower. These apples from the grower come in trays that hold exactly 32 apples each. How many apples did the little old lady actually buy? Fnag aton 18:00, 11 April 2008 (UTC)[reply]

I don't know, but I'm curious. Hoy many apples did she buy? Thunderbird2 (talk) 08:53, 12 April 2008 (UTC)[reply]

T-bird, I wouldn’t emulate what is being done with MB/s. But that point only unnecessarily clouds the issue we should be trying to address here. Paint me skeptical that you are really arguing that the best way to disambiguate baud speeds is like this: “Comcast’s normal speed is 6.00 Mb/s (5.59 mebibits/s)” when we could do it this way: “6 MBaud (six million bits per second)” or any number of available options that don’t rely on either ambiguous usage of units or the unfamiliar IEC prefixes. Can we stay focused on the central point here? The question we should be addressing is this: Can we agree that when writing “the Testosterone-9000 microprocessor has 512 MB of Level 1 cache,” that it can be successfully disambiguated using any of a handful of techniques without resorting to the IEC prefixes? Why not do it the way our own computers have been reporting it to us all along: “512 MB (536,870,912 bytes).” Is that so hard? Why are we still fighting over this IEC prefix issue when there are many perfectly suitable alternative ways to disambiguate that don’t rely on the use of units of measure the average reader has never encountered before? That just so violates the basic principals of technical writing. Greg L (my talk) 16:33, 11 April 2008 (UTC)[reply]

Can you honestly see editors replacing 512 MB with 512 MB (536,870,912 bytes) every time it appears? I can't. And even if they did it would make almost every article unreadable. That is the point you are missing. Where one can disambiguate without IEC prefixes and without confusing the reader then I agree that's a good way to go. It worked well with the DEC alpha article, but I don't believe it will always work well. Thunderbird2 (talk) 16:45, 11 April 2008 (UTC)[reply]

Stay reasonable T-bird or I won’t debate with you. I know full well that you read my hybrid proposal (you voted on it) and it is clear as glass I have not and am not proposing that my above-mentioned parenthetical disambiguation would appear with each occurrence. You were attempting to dodge the issue when you wrote “…but I don't believe it will always work well.” There is no point leaving the door open on this issue because the advocates of the IEC prefixes simply want to keep on using them. That is totally unnecessary given that there are many, perfectly suitable ways to communicate storage capacity without resorting to unfamiliar units of measure. This “IEC prefix” policy has been controversial and fraught with conflict since its inception. Let’s stay focused on the central issue here, shall we? Greg L (my talk) 17:04, 11 April 2008 (UTC)[reply]
I don't accept I was being unreasonable, or attempting to dodge the issue. I could have phrased my question more carefully, though, so let me do so now. Can you honestly see editors replacing 512 MB with 512 MB (536,870,912 bytes) every time it becomes necessary due to a change of meaning half way through a sentence? And even if they did it would make almost every such article unreadable. Thunderbird2 (talk) 17:13, 11 April 2008 (UTC)[reply]

Thank you; that makes more sense. And the answer to your question is “No.” Fortunately, real-world practice rarely has to deal with situations like a mid-sentence change in context, otherwise the IEC prefixes would have likely found traction in real-world usage. Usually, a discussion of hard drive capacity is just that—a discussion of hard drive capacity—and a one-time disambiguation applies for an entire article. Changing the meaning of the unit “GB” would very rarely occur within a sentence (as per your example above). I nevertheless anticipated this worse-case-scenario in my hybrid proposal with my example of a Seagate hard drive. No, it’s not “pretty”, but it works and is only one example of a way of dealing with an extreme situation that only rarely arises. Computer magazines manage to get around problems like this without using unfamiliar units of measure; I think we could too if we really set our minds to it. Would you agree?
I don’t think extreme exceptions to the rule should allow us to get pulled away from the basic principal we should be addressing. Why would we say “in this article, where we discuss hard drive storage capacity and file sizes in near proximity to each other, we will use “GB” to refer to hard drive storage and “GiB” (click the link and learn all about it here) to represent binary values,” if the reader won’t encounter such terminology after they leave Wikipedia?
I agree, I agree, I agree that the IEC prefixes were/are a good idea that addresses a legitimate issue (ambiguity) and should have been adopted by the computer industry. But they weren’t. Just like “20 ppb” (parts per billion) has an ambiguous meaning because “billion” means different values in different countries and the NIST and the BIPM doesn’t endorse its use (although the BIPM tolerates “ppm”), we mustn’t start using “20 nanouno” here on Wikipedia just because it was a good proposal from an influential organization. Why? Because even though the uno is a good idea and has an absolutely unambiguous meaning, we’re not doing our readers a service when we routinely use and promote a unit of measure they don’t initially recognize and won’t encounter elsewhere in the real world. This is simply the reality of the way we must communicate in technical writing. Greg L (my talk) 19:02, 11 April 2008 (UTC)[reply]

