Wikipedia talk:Manual of Style (dates and numbers)/Archive B3

Two quick notes

Hello. I just want to mention two quick things.
First, just a reminder, when a machine had some number of 'K' of memory (eg. 4K, 5K, etc), it wasn't necessarily kilobytes. As such, it wasn't necessarily KiB. Not all machines express their memory in bytes. Some expressed in 'words', and some even used peculiar word-size. (Heck, in really old hardware, not everything even used bytes at all!) Changing K -> KiB is not allowed unless you have direct knowledge that they definitely meant 1K=1024 bytes. That is, not just 1024, because that could mean 1024 words, but specifically 1024 bytes. Just wanted to make sure that was understood.
=
Second, there's clearly a lot of people who prefer to use the de facto standard for prefixes than an artificial standard that isn't accepted by professionals or in common use. The fact that it's only a couple editors at a time occasionally drifting in doesn't matter. A small handful of editors who keep this page on their watchlist shouldn't be allowed to override a scattered majority. (That is, if three people come in one month, and three another, and three another, then they shouldn't be overridden by four very persistent editors who never leave this page.)
I think it's time for another official vote. I know, votes are typically discouraged, but this guideline has been outright abused.
It's been used to hurt articles by giving false/misleading information. It's been presented as the only exception to IAR. It's been used as the only example I can think of where the "common usage" of the first major contributer takes a back-seat over someone else's personal preference.
Fact is, it's been abused, greatly. If Sarenne hadn't taken on such a crusade (and engaged in wikilawyering, for that matter), then you might've been able to work things in eventually, but unfortunately, you've attracted the attention of too many editors, so the time has come.

If it doesn't go by a vote, then it'll have to be the DR process, and I don't think anyone really wants that, do they?
For reference, canvassing is just fine if you're specifically canvassing for people who've already expressed their opinions on this subject. I've already commented on this MOS page, so it was very much acceptable to refer me to this page.
And that brings us to a vote. We need one. I'm sorry, but we need one. And we need to establish the rules ahead of time. I'd suggest:

  1. A vote of one week.
  2. Canvassing is acceptable only to users who have already specifically discussed the KiB vs. KB issue. (Regardless of which side they chose)
  3. To prevent excessive arguing, single-line summaries for your "vote".
  4. No votes of "per nom", or "per <insert-name-here>". Keep it short, but still provide a reason.

If you don't believe we should have a vote, then please provide an alternate method of resolving this dispute. Because the constant arguments are getting tiring.
Out of respect for the growing lack of consensus, it would be appropriate if all involved editors were to refrain from continuing the changes. That is, if you know that more and more people disagree with you, you really shouldn't be making any more changes to KiB. Bladestorm 14:54, 10 April 2007 (UTC)

I don't agree with several of the statements you've made, but let's not argue unnecessarily. What exactly do you think a vote would accomplish? Say we hold a vote and the policy is upheld. Do you think that would put an end to this ongoing debate? I certainly don't. Say it isn't upheld. Then we're stuck with trying to figure out how to consistently deal with very inconsistent usage of these terms. I can't fathom how the floppy disk article would be sane without using binary prefixes; you'd have to explicitly interpret nearly every quantity parenthetically.
I didn't originally have a problem with a straw poll, but I'm sorry, I do not consider canvassing acceptable. On a small scale it's vote stacking and on a large scale it's spam. I don't see a vote as necessary at this point because none of the arguments have changed. The only thing that has changed is the influx of several new editors who object to the prefixes on personal grounds. When I see people proposing to "find new voters" and bring in people from outside Wikipedia, I lose a lot of faith in the validity and utility of any such straw poll. Additionally, I haven't seen any reasonable proposal for dealing with ambiguity without the binary prefixes, and I don't consider deciding on a per-article basis acceptable because it would allow this horrible debate to be re-hashed on hundreds of computing-related talk pages. When someone proposes a decent alternative for dealing with the byte multiplier ambiguity, then I might be keener on the notion of a vote. -- mattb @ 2007-04-10T15:36Z
P.S. - I agree with your last statement, but in fairness don't you think that editors should refrain from removing binary prefixes as well? Edit warring over a disputed guideline isn't ever acceptable behavior, but it doesn't apply to just one side of this. -- mattb @ 2007-04-10T15:40Z
All changes (regardless of the style used now, which is the most fair solution for everyone) that gave style based reasons (any that quoted the MoS or binary prefixes) should defer to the original style of the first major contributor, which is what the root MoS guideline says.Fnagaton 15:58, 10 April 2007 (UTC)
Well now, hold on. There's currently a very significant dispute going on, and edit-warring. Once edit-warring starts, the current version is what it stays at until it's resolved. That is, anything currently using (ugh) KiB must stay, and anything currently using KB also must stay. So please don't revert anything back to KB until this is resolved, and I hope Sarenne will similarly refrain from changing anything to KiB until it's resolved. Bladestorm 16:04, 10 April 2007 (UTC)
You're right and I can live with waiting until this is properly resolved. Fnagaton 17:20, 10 April 2007 (UTC)
That's a gross misinterpretation of what it says. Go read the paragraph again. It says, FIRST you use the Wikipedia style, THEN if still not resolved, you use the style of the first major contributor. --SLi 16:05, 10 April 2007 (UTC)
In the articles I'm thinking of the MOSNUM style was tried (using style as the reason) then there was conflict (sometimes on a talk page) that was not resolved. The next step should be to defer to the style of the first major contributor because MOSNUM is all about style, but that has not happened in some cases. What has happened is that the editor using binary prefixes has hidden behind MOSNUM again, making it a style issue but refusing to defer. It's a circular argument to repeatedly do that. Fnagaton 17:20, 10 April 2007 (UTC)
Well, currently, the current (weak) phrasing of the MOS is being used as a weapon. It's being used for wikilawyering (I really hope that isn't one of the parts you disagree with). Fact is, sweeping (sloppy) changes are being made.
Canvassing is appropriate if you choose the right people. For example, would you concede that a handful of pro-IEC editors tend to monitor this page pretty religiously? And that you have to constantly address different editors who want to change it back to KB/MB? Answer honestly, it happens all the time, doesn't it? Why should the fact that a minority of IEC-followers should appear to have a consensus that they don't? I'm not saying it's fair to just pick random people. However, if people have already expressed their opinion (for or against the IEC convention), then it's fair to notify them that their opinion is desired again. It isn't spam if it's in direct response to their previously expressing their views. (And I would expect you to notify at least a few pro-IEC editors as well, because I don't want to have to keep going over this either!)
What would a vote accomplish? Well, currently, a few editors are claiming consensus (or overhwelming majority). I want it to be directly known, and indisputable, just how many people actually agree on this subject. Fact is, if you aren't willing to tally it up, you shouldn't claim to have consensus on your side.
And, also, even if you don't get the results you want, votes do tend to bring up alternate ideas. (I've learned this from a coupld of AFD's, CFD's)
And, if nothing else, what other solution is there? There are a lot of editors who disagree with skimming all the articles, making changes. (Especially the sloppy/incorrect ones, but even when they're technically accurate) So, what's your solution? And, for that matter, what's your solution to the sole IAR-exception? What's your solution to the first-major-contributer-wins problem? What's your solution to hiding behind MOSnum instead of discussing? In short, how do you expect to resolve this dispute if not to ask people what they think?
I contest the claims of "sloppy/incorrect" changes made using this guideline as a justification. Thus far I have only seen conflicts arise when people simply don't like the prefixes.
You still have not addressed how a vote now will prevent further dispute on this in the future. It can't, so I don't support a vote simply on the grounds that it will end this debate. If enough people want to have the poll to see if consensus still upholds the guideline, fine, I'm not going to try and "block" such a thing. I'm only pointing out that I don't think any amount of votes will ever stop the disputes over this.
I'm not the one making mass changes, but as I pointed out, most editors do NOT object to people who copy-edit lots of articles. I agree that Sarenne has hid behind the MOS a bit, but I understand the justification. What good does it do to debate this on every single talk page where an editor dislikes the prefixes when the debate has raged several times here? I do not think that all of Sarenne's actions have been wise or appropriate, but it's a straw man to use the actions of one editor to try and remove a guideline that has a lot of reasoning behind it.
Anyway, as I said, I'm not going to attempt to block a vote, but I go on record as saying that it won't resolve this dispute regardless of the outcome. I still don't agree with your point of view as regards canvassing, but if editors are going to engage in it, they will have to contact all fifty or so editors who have participated in polls/discussions previously in order to avoid vote-stacking. -- mattb @ 2007-04-10T16:03Z
I do not believe that "a handful of pro-IEC editors tend to monitor this page pretty religiously". Not at all. On the other side, as I stated above, I do believe that the actions by at least Sarenne and me have brought probably a lot of opponents of the current wording to this talk page, but no supporters of the current wording. That's not a good basis for a poll that tries to gauge general editor opinion on the issue, and so I strongly oppose voting on the issue at least for now. I also have trouble with the canvassing done, although I think if we allow pretty unlimited spamming of people who have expressed an opinion of any kind on the issue, we gain a lot more supporters for the current wording / IEC prefixes than opponents of the IEC prefixes. --SLi 16:10, 10 April 2007 (UTC)
I have to agree with Bladestorm when he wrote "a handful of pro-IEC editors tend to monitor this page pretty religiously" as it certainly seems to be the case. Now is an excellent time to conduct a poll since the guideline has been tested in the wild, warts and all. Fnagaton 16:51, 10 April 2007 (UTC)
Without any concern to the (to me quite obvious) thing that making such controversial edits will attract people here quite onesidedly? Or is it simply not important that the poll is a representative sample of the editors in Wikipedia? --SLi 14:41, 11 April 2007 (UTC)
An example of a "sloppy/incorrect" change [1] On the megabit page the link to megabyte was changed to mebibyte. The edit summary is "Megabit (mebibytes where appropriate (see Manual of Style (dates and numbers)))" -- SWTPC6800 19:20, 10 April 2007 (UTC)
Which part of that edit is sloppy/incorrect? Are video game cartridges measured in decimal megabits? — Omegatron 21:21, 10 April 2007 (UTC)
First off, an article on "megabits" should link to the article for "megabytes". Changing that 'megabyte' to 'mebibyte' had no justifiable purpose. What's more, he directly equated 125,000 bytes with 122 KiB. That's incorrect. In the edit summary, he acknowledges (on his second attempt at the change) that it's approximate. However, that sort of edit is explicitly prohibited even by the current MOS. (Note where they tell you not to convert hard drive sizes to GiB.) Incorrectly converting bytes to kibibytes for no reason other than to include KiB is absolutely unacceptable. As part of the edit war truce, I won't be reverting other changes, but there's no way that one could stay, because it was outright wrong (as in, factually inaccurate). Bladestorm 21:26, 10 April 2007 (UTC)
I am loath to point out that 125000/210 = 122.07. There's absolutely nothing factually incorrect in that edit. In actuality, the original "122 kilobyte" phrasing was in ITSELF an approximation (using the 210 definition of "kilobyte"), which you so strongly objected to. Intgr, who initially reverted SLi's edit, later reverted your chastising edit after realizing that there was nothing wrong with SLi's version. [2] Video game cartridges are not hard disks, either, so I am failing to see your point. Allow me to humbly suggest that perhaps what you have perceived as a harmful edit is merely a misunderstanding. -- mattb @ 2007-04-10T23:08Z

(outdent)Rargh. I'll concede that I may have been brash (is brash a word?) with that one, though I still maintain that outright removing a link to megabyte from an article on megabit is... peculiar? Bladestorm 23:48, 10 April 2007 (UTC)