I can see editors choosing to add a footnote to "512 MB" that says "536,870,912 bytes". It is completely unambiguous, it is easily understood because it is a number, it is also easily understood because it doesn't introduce any other confusing or virtually unused prefixes. Adding a footnote also means there is less visible cluttering of the article text compared to adding extra parenthesis containing a wikilinked IEC prefix. Fnag aton 18:30, 11 April 2008 (UTC)[reply]

A Tor exit node is making disruptive edits

These two edits have been made by a Tor node and are clearly disruptive. I thought the early edit might have been Omegatron editing his own comment but now I check the edit comment and IP it is also a Tor exit node. [10] [11] Fnag aton 18:34, 10 April 2008 (UTC)[reply]

It's funny how many Tor users and anonymous editors have a strong interest in this topic, don't you think? And it's good to know that our Help documentation has improved so much that users can register accounts, figure out our wiki syntax and policies, and jump right into this discussion without any learning curve at all. — Omegatron (talk) 01:05, 11 April 2008 (UTC)[reply]

I don’t get it, Fnagaton. Please dumb-down your language and express your concern using plain-speak as to what you think Omegatron did that was inappropriate. I do see that it appears a comment by our anonymous “217.87” was deleted but it’s still there now. Is that the issue? I’ve been tolerating 217.87’s comments as long as they weren’t vandalism or somehow violated Wikipedia policy. Greg L (my talk) 19:45, 11 April 2008 (UTC)[reply]

When I first saw the earlier edit I thought Omegatron and forgotten to login and was editing his own comment, at that time I had not seen the change comment, just the text difference. But an IP user reverting material on this page is not unusual, especially with this topic being discussed. So I reverted the edit and placed a comment on Omegatron's page to give notice that I reverted the change that he might have made and if it wasn't him making the change give a warning that someone editied his change. Then I saw the second edit and noticed it was reverting the IP user comment. This is disruptive. My standard approach is to check the Tor status of IP addresses that are making edits that are obviously disruptive. I noticed this edit is a Tor exit node, i.e. someone trying to hide their true identity. So I reverted the edit. I then checked the other IP address and this was also a Tor exit node. Two Tor node edits on this subject then reminded me of User:Sarenne who was the last user to be blocked for severely disruptive behaviour by forcing IEC prefixes into hundreds of articles and used Tor with multiple sock puppets. Fnag aton 20:12, 11 April 2008 (UTC)[reply]

So… Nothing can be proven and Omegatron gave a Cheshire cat-like, cryptic answer to tweak your nose? Is that right? Greg L (my talk) 20:35, 11 April 2008 (UTC)[reply]

Looks that way, yes. Fnag aton 21:17, 11 April 2008 (UTC)[reply]

Stunts like that is just horse crap we can deal with. Even as an admin (a power Omegatron has apparently abused since he first got it), he can’t make history files disappear. This is the opposition’s leader? Would you rather be up against a formidable one? He rammed through the current MOSNUM policy without a proper consensus and the years of endless bickering that have gone on ever since is all a result of his initial goof (and intransigence afterwards). Greg L (my talk) 21:44, 11 April 2008 (UTC)[reply]

That's not correct. Admins can erase edit histories. At least in so far that they won't be available to mere users anymore. Further, Omegatron isn't a leader in anyway. Different people just happen to be on different sides of the debate. Some are more vocal than others. Some have more stamina than others. Some care more, some care less. What you hold against Omegatron and similar could just as well be held against many participants on either side including you and me. I think most participants have cooled down and try to avoid getting emotional even if the tone is often still less than friendly. I didn't notice any questionable comments from Omegatron ever since you've joined the discussion and nobody should try to judge another by looking specific edits and excerpts of discussions that they weren't involved in. I'd really appreciate if everyone focused on the discussion at hand instead of questioning others' motivation. Too much time has already been wasted by trolls like those Tor edits and similar trolling comments from logged-in users who contribute little more but one liners to spit on either side. --217.87.103.237 (talk) 23:27, 11 April 2008 (UTC)[reply]

Very well. Greg L (my talk) 23:58, 11 April 2008 (UTC) P.S. But would you agree that referring to tracking down suspicious Tor edits as “trolling” is too harsh? If we didn’t have someone really in the saddle, tracking down this sort of activity, it could get out of hand really, really fast if certain others didn’t have a good idea they’ll be called to the mat for it. Fair enough? Greg L (my talk) 01:54, 12 April 2008 (UTC)[reply]

That's not what I was referring to as trolling at all. What I consider trolling were the edit summaries used by the Tor user giving the impression that he was a certain someone in order to stir things up and derail the discussion. I assume these two edits were not submitted by anyone involved in the discussion but a bored lurker. Another example are several edits by Swtpc6800, especially his most recent contribution [12]. --217.87.103.237 (talk) 02:09, 12 April 2008 (UTC)[reply]

I see. Very good. Case closed then. Thanks for the clarification. Greg L (my talk) 02:16, 12 April 2008 (UTC)[reply]

- No, the case is not closed. For the sake of a balanced and calm discussion, I think that the uncivil and patronising remarks about Omegatron's role are unjustified and should be withdrawn. I will resume the debate once that happens. Thunderbird2 (talk) 09:07, 12 April 2008 (UTC)[reply]

I actually took Omegatron's comment ( 01:05, 11 April 2008 ) to mean that he is implying it is someone here making those edits with Tor and with the IP user. So he should remove his comments first since they are uncivil. Fnag aton 14:01, 12 April 2008 (UTC)[reply]

I don't quite get what Omegatron is saying there since his comment seems to be a mix of sarcasm and irony but I wouldn't consider this comment as uncivil and he clearly isn't attacking a single specific participant. For the record, unregistered users are not necessarily less familiar with Wikipedia than registered users. Being familiar with Wikipedia certainly doesn't imply that the person already has a registered account. The most used wiki syntax isn't rocket-science anyway and it's MediaWiki syntax not specific to Wikipedia at all. Nobody claimed that the edits from 217.87.x.x in this discussions up to now are from different users, it safe to assume the opposite, not to mention that it wasn't used in any voting process. --217.87.83.213 (talk) 16:09, 12 April 2008 (UTC)[reply]

I think this is what Greg meant when he referred to the "cryptic answer". This should be easy enough to sort out. If Omegatron did not mean to imply that someone here is using Tor then he can strike the unhelpful comment, then I'm sure Greg can do the same. Fnag aton 14:01, 12 April 2008 (UTC)[reply]

he can’t make history files disappear.

Yes, admins can make history files disappear, by deleting a page and then undeleting only the history they want to keep.

Are you accusing me of this? If I deleted a diff from history without justification, that would be an abuse of admin powers. Which edit do you think I have "vanished"?

If you ever suspect that an admin has done this, it's very easy to confirm: ask another admin to check the history page. It will still appear in the "deleted edits", visible only to admins. — Omegatron (talk) 17:06, 12 April 2008 (UTC)[reply]

And my "cryptic answer" is because I know it's bad form to accuse specific people of sockpuppetry (or strawpuppetry) without specific types of evidence, and I'm sure the people involved are experienced professionals by now and know exactly how to evade that type of evidence (Tor). I think it's obvious who's behind them, but I'll let you all draw your own conclusions. I'll just say to those responsible that your position must be pretty weak if you can't convince anyone of it without resorting to this kind of childish nonsense. — Omegatron (talk) 17:19, 12 April 2008 (UTC)[reply]

So you are bad faith accusing someone here of sockpuppetry without specific types of evidence, that is being uncivil. Greg's comment is completely accurate then. So Thunderbird2 and 217.87.* now we have an admission from Omegatron I expect you'll now retract your posts? Fnag aton 17:27, 12 April 2008 (UTC)[reply]

Why can't we just attribute any malicious anonymous edits to User:Sarenne as before and be done with this? --217.87.83.213 (talk) 18:30, 12 April 2008 (UTC)[reply]