Sarenne's "crusade" has been encouraged by one of the "small handful of editors" who support the new binary prefixes. Seraphimblade gave Sarenne a Barnstar award "For tireless contributions at the thankless (often very thankless) job of converting articles to use binary prefixes where appropriate, I award Sarenne the Working Man's Barnstar." -- SWTPC6800 20:54, 10 April 2007 (UTC)
You're going to hold one user's gesture of good will towards another against them? -- mattb @ 2007-04-10T23:13Z
When Sjenkins7000 and Sarenne were in an edit war on the MacBook and such. Seraphimblade intervened with an official "Request for clarification." This would have been great except Seraphimblade misrepresented the results to favor Sarenne. User_talk:Sjenkins7000 He said that the MOS guideline was consensus and Sjenkins7000 must follow it. I read the response as the opposite. -- SWTPC6800 03:19, 11 April 2007 (UTC)
I agree that linking to megabyte from megabit is the correct thing to do and am sorry for making a mistake there. However if you view my edit history, you see that I have copyedited hundreds of articles, some for more trivial things like adding a space between a number and a unit, removing PBUH or "Peace Be Upon Him" outside quotes from articles on Islam, fixing typos, decapitalizing words that should not be capitalized, and so on, and some for things that require some domain expertise, like the IEC prefix issue. I'm not perfect. Don't you think showcasing the occasional mistake as evidence of malice is a bit harsh? --SLi 14:16, 11 April 2007 (UTC)
First, just a reminder, when a machine had some number of 'K' of memory (eg. 4K, 5K, etc), it wasn't necessarily kilobytes. As such, it wasn't necessarily KiB. Not all machines express their memory in bytes. Some expressed in 'words', and some even used peculiar word-size. (Heck, in really old hardware, not everything even used bytes at all!) Changing K -> KiB is not allowed unless you have direct knowledge that they definitely meant 1K=1024 bytes. That is, not just 1024, because that could mean 1024 words, but specifically 1024 bytes. Just wanted to make sure that was understood. (said by User:Bladestorm, above)
If you mean that we need to find a source for every single instance where the memory size is given as 64K (especially in old specifications), I strongly disagree. You probably know that in 99.9 % (ok, I concede it's only 99 % --SLi) of the cases it is true that it means 64 KiB, and if we claim it is 64 kB (or equivalently 64K), that is simply incorrect if we say kilo=1000. The claim that it means 64000 bytes is in my view the extraordinary claim that requires evidence, not the other way round. --SLi 14:29, 11 April 2007 (UTC)
However, if there's any reasonable question as to whether 210 or 103 was meant, it shouldn't be changed until a source can be found that clarifies things (such as what happened with the flash memory capacity on the Wii article). You're right that there should almost never be a question as regards modern microcomputers (late 1970s onward), but earlier than that you need to be particularly careful. However, I haven't yet seen anyone make an egregious technical error (ignoring stylistic quibbles) in changing the unit prefixes in articles about older computers, so I really can't accept the statement that this guideline has directly encouraged "harmful" edits. It has caused conflicts between editors, certainly, and the resulting edit wars are indeed harmful, but as far as I can see there is no reason to discredit the merits of the guideline on an accuracy basis. Note that I'm not including the consistency with sources and common usage arguments, which are separate and valid issues. -- mattb @ 2007-04-11T14:39Z
I can understand and accept as very valid the issue about common usage, but the consistency with sources argument I just cannot understand. We can just as well argue that when we are talking about 2000 year old Greek writings, the articles themselves should be in Greek for consistency. If we always use the style of the publication we refer to, it's going to be a truly horrible mess. --SLi 14:46, 11 April 2007 (UTC)
That does not logically follow because in the matter of Greek writing there are reliable sources that have done the translation from Greek to English, for example. It is therefore different to this case because the sources are using the terms and those terms are still in current use. Fnagaton 13:03, 12 April 2007 (UTC)
Sorry, but that's just ridiculous. Have you ever translated text yourself? I have and I can assure you that nobody can make a perfect translation from <insert nearly any language here> to <insert nearly any other language here>. On the other side, figuring out what a source means by kilobytes is nearly always trivial. --SLi 13:38, 12 April 2007 (UTC)
Actually, when they covered hardware history here (uh, comp sci at brock university, if it matters), we looked at some screwy hardware. If you like, I'll walk up to the 'museum' (glass case full of old computing crap) and see if I can find a couple good examples. I'm just saying that if you're going to change K to KiB (or to KB, for that matter), then you need to personally confirm that it's in bytes. And then you need to personally confirm that it really was 1024. Obviously, for more recent stuff (I'd prefer to say as recently as the 80's, than anything since 1970), this isn't an issue. But for history-of-computing stuff, you really need to do the extra work. And the wii article changes did hurt the article, though I think we can solve some of those problems in a bit. (Somebody remind me if I forget, but, before doing a vote, I'd like for us to outline just what we hope to get from it) (And okay okay okay, so I was unfair to your edit there. Shaddup shaddup shaddup!) :)Bladestorm 14:45, 11 April 2007 (UTC)
:) Yeah, I know about all the creepy 36-bit computer stuff and things like that. That's why I consider myself qualified to do this kind of copy editing on these issues. BTW flash memory really is nearly always 2^n bytes, AFAIK. --SLi 14:52, 11 April 2007 (UTC)


Binary prefixes straw poll

Okay, so if we do end up running a straw poll, I'd like to make sure we can come to terms on a couple of things:

  1. Both parties have to agree with the other's phrasing when they make their case. The summaries should get to the point without a lot of rhetoric.
  2. If either party wants to inform other users that this straw poll is going on, they must inform both those who may support and those may oppose their position. No vote stacking.
  3. We do not count votes from new accounts with no significant editing history. Nobody wants the results of a poll to be questioned due to perceived sockpuppetry.
  4. Voters are asked to keep their comments with their vote to one or two sentences. Once we have all agreed on the summaries for the vote, there should be no arguing in the polling process.

Let me know if I've missed anything important. Here is how I would summarize both parties' points of view. If you have something to add or remove, let's discuss it:

Keep the current guideline. The IEC binary prefixes allow articles to express binary capacities unambiguously. Common usage of prefixes like "kilo-" in the computing context allows for several conflicting definitions. Recommending the usage of the binary prefixes in contexts where binary power multipliers are implied shifts the interpretation of the exact quantity intended from the reader to the presumably more knowledgable editor. The current guideline intends to keep unit convention as consistent as possible; using the SI definitions of "kilo", "mega", etc, and using the binary prefixes where powers of two are needed. The guideline reflects the usage recommended by several relevant organizations: BIPM, IEEE, IEC, NIST, etc.
Remove the guideline. The binary prefixes are not commonly used by computer professionals or by any major computing standards organization. The de facto standard is for "kilobyte" and "megabyte" to denote binary powers of bytes except in certain contexts. The binary prefixes are neologisms that may confuse readers who have had no prior exposure to them. What's more, using these prefixes usually renders article text inconsistent with reliable sources, further compounding possible confusion. Wikipedia should use terms like "megabyte" as they are commonly defined and note whenever they take on a different meaning (as in the case of hard drives, floppy disks, and optical media).

If someone wants to add a voting option somewhere in between the two extremes, feel free to propose a wording. I think that adding a note that requires editors to provide verification of the capacity values before changing common usage to IEC prefixes might appease some people. -- mattb @ 2007-04-11T15:06Z

For the record, I'm perfectly fine with the MoS changing, if someone can provide a good reason why it should.
More votes and the stacking, polarization, and other bias that comes with them is not a real way to make that kind of decision. The idea behind our policies is that editors should come to an agreement and convince each other of the merits of their viewpoints, not to unwaveringly hold onto an idea and crush other's legitimate opinions under the treads of our majority rule votes. We shouldn't be trying to "win", but to understand where each other is coming from and judge which idea is best for writing an authoritative reference work.
If you can convince all of us that there's a good reason why this should be changed, we'll concede, but so far no one's being convinced. When you can't convince anyone that you're right, maybe you should reconsider. — Omegatron 15:33, 11 April 2007 (UTC)
In principle I agree with you, but what happens when you hit a point where there are irreconcilable opinions on a matter? I see the validity of both sides of this argument, I'm firmly supporting one option, but I can see how someone would reasonably prefer the other. Unfortunately this is a matter of consistency, so I don't think we can have it both ways. -- mattb @ 2007-04-11T15:38Z

Just a few concerns that I might like to be included, if possible:

  1. Wikipedia should be "descriptive", not "prescriptive". You can see evidence to this effect all over the site (for naming conventions, etc). We're expected to refer to things as they "are used", not as they "should be used". This is why gender-neutral pronouns aren't used in articles.
  2. In spite of that, I don't think anyone's saying you can't use binary prefixes. For example, if you have an article that you know mixes ^2 and ^10 a lot, then it could be really useful. I'd like to see it given equal treatment to any other stylistic issue (Colour vs color, etc. Except for where the distinction is of particular value to that specific article.).
  3. MiB and MB should not both be used in a single article unless every single case of MiB is known to be binary, and every single case of MB is known to be decimal. Using MiB for the definitively binary, and MB for both decimal and unknowns incorrectly gives the unknowns the appearance of being decimal. And, frankly, I was hoping we could all agree to this point regardless of the results of the vote.
  4. I don't think anybody is obligated to notify both sides. I do want both sides to know about it. However, anyone who already has someone in mind, who's already expressed an opinion on the subject, shouldn't then be forced to try to scrape up a possible opposing opinion. That is, everybody probably knows one or two people who already commented on it. However, I'd like to see a prohibition against inviting anyone who isn't already on record as having an interest in the topic. (Because otherwise, it would be spam) Bladestorm 16:35, 11 April 2007 (UTC)
Actually, I was just thinking, maybe there's a compromise? Post notifications of the poll on the talk pages of articles where conflicts have arisen, and maybe also a quick notice on the village pump? Bladestorm 17:15, 11 April 2007 (UTC)
Regarding #1, there are a significant number of editors who do use the IEC prefixes appropriately in their articles, and a number who do not. Without trite attempts at quantification, I think there are enough people who use the IEC prefixes in practice on Wikipedia to make it unnecessary to claim the guideline as phrased is too prescriptive. Other parts of the MOS work similarly; we strongly encourage placing a space between value and unit (per SI convention), but a significant number of editors don't do it.
Heh. Actually, I meant outside of wikipedia. As in, wikipedia shouldn't be trying any trailblazing for convention. Do a quick google search for 'wikipedia' and 'prescriptive', and you'll see just how often people agree that, no, we shouldn't tell readers what units we'd like them to use. :) Bladestorm 17:46, 11 April 2007 (UTC)
This point is covered under the common usage argument I presented. There's no need to restate it in a much more general form, it's just wordy. -- mattb @ 2007-04-11T18:09Z
I see this as a "do or don't" issue simply because it makes no sense to me to use binary prefixes in some articles and not others. That leaves a lot of opening for debate merely on grounds of editor stylistic preference and flies in the face of consistency. The guideline was initially enacted precisely to avoid this, and there's no point in even having a guideline on this matter if it doesn't make a strong recommendation.
Your third point is just beating this to death. The guideline already advises people not to change quantities that they aren't sure about, and Omegatron just added even stronger language. However, I don't object to adding another sentence to the guideline which makes it explicit that articles should only use SI prefixes as defined by BIPM if the article also uses binary prefixes. I have added a statement to the current guideline which explicitly advises against using "IEC binary" and "SI binary" notations in the same article.
I believe you misinterpreted. My concern isn't with using both prefixes for the binary sense. The problem is when an article initially uses decimal, and an editor changes the known binary values to IEC, and leaves the unknowns as decimal.
(Incidentally, your edit doesn't really make it any clearer, and you should probably link BIPM as well)
Take the Wii article. Some values were known to be binary. Some were unknown. The known values were changed to MiB, and the internal flash was left as MB, which turned out to be incorrect. However, this implied that the MB's were decimal. But, your solution is a bit backwards. The problem isn't with using MB only in the decimal sense. The problem is with mixing units when you don't know what all of the units are. That is, I should hope that IEC would only be used when all values are of known unit-type (decimal or binary). Bladestorm 17:46, 11 April 2007 (UTC)
Again, I think you're beating this to death and asking for instruction creep. The problem on the Wii page was quickly identified and fixed, so I don't think we need a bunch of extra rhetoric to describe one situation which most everybody already acknowledges. -- mattb @ 2007-04-11T18:09Z
I don't believe it's appropriate for someone to look at a list of users who have ever commented on this matter, pick out the ones who support one viewpoint, and notify them. That's vote stacking no matter who does it, and that's the kind of thing which will ruin the results of a straw poll. What happens if we tolerate advertising this poll to individual partisan users? Undoubtedly editors in the other camp will notice and feel obliged to do the same in order to compensate. Where does it end? When we've spammed a message to the talk page of each one of the roughly 50 editors who have ever remarked on this? "Having a few people in mind" is all well and good, but making a list of editors sorted by presumed vote on this matter sort of trumps that notion.
I would like to avoid vote stacking and campaigning and spammy behavior altogether. We can advertise this straw poll on the talk pages of an appropriate Wikiproject or two, and perhaps on the village pump as well. That will give it plenty of exposure without there being any possibility for vote stacking. -- mattb @ 2007-04-11T17:26Z
I never suggested that anyone should come up with a list to vote-stack. Pro-decies aren't ghouls, ya know. :) What do you think of just including a mention on the village pump, and on article talk-pages where there's been conflict? Eg. On the C64 page? Bladestorm 17:46, 11 April 2007 (UTC)
Including it on the talk pages where conflicts have arisen due to Sarenne or SLi's actions will only bring in editors who oppose the guideline. Let's just stick with the Wikiprojects and the village pump. -- mattb @ 2007-04-11T18:09Z
I can definitely sympathize with that position. On the other hand, though, for all the cases where MOS, rather than actual discussion, was used as a justification, I do think it's fair notice to let them know that MOS might change. I realize the edit warring might bring in a few disgruntled editors, but on the other hand, when people were told that consensus was already achieved, and that MOS was against them, they do have the right to know that that simply might not be so. Bladestorm 18:20, 11 April 2007 (UTC)
I agree with you here Bladestorm. Where an article has been changed to use binary prefixes and the MoS has been cited as the reason for that change then it is definitely of interest to the regular editors of that article that this MoS issue is being discussed and voted on. So to put a message on those relevant talk pages isn't spamming because it is informed reasonable communication. Fnagaton 11:15, 12 April 2007 (UTC)
I don't think this is about any "right to know", this is about trying to get a poll of a descriptive sample of wikipedians, and doing things which only attract one side of the issue is rigging it and absolutely unfair. I have argued that even the current setting, doing the poll after a campaign of changing these, creates an unfair advantage to one side. We don't want a bureaucratic flip-flop where next time some disgruntled people think the change was wrong they come here en masse as a result of similar circumstances and vote again to reinstate the language, do we? --SLi 19:12, 11 April 2007 (UTC)
You, um, might want to re-read what you just wrote. :) A poll after a campaign for change is wrong? How do you make changes to policy if you aren't allowed to ask for change when people have problems with it? As far as the 'right to know' is concerned, the fact is, the MOS and alleged consensus was used to trump people's concerns. Anyone who got trumped by a policy that's actually somewhat debated now should have a chance to have their say. (As far as things swinging the other way is concerned, if you can think of anyone that you think has a very serious concern with your side, and who you think wouldn't necessarily have already seen this page, I don't really see a problem with you telling them)
I guess what it all boils down to is... Yes, I can see the distinct problem posed by ballot-stuffing, etc. On the other hand, this is a policy that'll affect all of wikipedia, and I have to believe that the greater the number of people that know about it, the more legitimate it is. As it is, how many are currently actively involved in this debate here? Me, mattb, you, sarenne, 6800... anyone else? Obviously we need more than 5 people to establish new policy for the entire site. I know I'm probably naive, but I just think that anyone with an opinion should be invited to share it. Bladestorm 19:24, 11 April 2007 (UTC)
Well, no. You make the mistake of thinking about this as a vote instead of a poll. When we have a vote, we inform _everyone_ who is eligible to vote. When we have a poll, we have to try very hard to get a representative sample of the entire eligible voter population, which simply fails if you do things that introduce one-sided bias. --SLi 12:32, 12 April 2007 (UTC)
I do not think we need to make a binary choice on binary prefixes. The two choices were ban all MB or ban all MiB.
All of the historical sources and all of the current technical hardware sources use MB. Forcing XiB to eliminate the issue of 1000 vs. 1024 creates a conflict with citations. Wikipedia can not solve this and must live with both nomenclatures. Also KB and KiB are not the only choices. A major design feature/problem of the IBM PC was the "640K" memory limit. A historical discussion of that topic does not depend on the exact value of the memory limit. An audio tape of Bill Gates was just found and he does not say 640 kibibytes.
Omegatron was one of the primary authors of the new binary prefix guideline in July 2005 and has vigorously defended it. In the last three months 20 people have questioned the wisdom of this inflexible style guideline. When Seraphimblade did a "request for clarification" the answer was it is questionable if this MOS still had consensus. Omegatron now says "When you can't convince anyone that you're right, maybe you should reconsider." Maybe he should. -- SWTPC6800 23:09, 11 April 2007 (UTC)
You're making very little sense. Have you totally ignored the fact that there are plenty of contexts in which using "MB" to mean 106 bytes is appropriate? If you think this is about banning one usage or the other, you're severely mistaken. What's more, I don't want to debate this further here, just try to find common grounds on a straw poll. If you want to keep arguing, please go to the above section. I'd like to keep this one mostly to discussing what the straw poll should be deciding. -- mattb @ 2007-04-11T18:09Z


I think this should be a straw poll option.
Modify the current guideline. Remove this sentence. "If a contributor changes an article's usage from kilo- etc. to kibi- etc. where the units are in fact binary, that change should be accepted." The guideline will follow WP:MOS#Disputes_over_style_issues -- SWTPC6800 18:37, 11 April 2007 (UTC)
While although that would solve most of the actual problems, and make it fall more in line with general procedures, I doubt you'll get much support on that one. :) Bladestorm 19:24, 11 April 2007 (UTC)
I don't think that would solve anything; it still would say that the right thing is to use IEC prefixes. You need to add additional language explaining that such changes would in (some, specify in what) circumstances be unacceptable. --SLi 19:24, 11 April 2007 (UTC)
Not trying to push you on anything; just wouldn't mind knowing your opinion, SLi. What about something like:
  1. When an article needs to use both decimal and binary values, always use IEC for the binary values (for the sake of making it easier to read).
  2. When the choice of notation is disputed, always resort to the first major contributer (to make it conform to all other style issues).
  3. For all cases of both quotes and describing units used by specific sources, always use the same units they do. (I can explain this one if it isn't clear; but otherwise I'll try to keep things short)
  4. All changes from one unit to the other should include a friendly note on the article's talk page.
Again, not trying to push that on you. I just want to know what you think. :) Bladestorm 19:32, 11 April 2007 (UTC)
No problem, glad you asked :) I do not support different usages on different articles of the same encyclopedia, that would make things very confusing. Also I am entirely unable to see why it is relevant to Wikipedia, which is an entirely different work from the sources, what style they have decided to use. Internal consistency, on the contrary, is important. I would actually prefer using MB _always_ in the 2^20 bytes meaning and fixing articles where it's used in 10^6 bytes or 1000*1024 bytes meaning to your solution. --SLi 20:06, 11 April 2007 (UTC)
Very well. I tried. :) (btw, I'm just curious. What do you think of the current policy of spelling it "color" in distinctly US-based articles, "colour" in distinctly commonwealth-based articles, and a mishmosh of the two for everything else?) Bladestorm 20:17, 11 April 2007 (UTC)
I think there should be consistency inside an article for Br/Am spelling. In the ideal case Wikipedia would be written entirely in either style, but that would be quite a major dispute (I view this IEC thing as a small one). Also it is rather different from this issue, this isn't a pure matter of style because the SI prefixes are being used in at least three different ways. I mostly just long for consistency and accuracy, always using MB only for 2^20 would be some kind of a solution (but I don't think it's feasible, nobody wants to explain to every new contributor that you can't say 200 GB hard drive when the manufacturer says 200 GB hard drive). Inconsistency with how the SI prefixes are used in other cases is a smaller, but still very real issue. I just view the IEC prefixes the most obvious and easiest way out of this mess without sacrificing (internal, which I consider more important than with sources) consistency and accuracy.
Not being a native speaker myself, I find myself using mostly American spelling, but I do use some britishisms. I don't consider either style wrong. Consistency between articles even in that issue would be nice, but in that case we just don't have a way out like we do have the IEC prefixes (short of writing some automatic Br<->Am converter or stuff like that, which I think would be doable, but I'd rather leave that to the time of actually publishing Wikipedia, and only when Wikipedia is otherwise perfect ;). --SLi 22:38, 11 April 2007 (UTC)

(lose indentation) Matt, what do you consider a "new account with no significant editing history"? I agree that that is necessary, but I think we need to formalize that somehow. Should it be based on the time of first edit, number of (unrelated to this issue? non-editwarring?) edits (before some date?), or a mixture of the two? --SLi 19:28, 11 April 2007 (UTC)

I'd rather not put a fine point on it. It's usually very obvious when sock puppetry is going on. For example, if an account's first edit is to vote, that kind of thing should be ignored. I hope we won't see any of this, however. -- mattb @ 2007-04-11T22:36Z
Sounds good to me. I actually like that a lot more than what we/they use in Finnish Wikipedia, which IIRC has settled (but not formalized) to a history of 100 edits. --SLi 22:41, 11 April 2007 (UTC)

Omegatron was one of the primary authors of the new binary prefix guideline in July 2005 and has vigorously defended it.

Oh really? — Omegatron 20:36, 11 April 2007 (UTC)
I stand corrected, you were an early contributor.[3] - SWTPC6800 20:47, 11 April 2007 (UTC)
I have some objections to the formulation of the second proposal for the guideline. It is too suggestive as it stands. Here an annotated variant:
Remove the guideline. The binary prefixes are not commonly used by computer professionals.or by any major computing standards organization(see list in other proposal). The de facto standard is for "kilobyte" and "megabyte" to denote binary powers of bytes except in certain contexts. The binary prefixes are neologisms that may confuse readers who have had no prior exposure to them. What's more, using these prefixes usually renders article text inconsistent with different from(they are consistent) reliable sources, further compounding possible confusion. Wikipedia should use terms like "megabyte" as they are commonly defined used and note whenever they take on a different meaning (as in the case of hard drives, floppy disks, and optical media).
Woodstone 20:55, 11 April 2007 (UTC)
I'm fine with your proposed changes. -- mattb @ 2007-04-11T22:30Z
Provided we're specific, that's fine, but I think a footnote would be insufficient (especially as people might mistake it for a reference), and a parenthetical, while sufficient, would be confusing and ugly, especially in articles where (Note: "Megabyte" in this context refers to 220 bytes) would have to be used frequently. The "prescriptive" or "Wikipedia's trying to use it first!" or "It's a neologism" are spurious and fallacious arguments. The binary values are values clearly defined by major standards organizations. That makes them attributable, and not whatsoever a neologism. Nor would Wikipedia be doing anything but using them within their existing, predefined meanings. That is a descriptive usage, not a prescriptive one. It is unnecessary and divisive to throw "neologism" and "prescriptive use" around without knowing what those mean. Yes, we do avoid neologisms and prescriptive use, but no one is suggesting using a neologism or any term prescriptively here. Seraphimblade Talk to me 22:47, 11 April 2007 (UTC)
Seraphimblade, we're not asking you to agree with this position, only to agree that it is a fair representation of the arguments. I myself don't agree with the statements in that summary, but I tried to be as fair as possible when I wrote them. Please reconsider or suggest an alternative phrasing. -- mattb @ 2007-04-11T23:29Z
The valid (well, not valid to my line of thinking, but potentially correct) lines of reasoning for opposition that I've seen are as follows.
  • More sources use the xB format than xiB.
  • JEDEC recommends the xB usage.
  • xB can often be clear by context.