Since it is Omegatron throwing around accusations you had better ask him. Fnag aton 19:08, 12 April 2008 (UTC)[reply]

I am now going to quote Omegatron's edit "Is this seriously how you think? You really think that anyone who opposes something you do on Wikipedia is part of a vast conspiracy of sockpuppets? What if they are merely individual people who disagree with you on a particular issue for good reasons?". I think given the current situation Omegatron you need to take your own words to heart and stop making these kinds of accusations. Fnag aton 17:43, 12 April 2008 (UTC)[reply]

I do you a favour Omegatron by restoring an edit that was obviously made by a Tor user who was trying to misrepresent what you wrote and what do you do in return? You make baseless accusations. Greg's post was completely correct, you are being disruptive. Fnag aton 17:45, 12 April 2008 (UTC)[reply]

I advocate dropping it. There is no proof that Omegatron did the Tor edits. But it’s clear to me that his cryptic response to Fnagaton was intended to suggest that possibility without saying as much. There’s nothing wrong with that either, other than it was intended to keep Fnagaton suspecting Omegatron might have been behind them. Why don’t you guys just call each other a “poopy head” right here, right now, and be done with this so we can move on to more important issues than who failed to use deoderant this morning. Greg L (my talk) 20:37, 12 April 2008 (UTC)[reply]

How about this to break deadlock

I propose we bring in new blood on this. This argument had gone on for years (since the very inception of the existing policy) and there is no resolution in sight. Most of the people who voted on the hybrid proposal have tired of the fruitless bickering here and gone elsewhere to actually have fun at this hobby. Why don’t we post a very simple proposal on Talk:MOS and Talk:MOSNUM? We can have a lively discussion and debate and have a show of hands. The proposal to discuss would be this:

The following wording is intended to add extra specificity to Which system to use:

Units of measure: For any given article and target readership,^[1] Wikipedia should use the units of measure and methods of disambiguation commonly used in current literature on that subject.

To comment on how this policy could be modified and/or expanded here on MOSNUM, please go to this dedicated discussion page.

Notes

1. ^ A Wikipedia article about the mathematics of black hole evaporation necessarily requires the use of different terminology and units of measure—such as Planck units—than would an article directed to a more general-interest readership. A Wikipedia article on, for instance, x86 assembly language that is directed primarily to professional software developers might require units of measure and unit symbols that would be unfamiliar and unsuitable for use in general-interest articles about computers. In all cases, editors should use terminology and symbols commonly employed in the literature for that subject and level of difficulty. When in doubt, use the units of measure, prefixes, unit symbols, and methods of disambiguation used by the majority of reliable periodicals directed to that readership.

I think many more editors would be interested in weighing in on a simple policy guideline such as this.

Well, how about it? Should we solicit input from other editors and keep the issue extremely simple? Greg L (my talk) 21:16, 12 April 2008 (UTC)[reply]

Good idea. Let’s do.

Greg L (my talk) 21:25, 12 April 2008 (UTC)[reply]
SWTPC6800 (talk) 00:40, 13 April 2008 (UTC)[reply]
DavidPaulHamilton (talk) 10:10, 13 April 2008 (UTC)[reply]
Fnag aton 20:36, 14 April 2008 (UTC)[reply]

Bad idea. Let’s not.

That is a nice rule of thumb but it's common-sense. Therefore, there's no need to add it to any guideline. Due to the nature of the issue, it cannot be applied in this case anyway because we're not talking about different units. We're talking about conflicting definitions of the same prefixes. --217.87.100.251 (talk) 18:21, 13 April 2008 (UTC)[reply]

Not true. “Mebibyte” and its symbol, “MiB” incorporate a brand new and different prefix and are therefore different units of measure. If they weren’t “different units”, the IEC wouldn’t have accomplished anything with their proposal. And if it really is “common sense” as you say (I happen to agree), then it is a minor edit and is a change that anyone—as Omegatron pointed out about Wikipedia policy—can add to MOSNUM without prior consensus (although I would add an invitation for more discussion if it were me). That is, unless, you so vigorously oppose adding a “nice rule of thumb [that is] common sense” to MOSNUM that you intend to delete it as soon as you see it. Greg L (talk) 19:52, 13 April 2008 (UTC)[reply]