Opposition on the grounds of "neologism" or anything of the like is simply invalid, as is opposition based on prescriptive use. Both IEC and SI have prescribed the meaning of the binary prefixes. We would simply be using them descriptively, in keeping with those predetermined meanings, in exactly the same way as using a word in accordance with its standard dictionary definition. Similarly, any terminology approved by two major standard bodies cannot by definition be a neologism. A neologism is a word which was effectively "made up" at some point, not crafted by a standards organization. Seraphimblade Talk to me 09:32, 12 April 2007 (UTC)

It's not opposition on "prescriptive use", it's opposition on an article being insonsistent with the reliable sources used in that article. I have an example of taking the "standard above all else" approach to the logical conclusion. It is prescribed by the BIPM that the metre is the only unit of length to be used. If you really want to follow the standard then you must also remove the term "yard" and "feet" from the article on American football, so please start to do so. If you refuse then you're going to have to explain why you are not prepared to follow the standard. I await your reply with interest. Fnagaton 11:41, 12 April 2007 (UTC)


I agree with mattb with respect to the four terms suggested. Fnagaton 10:44, 12 April 2007 (UTC)

In general I also agree with the content of the keep/remove text produced by mattb for voting options, except to say that perhaps the langauge is a little too suggestive for example using the word "neologisms". I'll quote the two sections below and make some small changes to improve readability and separate the two "sides" of the issue.Fnagaton 10:45, 12 April 2007 (UTC)

Keep the current guideline. The IEC binary prefixes allow articles to express binary capacities unambiguously. Recommending the usage of the binary prefixes in contexts where binary power multipliers are implied shifts the interpretation of the exact quantity intended from the reader to the presumably more knowledgable editor. The current guideline intends to keep unit convention as consistent as possible; using the SI definitions of "kilo", "mega", etc, and using the binary prefixes where powers of two are needed. The guideline reflects the usage recommended by several relevant organizations: BIPM, IEEE, IEC, NIST, etc.
Remove the guideline. The binary prefixes are not commonly used by computer professionals or by any major computing standards organization and may confuse readers who have had no prior exposure to them. What's more, using binary prefixes usually renders article text inconsistent with reliable sources, further compounding possible confusion. The JEDEC recommends using "kilobyte" , "megabyte" etc. The de facto standard is for "kilobyte" and "megabyte" to denote binary powers of bytes except in certain contexts. Wikipedia should use terms like "megabyte" where they are commonly used by reliable sources used in articles and note whenever they take on a different meaning (as in the case of hard drives, floppy disks, and optical media).

Please note the change from "as they are commonly defined" to "where they are commonly used by reliable sources". This language change is important because I don't want to remove all binary prefixes, what I have been insterested in is having consistency with sources. This change allows for xB and xiB to be used depending on how the sources use those terms. This fits with other non-computer related articles where for example other units of measurement (yard, furlong) are primarily used because the sources use those terms despite the BIPM saying to only use the metre. Fnagaton 10:44, 12 April 2007 (UTC)

I don't think the proposed sentence: "using these prefixes usually renders article text different from reliable sources, further compounding possible confusion" captures the full argument. I would suggest: "using these prefixes often renders article text different from cited sources, making it difficult for readers to verify the information presented."--agr 11:44, 12 April 2007 (UTC)
I accept that updated wording from agr because it reads better than the version I proposed, except to say that I think "these prefixes" needs clarifying to be "using binary prefixes often renders article text different from cited sources, making it difficult for readers to verify the information presented". Fnagaton 11:46, 12 April 2007 (UTC)
"JEDEC recommends using kilobyte" is a straight lie. Please stop spreading that fud. --SLi 13:41, 12 April 2007 (UTC)
I based my text from the post by Seraphimblade above. If you don't agree with it then how about something factual such as "The JEDEC defines kilobyte and megabyte in the standard it publishes." There was a link to that provided showing this earlier on. Fnagaton 14:26, 12 April 2007 (UTC)
Perhaps even say that it defines them for use in that standard. OTOH IMO it would be crazy for it not to. --SLi 14:37, 12 April 2007 (UTC)
Also please note that being an organization that claims to be "the leading developer of standards for the solid-state industry" doesn't make it as important as you claim it to be. It's like saying "the leading producer of widgets for frobulators uses this terminology, hence it simply must be correct, and I refuse to listen to any argumentation against that". --SLi 13:51, 12 April 2007 (UTC)
You are wrong to try to dismiss the organisation just because it says something you personally disagree with. It's a lot different to your atempt to misrepresent the organisation by coming up with the anonymous "producer of widgets" example. For you to logically dismiss the JEDEC as "the leading developer of standards for the solid-state industry" then you need to disprove all of the evidence provided by JEDEC and the reliable source in that article. Fnagaton 14:26, 12 April 2007 (UTC)
I think you misunderstood what I said. I don't dismiss JEDEC because it uses kB in the 1024 sense in some of its standards. I'm willing to concede there are lots of such standards, especially older ones. Also while we should apply healthy doubt to any claims of being an industry leader (any company will do so), neither is that the crux of my argument. Rather the point is that even if it is "the leading developer of standards for the solid-state industry", it represents a very minor area of computer engineering and an even smaller area of units. It's like saying we should use furlongs everywhere because there is an organization that claims to standardize horse carriages and it happens to use furlongs in some of its published documents. --SLi 14:37, 12 April 2007 (UTC)
Again you're trying to misrepresent the organisation by using phrases like "minor area", again you're going to have to support your position with something better than "because I said so". Fnagaton 16:27, 12 April 2007 (UTC)
Are you serious in claiming that JEDEC plays such a major role in computer engineering? I have worked in the industry, including in embedded devices development, and I have never heard of JEDEC before, and I bet not many others have either. ANSI, NIST, IEC, IEEE etc. are major names which everybody and their mother know, JEDEC is not. I don't feel I have the proof of burden in proving that an organization is not a very central one, but here you are anyway. Google hits (updated) for ANSI: 25.6M; NIST: 11.9M; IEC: 22.9M; IEEE: 90.7M; JEDEC 1.46M. --SLi 16:38, 12 April 2007 (UTC)
I updated the counts above. I don't know if I mistyped ieee somehow in the search previously, but now I get 50 % more hits for it. Also the hits for the other organizations have increased a bit. I suspect this is not entirely explained by discussing these in Wikipedia :P --SLi 22:08, 12 April 2007 (UTC)
Here's some more: ISO: 191M; W3C: 67.9M; SAE: 14.5M (might be lots of false hits, limiting to "sae (car OR automotive OR auto)" gives 1.49M hits). SI is hard to measure, too much false hits; "si unit" gives 32.1M hits. --SLi 22:35, 12 April 2007 (UTC)
You're the one trying to misrepresent an organisation without any evidence beyond "I said so" or "I never heard of it" and it just so happens they have text on their website you don't like. Those are not valid reasons. I'm going to give you one last chance to come up with something concrete from reliable sources that supports your claims. Fnagaton 16:43, 12 April 2007 (UTC)
What? I don't understand you. I gave those google hit counts to give you a rough idea. What do you want me to do, get a signed note from some JEDEC chairman stating that their organization is not important? And when it's you who's claiming that JEDEC is of any importance, the burden on proof lies with _you_, not with the opposite side to show that JEDEC is not important. So far we have nothing but the statement of the organization itself (WP:RS?) that it is the leading standardizer of something that is not exactly the cornerstone of computer technology. And of course your "I said so" style argumentation. --SLi 19:44, 12 April 2007 (UTC)
No, the burden of proof, the preponderance of evidence, is still against you since you have provided nothing that is substantive to support your earlier statements. Fnagaton 14:12, 13 April 2007 (UTC)
I have given you just about the best possible evidence, which you simply ignore. I have asked you what would be good evidence, but you refuse to name any. I have pointed out that the one who claims something obscure is significant has the burden of proof, but you make it my job to prove that it's not. It has been pointed out that JEDEC does not even try to standardize the units, but you ignore that. Since it is obvious you do not listen to any amount of reasoning, I have nothing more to say about this issue. --SLi 16:28, 13 April 2007 (UTC)
You have not given good evidence because you have written your point of view and personal opinion. I have not refused to name anything either and your part about the JEDEC is misrepresentation, so I demand you retract your statements because they are not accurate. The JEDEC is a standards organisation and that is what the evidence from reliable sources already presented shows. To refute that you tried to claim "I have worked in the industry, including in embedded devices development, and I have never heard of JEDEC before, and I bet not many others have either." That is your personal opinion and it is not relevant evidence from reliable sources. For the last time you need to supply evidence from reliable sources to support your position instead of repeating your personal point of view. Fnagaton 10:25, 14 April 2007 (UTC)
No, that was not evidence nor an attempt at refutation. Instead, for example, the Google numbers above would be. And you still haven't said what you would count as evidence that JEDEC is not major. In my opinion the Google numbers give quite a nice idea over which organization is important and which is not. So go elsewhere to make your ridiculous demands. --SLi 00:49, 15 April 2007 (UTC)
So you retract your earlier statement where you tried claim "I have worked in the industry, including in embedded devices development, and I have never heard of JEDEC before, and I bet not many others have either." because you now say "that was not evidence nor an attempt at refutation". As for not saying what acceptable evidence is, that's also not true because it is a matter of record on this page many times already, to wit reliable sources. Fnagaton 19:25, 15 April 2007 (UTC)
No, I don't retract anything. And you _still_ didn't give a single example of what would be acceptable evidence that JEDEC is a minor organization, nor did you reply to the evidence (for example the Google numbers) I presented. You persist in arguing in circles and misrepresenting what I (and others) said. All that needs to be said has been said already a dozen times, and you haven't listened. I should have understood long ago to stop replying to you about this point. Now I will. If that's what you wanted to accomplish, congratulations. --SLi 23:02, 15 April 2007 (UTC)
You tried to claim the JEDEC is "minor" and when I called you on your personal point of view you then wrote "I have worked in the industry, including in embedded devices development, and I have never heard of JEDEC before", which is more personal point of view. When I again called on you to provide valid evidence you then repeated your personal point of view. Your personal point of view if not valid evidence and the only evidence is of course from reliable sources which I have already stated numerous times. You are misrepresenting what the JEDEC is. You are now also misrepresenting what I have written and this is your final warning, I demand you retract your statements because they lack any truth. Fnagaton 11:37, 16 April 2007 (UTC)


JEDEC standards do not use the words kilobyte, megabyte or gigabyte. They just use KB, MB, or GB. (Kb, Mb, and Gb for bit values.) In 1999 JEDEC and IEC signed a "Memorandum of Understanding" and the IEC binary terms were added as a note to the JEDEC definitions. You will only find the IEC note in the "Terms and Definitions" documents like JESD100-B. If you look at JEDEC standards published this year the terms MB and GB are used.[4] You can search JEDEC documents at www.jedec.org. -- SWTPC6800 15:30, 12 April 2007 (UTC)
The "Memorandum of Understanding" that you refer to involves IEC ESD standards and has absolutely nothing to do with binary prefixes. You're trying to build a case that just isn't there, and this is really grasping at straws. If you're hell-bent on mentioning that the JEDEC utilizes common use of the terms, go ahead, but it's misrepresentation to say that their usage and glossary definitions are an endorsement. For the umpteenth time, in their own words they are "documenting common usage", not recommending unit convention. The standards bodies I listed in the "keep" option actually explicitly recommend using the IEC prefixes. If you want to keep mention of the JEDEC, I suggest making it clear that you are listing them as an example of one standards organization that makes use of the common terms. -- mattb @ 2007-04-12T15:46Z
Oh, I forgot to add ANSI to the list of standards organizations that (explicitly) recommend usage of the IEC binary prefixes. I'd like to add that to the list. -- mattb @ 2007-04-12T15:50Z
Yes you are correct that the agreement was for common work on ESD. I was trying to point out the origin of a note that was placed in a JEDEC wide document. I was also correcting the statement that JEDEC used terms like megabyte. -- SWTPC6800 16:19, 12 April 2007 (UTC)
That's fine, but you didn't point out the origin; you found a totally unrelated agreement and attributed the note about binary prefixes to it. -- mattb @ 2007-04-12T16:24Z

Just as a response to the "descriptive" versus "prescriptive" part — we also strive for accuracy as much as possible, and quite simply, using the SI prefixes to mean anything other than powers of ten is inaccurate and ambiguous. If we have to use a less common naming scheme in order to be accurate, so be it. But I am not comfortable with sacrificing accuracy in our encyclopedia merely because other places live with the inaccuracy. We shouldn't. --Cyde Weys 16:48, 12 April 2007 (UTC)

Exactly. Remember that "common usage" means "completely inconsistent" usage. — Omegatron 17:02, 12 April 2007 (UTC)
It's not "completely inconsistent". Are either of you going to remove feet and yards from the article on American football? Fnagaton 17:15, 12 April 2007 (UTC)
But they are not inconsistent, and repeating the argument a gazillion times won't make them inconsistent. Everybody uses feet to mean 12 inches, it's not like someone defines it as 13 inches and yet another as 11 inches, as is the case with megabytes. --SLi 19:47, 12 April 2007 (UTC)
Actually they were inconsistent, if you know your history. Anyway, that's not the point. The point is people are trying to say "standards organisation says do this" but then they refuse to answer questions regarding measurements which are covered by the same standard organisations. That's the inconsistent part. Fnagaton 20:00, 12 April 2007 (UTC)

Wording of proposals

Collecting comments on the proposed wording here. Are we so far agreed on these reworded versions?