Breaking deadlock:Discussion

Transplanted from Ambiguity and understandability

Some current literature does conform to IEC 60027-2 and even explicitely uses MiB. Or are you talking about majorities again? In some cases different units are common in current literature. So a rule "Wikipedia should use the units of measure used in current literature on that subject" like does not really help at all. Again IEC 60027-2 is an international standard and it is therefore perfectly valid to use it when writing Wikipedia articles. I could care less about the motivation of magazine editors to stick to the old ambiguous convention. I could think of many reasons but it's nonsense to speculate about them. I'd guess it basically boils down to peer pressure though. Most of those authors can also assume that their readers are already familiar with the basics and terminology. If you really wonder you might want to ask some of these journalists. I'm also convinced that by now almost all people working somewhere in the IT business have at least heard once about IEC 60027-2 (not necessarily the standard itself but at least the new prefixes). I don't consider myself as one of a "few gifted experts who know more than others". Not at all, I don't even think you have to be much of an expert at all to understand the problem and know the new IEC prefixes. Apparently you consider me an elitist but I believe it's exactly the other way around because my aim is to disambiguate, to straighten things out and make them easier to understand for everyone. The opposition wants to stick to a convention that requires a lot expert knowledge which is full of exceptions because it's not even consistent among all field of computing. --217.87.83.213 (talk) 13:28, 13 April 2008 (UTC)[reply]

Now we’re getting somewhere. Note the wording:

Units of measure: For any given article and target readership*, Wikipedia should use the units of measure and methods of disambiguation commonly used in current literature on that subject.

Indeed; some computer-industry articles do use the IEC prefixes. Apple has a web page (buried somewhere in one of these archives that one of you readers pointed out) that uses them. But that whole Apple page was directed to professional programmers. Perhaps it was dealing with a memory overflow issue that created a security hole; I don’t know. For whatever reason, Apple perceived 1) that absolutely unambiguous information had to be delivered and that 2) the target audience (registered software developers) could readily recognize and understand the units of measure. A Wikipedia article about the mathematics of black hole evaporation necessarily needs and uses different terminology and math symbols than one directed to a more general-interest level. In all cases, Wikipedia’s articles should use the units commonly used for that particular subject for that particular readership. A footnote can be added to the policy statement (location shown with an asterisk above) detailing this common-sense rule of technical writing.

As for your observation “I'm also convinced that by now almost all people working somewhere in the IT business have at least heard once about IEC 60027-2,” let’s not take two steps backwards for each one forwards. No one disagreed with this statement: “The word “mebibyte” (symbol MiB) is not widely recognized by the typical Wikipedia reader.” If the computer industry adopted the IEC prefixes, the general-interest computer magazines would soon follow suit. But they both haven’t and pretty much covers the bulk of Wikipedia’s target audience because most of the computer-related articles on Wikipedia are directed to a general-interest readership.

Now, I was being a tad facetious when I wrote above about how the proponents of the IEC prefixes are behaving as if they are technical writing gods (although there is certainly an element of that god-like attitude when the proponents say “things are broken and we know what’s better”). It is simply not right for contributing authors to put Wikipedia in a position of effectively promoting change by being the only general-interest, English-language publication (among magazines—whether print or on-line—and encyclopedias) to have adopted a new unit of measure. It doesn’t do Wikipedia’s readers any good whatsoever to routinely encounter stuff like “To run Vista, a computer is best equipped with at least 3 GiB of RAM.” Hopefully, the reader would be smart enough to know that “well… that’s just Wikipedia.” But if some guy ran down to Best Buy with his friends to buy a computer and used language like that, it wouldn’t go over well—at least not until he explained that he had read it on Wikipedia. Ohhhhhhhh… *(knowing smiles among his friends)* We’ve got to be more mainstream here. The practices observed by Encyclopedia Britannica and the computer magazines works well enough for them without using the IEC prefixes so we can figure out a way to work without them too. And because the rest of the computer and publishing industries aren’t routinely using the IEC prefixes is why we shouldn’t routinely use them. We can write about them—in articles like Byte for instance; just not routinely use them. We’ve got to go with the flow here. Greg L (talk) 19:00, 13 April 2008 (UTC)[reply]

The manpage is not for "professional programmers" but anyone using the "raidutil" utility, for example, a system administrator. The manpage even explains IEC 60027-2 so it can be readily understood by people who have never heard about it before.