Keep the current guideline. The IEC binary prefixes allow articles to express binary capacities unambiguously. Common usage of prefixes like "kilo-" in the computing context allows for several conflicting definitions. Recommending the usage of the binary prefixes in contexts where binary power multipliers are implied shifts the interpretation of the exact quantity intended from the reader to the presumably more knowledgable editor. The current guideline intends to keep unit convention as consistent as possible; using the SI definitions of "kilo", "mega", etc, and using the binary prefixes where powers of two are needed. The guideline reflects the usage recommended by several major standards organizations: ANSI, BIPM, IEEE, IEC, NIST, etc (refs).
Remove the guideline. The binary prefixes are not commonly used by computer professionals. The de facto standard is for "kilobyte" and "megabyte" to denote binary powers of bytes except in certain contexts. The binary prefixes may confuse readers who have had no prior exposure to them. Using binary prefixes often renders article text different from cited sources, making it difficult for readers to verify the information presented. Wikipedia should use terms like "megabyte" as they are commonly used and note whenever they take on a different meaning (as in the case of hard drives, floppy disks, and optical media).

Woodstone 18:17, 12 April 2007 (UTC)

Before those organisations are listed in the Keep the current guideline I'd like to see specific cites, not general pages. Fnagaton 19:14, 12 April 2007 (UTC)
Every one of those are cited in previous talk archives. Give me a few minutes and I'll find where. Edit: actually, most of those citations are available on the binary prefixes page. IEC is obvious since it proposed the prefixes. ANSI adopted the IEEE's proposal (cited on the binary prefixes page) almost two years ago [5] (page 7). There's also CENELEC and CEN is pending, but I think the list is long enough. -- mattb @ 2007-04-12T19:51Z
There seems to be a nice list at Talk:Binary_prefix#Organization recommendations that lists IEC, IEEE, NIST, SI/BIPM, ISO, ANSI, W3C and SAE, with references. Each of these with the exception of SAE (I think) is way more known than JEDEC, so I wonder why the article itself gives that much weight to the JEDEC usage... --SLi 20:01, 12 April 2007 (UTC)
This is the second sentence of the Binary prefixes article. "When discussing things that naturally come in powers of two (such as computer memory sizes)." The organization that develops the standards for semiconductor memory should be relevant to the binary prefixes article. You cannot purchase a computer that does not use JEDEC standard memory. -- SWTPC6800 02:10, 13 April 2007 (UTC)
Disagreed. The binary prefixes are not neologisms, and the use of that term is unnecessarily prejudicial. A term defined and endorsed by multiple standards organizations cannot by definition be a neologism. I've adjusted accordingly. Seraphimblade Talk to me 20:41, 12 April 2007 (UTC)
I accept the removal of "neologism" (Fnagaton apparently does as well per his earlier comment). While I still think that you could call them neologisms (in the strict sense of the word), the link to WP:NEO was probably too much since it adds some connotation that is unnecessary in this case. -- mattb @ 2007-04-12T22:46Z
I aggree, binary prefixes are uncommon and maybe esoteric but not neologism. -- SWTPC6800 02:05, 13 April 2007 (UTC)
For what it's worth, I'm fine with removing "neologism" as well. I think I was the one who wanted to call it that in the first place, but I think you've convinced me that they're more obscure (and currently non-standard) than neologisms. Bladestorm 15:34, 13 April 2007 (UTC)

I'm not quite happy with the Remove wording. I don't think we should opine on what the de facto standard is. I would rather say something like:

Remove the guideline. There is a long tradition of using "kilobyte" and "megabyte" to denote binary powers of bytes in certain contexts. Despite efforts since 1999 to promote binary prefixes, they are not commonly used by computer manufacturers. The binary prefixes may confuse readers who have had no prior exposure to them. Using binary prefixes often renders article text different from cited sources, making it difficult for readers to verify the information presented. Wikipedia articles should use terms consistent with widely available reliable sources pertinent to the article, with additional verifiable clarifications added as appropriate.

I'd rather see MOS be less prescriptive and place final decisions in the hands of editors who work on the articles or projects in question. Then maybe we can work on helpful ways to clear up ambiguity without triggering constant edit wars.--agr 03:48, 13 April 2007 (UTC)

I prefer the first wording and I prefer that the MOS be very clear on this. What use is a style guide if it doesn't strongly recommend a certain style? Words like "should" are totally appropriate. -- mattb @ 2007-04-13T04:00Z
MOS should be prescriptive, especially in a case like this. The purpose of the MOS is to take a consistent approach. Especially in a case like this, where there's already an ambiguity problem, we would compound it even worse by using different approaches in different articles. Seraphimblade Talk to me 04:03, 13 April 2007 (UTC)
"MOS should be prescriptive"? Give one good reason why. Fnagaton 10:52, 13 April 2007 (UTC)
I prefer the wording proposed by agr at 03:48, 13 April 2007 (UTC). Fnagaton 10:52, 13 April 2007 (UTC)

Let's not rehash all the arguments that are presented at length above. It's clear that the editors arguing for Keep the current guideline prefer a prescriptive approach. I think the other side believes that in the absence of a consensus on the matter outside Wikipedia, MOS should not impose a solution. The poll should reflect that difference in opinion. --agr 11:15, 13 April 2007 (UTC)

That's fine, and I have nothing in particular against your wording. I think you're hurting yourself, though, since a significant number of editors are concerned about consistency within Wikipedia more than anything else. However, as I said, I don't take any issue with your proposed wording. -- mattb @ 2007-04-13T13:28Z

Standards Organizations: Over on Binary_prefix#JEDEC_units_of_semiconductor_storage_capacity there is a proposal to remove the quoted JEDEC standard information as POV. There was also the addition of this image, Windows XP Disk management for 160 GB disk, to give a non POV example of the problem. SWTPC6800 14:21, 13 April 2007 (UTC)

I'm not sure what this has to do with proposal wording, but the Windows XP screen shot makes it clear that Microsoft has not adopted binary prefixes, tho a Vista shot would be better. --agr 15:02, 13 April 2007 (UTC)

I've read document JESD100B.01 (from the JEDEC) again and actually think the language inside it is prescriptive. Certainly it's strong enough for the Remove text to include a reference to the JEDEC ("the leading developer of standards for the solid-state industry") recommending the use of kilo-, mega-. Fnagaton 15:45, 13 April 2007 (UTC)

I oppose the wording that claims using IEC prefixes makes articles inconsistent with reliable sources. There are hardly more reliable sources than the documents published by the major standardization organizations, which have practically all endorsed the IEC prefixes. --SLi 16:30, 13 April 2007 (UTC)

To respond more directly to SLi, there are hundreds of articles in Wikipedia that discuss specific models of computers, and still more that describe various peripherals and other electronic devices that include memory. Some are current models, others are historic. The information in these articles almost always derives from sources that do not use binary prefixes. The point being made in the remove proposal is that using units which differ from these sources makes it hard for a reader to verify information in the article. --agr 17:08, 13 April 2007 (UTC)
Is Micron a reliable source?
"Evaluation samples of Micron's leading-edge 1Gb DDR3 components are available to select customers with production expected to begin early next year. Micron's 1Gb DDR3 components will be available in various output configurations (x4, x8 and x16), and will be fully compliant to the most recent JEDEC DDR3 specifications (JEDEC is the leading developer of standards for the semiconductor industry). These components will support module densities from 512 megabytes (MB) through 4 gigabytes (GB) and a variety of module types including FBDIMMs, UDIMMs, SODIMMs, and RDIMMs." [6] -- SWTPC6800 16:53, 13 April 2007 (UTC)
If you were consistent with that quoted passage and used "1Gb" in articles, you'd be in contention with the guideline to include a space between value and unit. Does adding the space required by SI and recommended by MOSNUM render the article inconsistent with the source? How about adding an 'i'? -- mattb @ 2007-04-13T17:04Z
Oh, come on now matt. You know that isn't even remotely the same thing. The very fact that there are so many people saying that MiB is entirely different from MB inherently illustrates that you're talking apples and oranges here. Bladestorm 17:08, 13 April 2007 (UTC)
Perhaps so, but there are many many people who omit the SI-prescribed space between unit and value. Correcting that for inclusion on Wikipedia is a one-character change that makes the text different from the source but doesn't change the meaning. Just like IEC prefixes! My only point is that JEDEC does not claim to issue or mandate consistent unit standards. Do with that what you will; as I said earlier, I don't oppose mentioning JEDEC in your poll option, but I think it's hinged on an incredibly weak argument. Anyway, we're arguing again, let's concentrate on getting wording that we agree upon. -- mattb @ 2007-04-13T18:24Z

New Version for remove

I have a problem with the remove the guideline argument as it misses the crux of the issue. My real problem with the mebibyte crowd (ironically Firefox is flagging "mebibyte" as a misspelling) has nothing to do with using the MiB abbreviation for the binary prefixs. My main problem is with the flipside that is attempting to incorrectly redefine "KB" & "Kilobyte" and "MB" & "Megabyte" to only mean Thousands or Millions of Bytes contrary to four decades of computer science teachings. Yes I understand the Hard Disk marketers purposefully confused the issue a few years ago, but they were wrong then; they are still wrong now. A megabyte is 1,048,576 bytes. There are hundreds of thousands of authoritative sources that say that. -- KelleyCook 17:01, 13 April 2007 (UTC)

Remove the guideline. Promoting the binary prefixes has the incorrect side effect of redefining the long tradition of using "kilobyte" and "megabyte" from denoting binary powers of bytes to their SI counterparts. Despite efforts since 1999 to introduce binary prefixes, they are rarely used by computer professionals and manufacturers. The binary prefixes may confuse readers who have had no prior exposure to them. Using binary prefixes often renders article text different from cited sources, making it difficult for readers to verify the information presented. Wikipedia articles should use terms consistent with widely available reliable sources pertinent to the article, with additional verifiable clarifications added as appropriate.

For the transition period, i.e. while people still incorrectly use mega in the binary sense, for the reason you mentioned using the IEC prefixes or specifically mentioning that mega is 10^6 should be done when it is important that the exact amount of bytes is 10^6. I assume that would happen quite rarely, though, much less so than where it's important that the exact amount of bytes is 2^n. --SLi 17:06, 13 April 2007 (UTC)
Kudos for succinctly putting why is issue is likely going to end up going to a DR with "while people still incorrectly use mega in the binary sense". Contrary to what you and others are trying to WP:PUSH, "Megabyte" always has meant 1,048,576 bytes and that is not about to change soon. In fact the vast majority of the english speaking world would maintain that it is incorrect to use megabyte to refer to 1,000,000 bytes. -- KelleyCook 17:47, 13 April 2007 (UTC)
Except for hard disks, floppy disks, optical disks, flash drives, and the strict SI definition that the rest of the scientific world uses... This isn't a small issue isolated to hard drive capacities. Saying that megabyte has "always meant" 106 bytes also ignores a bit of history considering that some of the first usage of SI prefixes to denote binary powers did indeed use kilo in the decimal sense (65 kilobytes rather than 64 to express 65536 bytes). The common usage today evolved from a number of factors, and it's rather silly to insist that "megabyte always has meant" one thing when there are plenty of counter-examples, both modern and historical. Your proposed wording is very POV. This is a grey issue and neither your view nor mine is "correct" or incorrect. -- mattb @ 2007-04-13T18:18Z
Yes I said Hard Disk if you had bothered to read what I wrote. Floppy Disks are provably wrong, for example, the original 3.5" 720K floppy (9 sectors/track * 80 tracks/side * 2 sides * 512 bytes/sector) = 1440 * 512 = (720 * 2) * 1024 = 720 * 1024 = 720KB = 737,280 bytes. It has never been called a 720KiB floppy disk. (And yes as a long time computer professional, I understand they completely screwed the pooch when they named the 1.2MB and 1.44MB disk which were actually 1200K and 1440K respectively, where K properly refers to 1024). Furthermore, all Flash Drives also use the MB=1047576 or GB=1073741824 units.
With all that being said, the real Bottom line is that you are obviously on the other side of this issue and should have no input on the wording of our side argument -- KelleyCook 20:23, 13 April 2007 (UTC)
No, "your side" is not allowed to lie in the wording regardless of whether you reach consensus on that or not. --SLi 20:32, 13 April 2007 (UTC)
You're coming very close to violations of WP:AGF and WP:CIVIL. Crotalus horridus (TALKCONTRIBS) 21:00, 13 April 2007 (UTC)
Umm, the grandparent claimed that the other side should have no input on the wording of one side's argument, and I pointed out this is not how it works. I don't understand where you see WP:AGF and WP:CIVIL violations there. --SLi 23:39, 13 April 2007 (UTC)
You accused someone of wanting to "lie in the wording" - such accusations are generally considered uncivil and a failure to assume good faith. Crotalus horridus (TALKCONTRIBS) 23:49, 13 April 2007 (UTC)
That's not true, USB flash drives are (always ?) advertized using MB=106 B and GB=109 B. [7]. Sarenne 20:47, 13 April 2007 (UTC)
I don't really care which terms are used in modern computer articles. But articles about systems from the 8-bit and 16-bit eras must use the SI prefixes since that is what every single contemporary source uses. The English language isn't defined by the IEC; it's defined by common usage and custom. And the common usage and custom was that "kilobyte" referred to 210 bytes, "megabyte" referred to 220 bytes, and so forth. Crotalus horridus (TALKCONTRIBS) 19:53, 13 April 2007 (UTC)
I don't understand the artificial distinction, but if you'd like to propose an alternate or additional poll option, by all means do so. -- mattb @ 2007-04-13T19:59Z
The lines above by Crotalus confirm very clearly the extent of the confusion, where he says that the SI prefixes mean 2n. This is most certainly wrong. Admittedly MB is often used to mean 1048576 bytes, but this is definitely not using the SI prefix M, but forms an alternative use of the capital letter M. −Woodstone 20:05, 13 April 2007 (UTC)
As per WP:V: "The threshold for inclusion in Wikipedia is verifiability, not truth." Every single source from the 8-bit era - without any exception as far as I can see - uses "kilobyte" to refer to 210 bytes. I'll change my mind if you can find a single source - just one - about the Commodore 64, dating from the era when that computer was produced by Commodore (i.e. not Wikipedia or a Wikipedia mirror), that uses "kibibytes" or any of the other neologisms. Wikipedia is not a soapbox; it's not our job to push the new binary prefix terms, it's our job to report accurately what the sources of the time say about these systems. I also question the claim that there ever was a consensus to use binary prefixes. There may have been a consensus among Manual of Style regulars, but not among the people who actually created and edited these articles. Virtually everyone who is actually knowledgable about the subject of 8-bit and 16-bit computers has opposed inserting these neologisms. Crotalus horridus (TALKCONTRIBS) 20:23, 13 April 2007 (UTC)
If it is verifiable that the Commodore 64 has 64 KB of memory with KB=1024 B, it is verifiable that it has 64 KiB of memory. What needs to be verifiable is the meaning. Sarenne 20:47, 13 April 2007 (UTC)
No, what you have just done in this statement is original research by synthesis. Crotalus horridus (TALKCONTRIBS) 23:05, 13 April 2007 (UTC)
Then converting units is original research ? Paraphrasing is original research ? "Verifiable" doesn't mean that every word must be exactly the same as in the source. Wikipedia is not a copy/paste of what you can find somewhere else. You should read WP:NOR. Sarenne 23:22, 13 April 2007 (UTC)

True. And it's perfectly verifiable that standards organizations have defined and advocated use of the binary prefixes (which also means that they cannot, by definition, be neologisms), and that they have perfectly clear and verifiable meanings. We paraphrase sources all the time, that's no issue so long as our paraphrases are in keeping with what those sources were saying. If the source says "512 MB of RAM", it is meaning that the computer has (512*220) bytes of RAM. If we then say it has 512 MiB of RAM, we mean that it has (512*220) bytes of RAM. Same thing! Therefore, the paraphrase is totally acceptable, since we aren't changing the source's meaning, simply paraphrasing wording. (If we didn't do that, our articles would just be collections of direct quotes!) It's actually more verifiably accurate and unchanged than taking a critic's 3 page rant about how bad a movie was and summarizing that "John Smith heavily criticized That One Movie." This being said, polling is a bad idea here, given that canvassing has verifiably occurred (unless, in the interest of fairness, we shall contact everyone who participated in the last poll on this issue.) There have been effectively no valid arguments raised, and "I hate them" is not a good reason not to use them. I've really only seen a couple arguments that make any sense at all, and those are very weak. In that case, a poll is simply a referendum on whether we like the change or not, which doesn't matter at all. For reference, here's all the arguments I've seen, and why they have no validity.

  • "They're neologisms."
    • Anyone stating this needs to read the definition of a wikt:neologism. Were the terms recently coined? For the definition of "recent" that applies to the computing world, not really. Were they "coined" at all? Absolutely not-they were forged by standards organizations. It's not something made up in the server room one day that's gained a modicum of currency, it's something carefully crafted by a lot of people that get paid to do this type of thing all day long.
  • "JEDEC endorses the decimal terms."
    • First, no they really don't, they accept them. Secondly, in the face of ANSI, IEEE, IEC, SI, and so on rejecting them and specifying binary prefix use, that really doesn't matter. To use XB rather than XiB in that case would be to give JEDEC's opinion undue weight.
  • "That's prescriptive rather than descriptive use. That's not our place."
    • Exactly backward. Actually, it is prescriptive use to go against what almost all of the major standards bodies recommend, and say "Nah, we'll do it our way because (we like it better/we think it's less confusing/we don't think the change should have been made/it's always been done that way/XB gets used more often in practice)." It is descriptive use to utilize terms in compliance with the way they're specified by those bodies, whether we like those specifications or not.
  • "It's always been that way."
    • Cubits, hands, and furlongs have been in use a lot longer than meters, inches, and feet. Compared to those terms, meters, inches, and feet are very recent usurpers. Shall we take the terms which were used longest in that case?
  • "We shouldn't change what sources say."
    • We paraphrase sources all day long. So long as we don't change their meaning, this is perfectly acceptable. If we were using a historical text as a source which used Middle English, we would paraphrase that into modern English for our readers.
  • "KB, MB, etc., are no longer ambiguous, they've been redefined as binary anyway."
    • First, that's our take, not that of the standards bodies, so even if that's correct it doesn't matter. Second, the fact that some values (mainly hard drives) really are decimal values does render the term ambiguous.
  • "KB, MB, etc., are used more often in practice."
    • Informal English is used much more often in practice than formal, but we still use a formal tone to write our articles. The word "fuck" is used gratuitously a lot in practice, but we don't insert it gratuitously into our articles. Not everything we do does or should follow common practice, and accuracy and unambiguity are higher priorities.

In sum: There are no valid arguments in favor of decimal prefixes, only a lot of pseudovalid ones. The real issue here is that a lot of people don't like the binary prefixes. In all truth, I think they're kind of lame myself, and they could've picked better terms. But those are the ones that did get picked, they mean what the standards bodies say they mean (verifiability, not truth, again), and overwhelmingly, the standards bodies say decimal means decimal and binary means binary. That's not something we put to a poll or referendum. We don't put in things that are verifiably incorrect, even if we know full well that most everyone would interpret it correctly anyway. In this case, the highest reliable source on standards are standards bodies. And they've spoken. It's not our choice to follow them or not. Seraphimblade Talk to me 21:01, 13 April 2007 (UTC)