You nicely explained what I referred to as "peer pressure". Does this really happen in reality or is it just your imagination? You could simply call a couple of computer equipment dealers by phone and ask them the price of 1 GiB DDR RAM and see what happens. That's, of course, "original research" but I suggest it anyway, so that you can convince yourself whether it's just imagination or reality. I don't think the fear of causing imaginary traumatized readers is a valid reason to avoid IEC 60027-2.

Also I still think you have it backwards: IEC 60027-2 is mostly of benefit for non-experts. So arguing "only experts know IEC 60027-2" is completely backwards, in my opinion. I would also assume that the average computing article mentions many other terms that the average reader will have to lookup by following the hyperlink - if he cares. This is hardly be avoidable in highly technical articles and with "highly technical" I don't even mean astrophysics or rocket-science. Even if IEC 60027-2 is used in millions of articles, you only have to understand it once. --217.87.100.251 (talk) 20:10, 13 April 2008 (UTC)[reply]

Perhaps what you say is all true, 217.87…. I do however, note that Apple’s Web page, that you say is “not for professional programmers” is titled “Developer Connection” and under that, is this: → [Log In] | [Not a Member]

Greg L (talk) 20:25, 13 April 2008 (UTC)[reply]

Please have a look at manpage or this. --217.87.100.251 (talk) 20:38, 13 April 2008 (UTC)[reply]

The manpage of the cmp utility might also be of interest. --217.87.100.251 (talk) 21:30, 13 April 2008 (UTC)[reply]

In light of what we are discussing here, and from a “big picture”, goal-oriented view, what—in your view—will this prove? Greg L (talk) 21:53, 13 April 2008 (UTC)[reply]

Well, you brought it up. Anyway, examples prove little but these disprove the claim that Apple never uses IEC prefixes or a decimal MB in the software they ship. Other than that it proves nothing but shows that Apple doesn't seem to have a problem with exposing their users to IEC prefixes or a decimal MB. If it's good enough for Apple's customers, I don't think it's far-fetching to expose Wikipedia readers to this. --217.87.100.251 (talk) 22:14, 13 April 2008 (UTC)[reply]

I’ve tried throughout these discussions to avoid using terms like “never” since that’s about as useful as posting a sign on one’s butt saying “FIND A SINGLE EXCEPTION AND KICK HERE.” I try to adhere to terminology such as “the typical Wikipedia reader is unfamiliar with the IEC prefixes and will rarely—if ever—encounter them elsewhere in real-world usage, such as computer advertisements, computer owners manuals, general-interest computer magazines, and professional encyclopedias.” The above policy speaks of using units of measure “commonly used in current literature on that subject”. I think you and I know what that means.
Apple’s Apple’s Developer Connection Web site, is clearly a web site for developers and programmers. How do I know this? Because it’s part of a section called “Apple’s Developer Connection” at http://developer.apple.com/. Even I can figure that one out. Note too that this particular site contains this choice little bit for developers to help them write to a RAID:

The following command would create a volume named "MyVolume", on the RAIDSet named "MyRAIDSet", using one fifth of all available space on that RAIDSet:
# raidutil create volume -n MyVolume -s 20% -r MyRAIDSet

Wow. Sounds like some sorta “Paris-talk” or something. I seem to have missed that one in my owners manual here. Hold on. I’m looking through my 12-page Mac manual for something like that… I think we’re on pretty safe ground here to say that Apple’s Apple’s Developer Connection is for developers. When you register, developers have access to Apple’sXcode Tools, SDKs, and DDKs. Saying that Apple’s Developer Connection is for regular computer users is like putting lipstick on a pig and trying to pass it off as a prom date. It’s not working for me here.

I note your above statement “ Apple doesn't seem to have a problem with exposing their users to IEC prefixes” (my emphasis). It is abundantly clear on the face of it as to the truth of the matter. It is clear that Apple draws a clear line between a “developer” and a “user.” How do we know this? Because on that very same page is this little tidbit to Apple’s develpers that might be of interest:

For our purposes, we will treat all size units as 1024 based numbers. We will allow the user to use the historically prevalent two character notation (KB, MB, GB, TB, and PB) but they will be treated as their more modern counterparts (KiB, MiB, GiB, TiB, and PiB).

So here is well-documented evidence of a computer manufacturer doing the right thing: using the IEC prefixes when they are speaking to an advanced audience that is accustomed to such units; for their “users”, Apple’s OS X, and their advertisements, and their packaging, and their owners manuals, etc., all use “MB” but treat it as a base-2 value. All everyone here on this side of the fence is trying to do is get Wikipedia to consistently adhere to the same philosophy.

Greg L (talk) 00:49, 14 April 2008 (UTC)[reply]

I was talking about this statement in which the editor expressed support based on a misconception. Also Fnagaton keeps claiming that MiB is ambiguous based on a single example he found on the web in a mailing list archive. However I apply logic correctly in so far that I prove the existence of some by examples. That's valid. Fnagaton tries to prove that MiB is ambiguous in general or at least to a large audience. Making this claim based on a single example is not valid. Does anyone disagree?

You, Greg L, don't have to know what a manpage is or what RAID is but documentation does not become "development documention" because it's mirrored on a website for developers. This documentation is for users of Mac OS X not just developers. I don't know what "Paris-talk" is but it seems you're resorting to offensive language again. What exactly do you expect from a "12-page Mac manual"? I have a lot of simple, single-purpose electronic equipment that comes which more documentation than that.

Regarding the raidutil manpage, that's one horrible way to use IEC 60027-2 but that's not the point. The point is, it's used in software shipped by Apple. --217.87.100.251 (talk) 11:00, 14 April 2008 (UTC)[reply]

Would that man page be considered reliable source under Apple's editorial control. It appears to be from the Apple developer's community. The disclaimer at the top of the page says the following.

"This document is a Mac OS X manual page. Manual pages are a command-line technology for providing documentation. You can view these manual pages locally using the man(1) command. These manual pages come from many different sources, and thus, have a variety of writing styles."

I am sure that a few Apple employees believe in kibibytes. But when Steve Jobs introduced the MacBook Air, the press release said it had 2 GB RAM and an 80 GB hard drive. -- SWTPC6800 (talk) 00:57, 14 April 2008 (UTC)[reply]

And still does. Greg L (talk) 01:06, 14 April 2008 (UTC)[reply]

The MacBook Air has 2 GiB RAM. Its hard drive has 80,000,000,000 bytes (and some) storage capacity, not 80 GiB. So it's only half-wrong that is the amount of RAM is underspecified. Steve_Jobs#Early_years doesn't really seem to be a reliable source because it appears he considers US$600 and US$5000 as equivalent. --217.87.100.251 (talk) 10:23, 14 April 2008 (UTC)[reply]

So… Your point seems to be that Apple has no logical high ground here to stand on in their choice of units of measure and how they communicate the quantities of bytes to their customers. But we have all the other computer manufacturers now for you to show why they too have no moral standing in this regard. Whereas I’m sure you can dig up exceptions to the rule with Dell, Hewlett-Packard, Sony, and all the rest, that wouldn’t change the fact that they near-universally and consistently use GB to mean binary math with regard to RAM and decimal math for hard drives. We can all go collectively rap their knuckles with a ruler but that’s the way it’s been done and continues to be done. You guys’ argument *(once again)* boils down to the fact that 1) the current units are ambiguous (I agree); 2) shouldn’t have misappropriated the prefixes from the SI to denote binary math in the first place (I agree); 3) the IEC prefixes come from a wholesome, wonderful, standards organization whose members eat bran each morning with their breakfast (fine); and 4) even though the typical Wikipedia reader doesn’t recognize the IEC prefixes (everyone agrees to that one); which itself is because 5) the manufacturers of RAM, hard drives, and computers near-universally don't use the IEC prefixes for communicating to the average Joe; which is the reason 6) no general-interest computer magazine (print or on-line where they could use hyperlinks) uses the IEC prefixes and neither does any, proper, professional encyclopedia; 7) none of that matters because you guys know better and—with the assistance of User:Sarenne (blocked for life)—have put Wikipedia in the lead position of effectively promoting a change in the very language our readers should speak—but which they won’t likely encounter anywhere else. In your view, this is good, sound practices for effective technical writing. A majority of editors on Wikipedia disagree with you on that last point. Greg L (talk) 17:00, 14 April 2008 (UTC)[reply]