Repeating your personal point of view while ignoring all of the arguments that refutes most of what you've just written and trying to use language like "pseudovalid" does not help and is not a valid argument. Fnagaton 11:49, 16 April 2007 (UTC)
About informing everybody who took part in earlier polls, I understood from the discussion of canvassing that (curiously, mainly) those opposing the IEC prefixes agreed that it is OK to do so (in justifying their own deeds). And indeed after the canvassing/advertising by IEC prefix opponents, I pretty much consider that the condition of me having anything at all to do with this poll. --SLi 23:54, 13 April 2007 (UTC)
So how did that (repeated, I may add) diatribe help in the wording of the opposite side of the poll? -- KelleyCook 21:26, 13 April 2007 (UTC)
Hopefully, it didn't. We don't poll over whether or not to use sourced information. Seraphimblade Talk to me 21:32, 13 April 2007 (UTC)
A standard isn't really a standard if no one follows along, now is it? IEC attempted to create a new standard, and that standard was overwhelmingly rejected by the IT community as a whole. Furthermore, it's POV to say that the IEC is somehow a "higher authority" than Commodore, Atari, and all the thousands of other companies that used the traditional meanings over the years. De facto standards trump de jure ones. Furthermore, we do indeed use terms like feet and yards (rather than their metric equivalents) if that is what is widely used by people in the field - e.g. American football. Or perhaps you'd care to move 10-Yard Fight to 9.144-Meter Fight? Crotalus horridus (TALKCONTRIBS) 21:09, 13 April 2007 (UTC)
Firstly, that's a title. We wouldn't change that even if it did use "furlongs". If a computer were named the "512 Megabyte Wonder", we would call it that when referring to it by name, even if the value in question was binary. The same applies to direct quotes (as opposed to paraphrases), we wouldn't change those (even if, as above, the direct quote were in Middle English). But when referring to its memory capacity, we would use the verifiable meanings of terms established by standards bodies. Also, it is not POV to look at the value of sources. In terms of what a standard is, yes, the standards bodies do overrule Commodore, Atari, Apple, or anyone else. They define and set the standards. That is their function. Of course, the binary prefixes article should indeed note that the old decimal prefixes are still common in practice. But for our purposes here, that's irrelevant. If Dell wants to ignore major standards bodies in material it produces, that's Dell's prerogative. But that's not ours. It would be original research to infer that something other than the verifiable standard (the one that the standards bodies have set) is the "real" one. Seraphimblade Talk to me 21:20, 13 April 2007 (UTC)
You have things backward. It's not original research to use the terms that were used in all of the source material. It's original research to replace those terms with seldom-used neologisms in order to enforce the POV that the IEC standards should be followed in preference to long-standing de facto standards. Childish-sounding nonsense like "kibibytes" was not used in the source material for 8-bit computers and game consoles even once. Not once. NOT ONCE. I'll repeat it in case anyone didn't hear: NOT ONCE did ANY original source use these neologisms. It isn't the longtime editors of these articles who want these IEC terms used. It's outsiders who know nothing about 8-bit systems and probably were not even born when they were released, who are intent on ignoring editors who know what they're talking about. Crotalus horridus (TALKCONTRIBS) 23:04, 13 April 2007 (UTC)
Actually, my first system was a used 8-bit (though be damned if I can remember what it was. I wish now I would've kept it, those things can sell for a good bit sometimes!) And I certainly remember when RAM was measured in kilobytes and HD space was measured in megabytes. Even at the time, yes, I knew those terms were ambiguous and not exactly correct. Didn't know of the alternatives then. But we sure do now! As to original research, when it's sourced and verifiable that a value is indeed binary, and that standards exist for reference to such values, that is not original research. It is original research to decide that the term sounds "childish" and doesn't apply to systems past a certain age. I don't particularly like the terms they came up with myself, but those are the ones that the major standards organizations chose and endorsed, so those are the ones we use. Seraphimblade Talk to me 23:32, 13 April 2007 (UTC)
The thing is, no one uses these terms but Wikipedia. For example, a quick search of Newegg.com - generally considered one of the most popular and reputable online retailers - shows 888 matches for "megabyte" and about 26,000 matches for "gigabyte" (admittedly, some of the latter refer to the company). But it does not show a single match for mebibyte and only one match for gibibyte (from a reader comment). No one uses these terms. Laymen don't use these terms and IT professionals don't use them. We are not required to blindly follow a dead-letter "standard" in defiance of all sourcing policy. Crotalus horridus (TALKCONTRIBS) 23:55, 13 April 2007 (UTC)
True. However, Newegg is not tasked with setting standards, so also irrelevant. The discussion here is about what is correct, not what is common. Seraphimblade Talk to me 00:45, 14 April 2007 (UTC)
No, the question is about writing an encyclopaedia, and whether that is best done using unambiguous terms defined by standards bodies, or ambiguous terms used by our sources. And this is not a simple question: there is a tension here, and people can legitimately have different views which way to resolve it. Stephen Turner (Talk) 06:46, 14 April 2007 (UTC)
Please, don't assume so much about what we know about 8-bit systems. That's childish. --SLi 23:48, 13 April 2007 (UTC)
Have you made any contributions to articles on 8-bit systems other than style warring? Crotalus horridus (TALKCONTRIBS) 23:57, 13 April 2007 (UTC)
Just for reference, my own contributions on 8-bit-related articles include: virtually the entire DOS Wedge article, major portions of the SpeedScript article, substantial contributions to the Commodore 1541 article, much of MOS Technology 8563, most of MLX, and various pinout graphics like Image:6510 CPU Pinout.svg, Image:6522 VIA Pinout.svg, Image:6526 CIA Pinout.svg. Crotalus horridus (TALKCONTRIBS) 00:04, 14 April 2007 (UTC)
If you take a look at my edit history, it shows a history of mostly copy editing. I think you underappreciate that work (but don't worry, so do in my opinion most others, and I don't exactly do it for the fame). Also what I do on Wikipedia hardly tells too much about my knowledge about old systems. I generally don't have too much interest in writing new text for articles about subjects I encounter in my work for living. If it matters to you, I have written assembly code for quite a large number of old and new 8-bit and 4-bit processors, both as a hobby and in large industrial projects. These processors include the 6510. I think it shouldn't matter, though. --SLi 00:12, 14 April 2007 (UTC)
Please people. Can you stop bickering long enough to try and agree upon the phrasing? If we cannot agree on voting options then this poll will not happen and the guideline will stand due to lack of clear consensus to change it. The amount of arm waving and finger wagging is getting to be childish at this point. Neither point of view here is absolutely right or wrong, and insistance upon absolutes just makes you look closed minded and stubborn. I tried my darndest to be as neutral as possible when I wrote the summaries, so please think carefully before trying to alter summaries to endorse your perceived "correctness" of the issue. We must only present the facts on both sides in a clear and dispassionate manner. -- mattb @ 2007-04-13T21:17Z
Actually, that's exactly what I'm saying here-one position is verifiably right, one verifiably wrong, and this poll should not happen. It has already been reliably sourced that standards bodies have made their definitions. Our poll here will not change the verifiability of those definitions. If anyone here believes the standards bodies did wrong, they can and should petition them to review their findings. But it's not our place to "correct" them. Straw polls do not override core policy. Seraphimblade Talk to me 21:23, 13 April 2007 (UTC)
Seraphim, I don't believe many people think that it's Wikipedia's obligation to uphold every writing standard that an organization has drafted. Journalistic organizations maintain their own style guides which usually draw from major published standards but also have their minor differences (cf. APA style vs MLA style; you can't follow both to the letter, which one is "right"?). I don't see this as an issue of right and wrong, but as style vs style. However, whatever we choose should be applied consistently, and for that to happen the arguments for both positions need to be presented succinctly and factually. You can strongly support one style without resorting to bickering over which is "right", but the continued arguing and finger pointing and patronizing policy linking doesn't get us any closer to (re-)establishing consensus on this matter (e.g. "if you don't agree with my opinion you obviously haven't read [[WP:XYZ]]"). Please weigh your words carefully before insisting upon absolutes in a stylistic decision. -- mattb @ 2007-04-13T21:33Z
I would never insist on absolutes in any stylistic decision, for example British vs. American spelling. However, this is not a stylistic decision. This is an attempt to hold a straw poll on "Should we allow original research (which includes personal experience and knowledge) or not?" This is not the appropriate venue to hold that discussion. That venue is at WT:NOR, or the village pump, or on the mailing list. (For such a fundamental issue, probably all three). In this case, it has already been sourced that the binary prefixes are the means established by major standards bodies to express binary values. The only countervailing source presented so far is that a minor body accepts other means. The issue has already been decided, then, by prevalence of sources. The reliable sources as to standards are the bodies accepted and entrusted to define and create them. We do not disregard such things because we dislike what they say, and no poll can change that. You could argue to change that policy, as you are welcome to argue to change any policy, but until and unless such happens, we don't do "backdoor" polls as to whether or not they apply. They always do. Seraphimblade Talk to me 21:49, 13 April 2007 (UTC)
I agree we shouldn't argue the merits here, but I must take exception to the notion that changing MB in a source to MiB is "paraphrasing." Such a change implies a certainty as to what MB meant in that source and as has been pointed out above, usage varies, so that certainty must be verifiable. The requirement for verifiability cannot be negotiated away here. To move this along, I would propose we reject the Remove wording that started this section as substitute my previous wording which does not make contentious claims that something is "incorrect."--agr 21:51, 13 April 2007 (UTC)
It's already been suggested that, in order to resolve this argument, any reasonable ambiguity in whether the value in question is binary or decimal should default to the use of decimal, and that the ambiguity should be explained. However, the occurrences of this happening in practice is far less than one percent of cases. In almost all cases, it is clear whether the values in question are binary or decimal. Seraphimblade Talk to me 21:56, 13 April 2007 (UTC)
It's clear by what standards? And who decides? And what is your basis for saying "far less than one percent of cases"? And how does a reader verify what is in the article when the cited source says MB?--agr 22:37, 13 April 2007 (UTC)
Actually, you raise a good point, but a great solution as well! There are plenty of sources that discuss what values are binary vs. which are decimal, and of course the standards bodies' positions can easily be sourced. It would be relatively easy to design a {{cite binary}} template, which could easily be used with any reference to binary prefixes to provide those things as references automagically. I really don't know that that's necessary if we wikilink the prefix to its article, where all those references are present anyway, but more references never hurt. Seraphimblade Talk to me 22:43, 13 April 2007 (UTC)
If you think the problems I mentioned are easily solved, come up with a detailed proposal and we can discuss that.--agr 23:01, 13 April 2007 (UTC)
Here's a very quick and dirty mockup of a template for it. Yes, it's currently rather ugly and requires substing (which to my knowledge anything using the <ref> tags will anyway, they won't work properly if they're transcluded), but any number of sources could be added to that, and we could certainly have "at-a-click" verifiability. Seraphimblade Talk to me 08:03, 14 April 2007 (UTC)

Well this talk exploded overnight. I'd like to point out here that I still think we need to concentrate on the wording of the proposals for the vote on this issue. I also think trying to sort out templates should come afterwards. Fnagaton 10:18, 14 April 2007 (UTC)

In the Keep the current guideline proposal several standards organizations are mentioned. Should the Remove the guideline mention that the U.S. Federal Trade Commission has spent the last 5 years supporting JEDEC's efforts to standardize semiconductor memory? This "Opinion Of The Commission"[8] in the Rambus case cites JEDEC as being "an industrywide standard-setting organization" in the opening paragraph. -- SWTPC6800 20:01, 14 April 2007 (UTC)

I come up empty in that document on searches for "binary prefix", "mebibyte", and "gibibyte". Could you please reference the page(s) where they discuss those? Seraphimblade Talk to me 20:13, 14 April 2007 (UTC)
The FTC note would point out JEDEC's status as a recognized standards organization. That appears to be a major criterion of the existing style.
The JEDEC standards do not use the words megabyte or mebibyte. They use the terms MB and GB (Mb and Gb for bits.) Here is a picture of DDR SDRAM.Image:Memory module DDRAM 20-03-2006.jpg They are labeled correctly as 512MB. (JEDEC Standard) -- SWTPC6800 20:46, 14 April 2007 (UTC)
For the twentieth time, the JEDEC standardizes memory, not unit prefixes. All of the organizations listed in the 'keep' argument have either directly and explicitly endorsed the IEC binary prefixes or have ratified another organization's standard which directly and explicitly endorses the IEC binary prefixes. If you can find a standards organization which directly and explicitly endorses common usage or explicitly recommends not using the IEC binary prefixes, that would be significant. Otherwise let's just stick with the simple (and significant) facts of common usage. -- mattb @ 2007-04-14T22:53Z
Not correct. The JEDEC standards document states that for a document to claim compliance with the standard published it needs to use the terms defined in the standards document. The prefixes defined in the standards document reflect the common usage and those terms are kilo and mega. This fact combined with the new evidence SWTPC6800 does support the argument that the JEDEC is a relevant standards organisation. As such the JEDEC link should be included in the remove text and noted as the relevant standards organisation and that to follow the standard means documents use kilo and mega. This is definitely the kind of new evidence from reliable sources that supports the need for changing the current guideline. Fnagaton 19:31, 15 April 2007 (UTC)
Fine, let's email the JEDEC people and ask them if their specification requires the usage of the prefixes as they use them to claim compliance. That should settle this ridiculous argument. It's amazing to me that you are reading so much into standard technical documentation practice. -- mattb @ 2007-04-15T21:15Z
Not needed, this is because it is already covered by JEDEC document JESD88B 3rd Edition. Page 1 "JEDEC STANDARD. JEDEC Dictionary of Terms for Solid State Technology". Page 2 says "No claims to be in conformance with this standard may be made unless all requirements stated in the standard are met." Page 5 "This dictionary, like any dictionary, is primarily intended to define concepts." i.e. The document defines a dictionary of terms to use in documents describing solid state technology. Page 102: "kilo (K) (as a prefix to units of semiconductor storage capacity): A multiplier equal to 1024 (2^10)". Roughly the same text is used for mega. That is to say KB, kilobyte are terms that comply with the JEDEC standard. Fnagaton 21:38, 15 April 2007 (UTC)
I think it is needed since it's quite apparent to myself and some others that you are reading something into that document that simply isn't there. If you feel that your assertion is iron-clad, you ought to have no problems seeking direct confirmation of it. -- mattb @ 2007-04-15T22:24Z
Matt. I think your side may be reading something into documents that simply isn't there. I suspect if we look at what most of the cited standards organizations said, it was that the binary prefixes should be used in their publications, i.e. standards and related documents, places where unambiguous language is essential. While these organizations recommendations deserve some weight in deciding what Wikipedia should do, there are other organizations that help set standards for usage: Microsoft, Apple, IBM, Dell, HP, Levano, Gateway, the New York Times, the Wall Street Journal, the Times of London. the Associated Press, the major book publishers, and so on. I am not aware of any that have adopted the IEC binary prefixes. --agr 22:31, 15 April 2007 (UTC)
That may be, but even if you're correct, you're arguing for their use here. This is not a newspaper, or a sales document, it is a reference work. In any reference work, unambiguous and precise language is essential. Seraphimblade Talk to me 22:36, 15 April 2007 (UTC)
May I point out that this is an encyclopedia, so unambiguous language is precisely what we're after. We are talking about technical terms, and all of the standards organizations I listed set the norm for technical documents and specifications. What's more, the organizations I listed explicitly recommend IEC prefix usage, there is no roundabout interpretation necessary to reach that conclusion. Have you taken a look at the documents in which they make these recommendations to see the context, or are you suggesting that I'm exaggerating just to go tit-for-tat with my assertion as regards the JEDEC? In any case, I've emailed the JEDEC asking whether their standard requires usage of "kilobyte", "megabyte", etc in order to produce a product that is compliant with their specifications. I will post the result when I get it. -- mattb @ 2007-04-15T22:50Z
Indeed we are an encyclopedia. Can you name a single published encyclopedia that uses binary prefixes? Does Britannica? Second, if accuracy is important we should stick to the usage of the sources we cite. Using binary prefixes when sources don't, as you propose, involves some technical interpretation of those sources, which is subject to error. There are hundreds of articles involved and thousands of editors. We can't check them all. We depend on our readers being able to check for themselves. That verifiabliity is vital to our credibility. --agr 23:03, 15 April 2007 (UTC)
Can we at least feign staying on track? Do you guys want this vote to happen or not? I've sent an email to JEDEC trying to resolve this dispute over "what they may have meant", can you guys try and continue to discuss the wording? Bickering forever will not show any consensus and won't overturn the current guideline. -- mattb @ 2007-04-15T23:39Z

Although I have said before I would not comment on this issue any more, because all had been said already and sometimes more than once, so much has been said since (much of which I did not bother to read) that I feel compelled to rejoin the debate.