Yes, we seem to agree on almost every point except the conclusion. Though, I'd like to ask you why you have to insert insulting remarks in about every longer comment you write (examples: "eat bran each morning", "none of that matters because you guys know better"). I mean what exactly is the purpose of that? I really don't care as much that I'd run away crying. Actually you're making it hard for others to support you and your position excluding those who are also keen on making offensive remarks. Let me also get back to what you wrote previously about getting laughed at if you used these new IEC prefixes in real life. You haven't provided any evidence for this. What's even more interesting is that Wikipedia contains information about China that would get Chinese people into jail if they dared to talk about that in public in China. So do you really care about imaginary ignorant people laughing about Wikipedia readers but you don't care about Chinese being thrown to jail? I'm not saying this information about China should be removed. I'd just like to show you how silly this worry is. --217.87.116.100 (talk) 18:43, 14 April 2008 (UTC)[reply]

The example of the minority of "Apple software" using the new IEC prefixes is because some of these modules are written by individual programmers who have changed the software and it is not because of any published policy from Apple itself. The majority of software Apple produces does not use the IEC prefixes. Fnag aton 20:39, 14 April 2008 (UTC)[reply]

Puuuuhleeze 217.87, now you’re just being silly. Little about what you wrote makes any sense to me. My being facetious about the ‘wholesome goodness’ of the IEC by alluding to how they “eat bran” in the morning wasn’t intended to be insulting to you or anyone else here on Wikipedia. Any reasonable interpretation of what I wrote is clearly this point: It doesn’t matter how meritorious the proposal from the IEC is or how you think they’re big shots. Official policy of the BIPM that is part of the SI is sometimes ignored by the rest of the world (such as leaving a space between a numeric value and the % symbol); and they’re the biggest kahunas on the block when it comes to standards. If the rest of the world doesn’t use the space in percentages, then Wikipedia shouldn’t either (and doesn’t, even though it’s “official”). I think you tend to seize upon perceived slights as a diversion rather than focus on the issue at hand: the merits of the proper methods of communicating via technical writing.
Next, having the courage to speak the truth on Wikipedia regarding a country that has no guarantees of freedom of speech and your transparent attempt to conjure images of brave students standing in front of army tanks doesn’t have the slightest relevance here. We’re talking about how best to not confuse readers by avoiding the routine use of units of measure that haven’t found traction in the real world and are unfamiliar to most readers. If you have to resort to the tactic of attempting to seize the moral high ground by linking this discussion to brave Chinese disidents who are thrown in jail for speaking the truth, you have apparently run out of anything valuable to say. I do wish you wouldn’t waste anyone’s time here and would try to find something more relevant to say.
Finally, as for my comment of how you guys seem to have an attitude of “we want Wikipedia to do it differently from the rest of the world because we know better,” that is not intended as an insult. That is a legitimate, truthful way of looking at the logical consequences of what that position amounts to. Ergo, the poll above (“As a matter of fact, yes, I am more enlightened than all the editors at all the general-interest computer magazines and professional print encyclopedias.”) That is a way of helping certain editors here to see that this is a logical consequence of embracing their position on this matter. My suggesting that your position amounts to an attitude of “because you guys somehow know better than the pros” is not a prohibited personal attack; it is a perfectly legitimate tactic of debate taught in high-schools and college debate classes. If you don’t want to debate me, don’t debate me, but please don’t tell me how I may think or express my thoughts; if you can’t take the heat, get out of the kitchen. Greg L (talk) 21:00, 14 April 2008 (UTC)[reply]

This is neither a debate class nor political chit-chat. You won't gain anything from such "perfectly legitimate tactics" here. --217.87.116.100 (talk) 21:25, 14 April 2008 (UTC)[reply]

"You won't gain anything" - Well actually that isn't true, he does gain the advantage of showing how bad your argument is since it uses fallacious points. Fnag aton 21:32, 14 April 2008 (UTC)[reply]

217.87: I agree; this is not a debate class nor is it political chit-chat (so it is most unfortunate that you tried to somehow draw a connection between the efforts here and the ongoings in China). However—and you may not like it—this is debate. Logic talks and B.S. walks. Greg L (talk) 21:48, 14 April 2008 (UTC)[reply]

^ http://www.inwap.com/tivo/disk-sizes.txt

[1] ttp://www.inwap.com/tivo/disk-sizes.txt

[1]