I think the basic problem is described but not solved easily: one side wants consistency within Wikipedia over all fields and the other side wants consistency inside and outside Wikipedia within one field. These are pretty much mutually exclusive as long as the predominant use outside Wikipedia stays the way it is.

Some thoughts of mine:

  • The use of the “IEC prefixes” (in a project as large as Wikipedia) is only of true benefit if they and their decimal counterparts are applied consistently.
  • A manual of style for an encyclopædia may well choose to establish guidelines uncommon to specific fields, because it has a broader scope.
  • Modern-era unit names come into language through their standardisation by eligible bodies.
  • There are comparable precedents, especially in the area of traditional English units bearing ambiguous meaning; e.g. mile, pound and ounce, gallon and pint. Disambiguation is required in many cases.
  • Most readers will probably skip over the i. Others may be confused and either might learn about the still rather new prefixes or misassume an error. (“M2B” might have been more intuitive.) It is impossible to know whether more of them welcome the notation or reject it.
  • The term “binary megabyte” (and analogous terms for other prefixes and the bit) may be acceptable in place of “mebibyte”. This does not work for symbols.
  • New units (or unit prefixes for that matter) may well be used to decribed older measures. Text quoted literally should not be changed, though, but such instances are assumed to be rare. Prose should not be changed to direct quotations just to avoid using binary prefixes. Christoph Päper 17:50, 16 April 2007 (UTC)

Lighten up time


* My goal is to promote the usage of ambiguous units, in order to maintain consistency with their inconsistent traditional usage. It's clearly not at all important what the units actually mean; the negligible difference between meanings would never generate even a few words of discussion on talk pages, so who cares? We should also abandon the use of the metric system, which is not common usage for most of our readers (Americans) and may confuse them. No one uses metric anyway, except overly prescriptive technical types in pedantic, obscure documents, and doesn't make a difference to most of our readers. I've certainly never heard of a "killa-meter". What a silly word. We should just use miles like everyone else. The NIST defines the meaning of these American units in their official documents, and I don't see why we shouldn't just follow this one organization and ignore widely agreed upon international standards that none of my friends actually use. It's the way things have always been done. When an original source specifies a length in cubits, we should use cubits and only cubits; we need to maintain accuracy and consistency with the original sources. We should not provide any conversions to standard units or explanations of which particular cubit we mean, as this might confuse readers. We should use Egyptian cubits in articles about the pyramids, and Babylonian cubits in articles about the Hanging Gardens. The 4.5% difference between articles is unimportant; readers who are familiar with the subject will know what we mean. Besides, if we made it policy to convert units to a common standard that everyone could understand, overzealous editors might convert units incorrectly every once in a while! This could never be fixed by other editors and would completely negate any benefits that the vast majority of correct conversions might provide. — Omegatron 16:53, 17 April 2007 (UTC)

Sir, that was worthy of Pope. Thanks for the smile. -- mattb
Sarcasm and a weak attempt at misrepresentation of the real arguments is not helpful in discussing the issue, especially considering somebody has attempted to use an open proxy to edit the Commodore 64 article to add binary prefixes again. Fnagaton 17:41, 17 April 2007 (UTC)
Lighten up, some humor isn't the end of the world any more than a couple of 'i's are. -- mattb
I put it to you that if you really think "those little 'i's" are not that important then stop insisting that they be used. Bladestorm has made a concerted effort to get this sorted out and then an editor comes along and makes comments that are at best unhelpful and in the wrong place, at worst since the act was deliberate I call that vandalism. Fnagaton 18:10, 17 April 2007 (UTC)
I know it looks like we may be overreacting a bit, but I was hoping we could get a very quick and easy-to-read synopsis of what everyone truly wants to achieve from this. And he decided to argue instead of cooperate, to use unconstructive sarcasm, and (possibly worse of all), he took up about the same amount of space as three real entries. Sometimes people get frustrated, and don't express it in the way we'd like them to. That's fine. But it isn't fair that we shouldn't be able to actually try to get something accomplished here. Bladestorm 18:00, 17 April 2007 (UTC)

I agree with matt, a little humor can be helpful: My goal is to get Wikipedia to use these new terms I heard about called binary prefixes. I think they are really, really cool! It's about time that everybody uses them, because they're sooo much more precise that the terms all the computer companies use. What do they know anyway? I could try to convince the editors of the individual computer articles to accept them, but they are a bunch of ignorant jerks who don't see the transcendent importance of clearing up a 4.5% ambiguity in memory size and have these silly ideas that readers should be able to verify the content of articles on their own using the cited sources. So me and my friends are going to write a guideline that everyone has to follow. And we'll even add language that says if we change something, those jerks can't change it back. So what if our plan conflicts with policies like WP:V, WP:OR, WP:RS, WP:AGF, WP:SOAP, WP:OWN? I mean nobody really takes policies seriously. Oh, and terms like kibibyte and mebibyte are not neologisms, even though they're new and not in dictionaries. They are in Wictionary! What it clearly says in WP:NEO doesn't apply to us. And if a few articles end up with incorrect information, so what? At least they will look the same as all the other articles. It's only a 4.5% difference anyway. --agr 18:41, 17 April 2007 (UTC)

Pulling it together

Holy crap... I go away for a weekend, and this is what I come back to?!?
"There are no valid arguments in favor of decimal prefixes, only a lot of pseudovalid ones"-Wow. Real mature. Countless valid reasons were given. Here's one: That's what most people use. Wikipedia is prescriptive, not descriptive. That means we call things what they are called, not what they should be called. If several organizations were to announce that there will now be gender-neutral pronouns in the english language, it still wouldn't be appropriate to adopt them until they were truly adopted by the public. Would gender-neutral pronouns be more precise? Sure. But they currently aren't "real" english. Frankly, I don't know why it seems like so many people are incapable of realizing something bloody obvious: There are two sides here.(I know, matt; you've been very clear that you realize this) Putting down, or outright ignoring the existence of, your opposition's argument will not solve a single bloody thing.
"About informing everybody who took part in earlier polls, I understood from the discussion of canvassing that (curiously, mainly) those opposing the IEC prefixes agreed that it is OK to do so (in justifying their own deeds)" -For reference, I thought that it would be fair to notify people. Specifically, I thought it'd be fair to notify anyone on both sides of the issue; simply because I want everyone with an opinion to know that it's currently wanted. I don't see what the problem with that is. And I am not justifying my actions so far. Since matt (who's been pretty fair about everything else, so I tend to trust his instincts on this) still didn't like the idea, I haven't notified anyone. We really don't need any bad-faith accusations here.
Seriously people, can we get back on-track here? Specifically, no more arguing about every little detail; and certainly no more arguments about whether or not people's arguments are valid. SLi and Sarenne have been very good at suspending their (related) editing while we're debating this, and I'd like to close this once and for all.

I think part of the reason we can't agree on phrasings is that we haven't even entirely established what both sides want.
Please, can everyone just (very quickly) summarize what they want here? No in-depth arguments. No more bickering. No more nonsense like, "there's verifiably only one correct option here." Just a resolution? All we need is to decide:

  1. What each side wants to happen.
  2. How to phrase it.

That's it.
So... What do you want to get out of this? Bladestorm 14:02, 16 April 2007 (UTC)
MB/KB:

  1. At least the potential to choose which units you want on an article-by-article basis, depending on which seems more appropriate within the context of that article.
  2. The ability to use IEC prefixes for binary and standard for decimal values, when it improves the readability of the article.
  3. The ability to simply say MB when referring to values taken from sources that use the same units.
  4. The ability to IgnoreAllRules, or defer to common sense, when there are specific problems with using either set of units.


MiB/KiB:

  1. Only IEC prefixes for values used in binary sense. (Direct quotes being an obvious exception)

There are two sides here. ... Putting down, or outright ignoring the existence of, your opposition's argument will not solve a single bloody thing.

Actually, there are three sides, and all of their arguments were already addressed in 2005. — Omegatron 20:16, 17 April 2007 (UTC)
Then why did you change it to use IEC prefixes? —Centrxtalk • 20:57, 17 April 2007 (UTC)
Actually you're wrong because new points have been made that have not been covered before and since it is after the MoS change then that is also cause for the new changes to be debated after they have been tried in the real world. I have a feeling BS may want to move your comment and this one up above out of the way... Fnagaton 20:37, 17 April 2007 (UTC)

Discussion of Goals

Please, only discuss what you want to achieve here. (It's okay to list something that's already being achieved by the current version of MoS)

  • What I want is where articles have a majority of sources that use kilobyte/KB/megabyte/MB etc then those articles should use those terms and not binary prefixes, this is especially the case for articles that discuss older computers. This is to improve the consistency of the article with the sources and allows the reader to verify the terminology used. The exact phrasing will depend on the reply from the JEDEC. Fnagaton 14:09, 16 April 2007 (UTC)
  • My primary goal is to put editing of computer-related articles on the same footing as other articles by removing the following sentence from the guideline "If a contributor changes an article's usage from kilo- to kibi-, for instance, where the units are in fact binary, that change should be accepted." My secondary goals are to keep Wikipedia as widely accessible as possible to a general audience and to insure that articles about computer systems and related devices, including historic ones, are not changed in ways that make them harder to verify against original sources.--agr 20:18, 16 April 2007 (UTC)
  • I want, for articles on 8-bit computers especially, to be able to use the same terms (KB/MB) that all sources use, rather than inserting unpopular neologisms. I can find you a hundred sources saying the Commodore 64 has 64 kilobytes of RAM; I can't find a single source but Wikipedia and its mirrors that insists it has 64 "kibibytes." Above all, this should be handled on an article level, by article editors who are familiar with the subject and know what terms are used by reliable sources in the field, not centrally at MoS. Crotalus horridus (TALKCONTRIBS) 03:25, 17 April 2007 (UTC)
  • My goal is to maintain accuracy and consistency. I understand the desire to keep articles consistent with sources, but have to disagree that adding a little 'i' to abbreviations will significantly impede that. More importantly, I think it's utterly self-defeating to accept inconsistent usage of the prefixes. They are only useful to us if applied uniformly over computing-related articles. I believe that the wording I already proposed adequately covers what I want to get across in a pretty unbiased manner. -- mattb 15:02, 17 April 2007 (UTC)
  • My goal is to provide the reader with unambiguous and accurate information. Interpreting the meaning of units used in sources should be the burden of the editor, not the reader. Verifying articles against sources is not something a reader wants to to, but a knowledgeable editor should do it. −Woodstone 18:59, 17 April 2007 (UTC)
  • Wikipedia is a reference work. SI prefixes are standardized and should be used consistently throughout. — Omegatron 20:04, 17 April 2007 (UTC)
  • Wikipedia articles are created by volunteers. Style police who ignore the consensus of the contributors will not encourage these volunteers to create more articles or add to existing ones. A style guideline that states that one person's opinion should be accepted is the antithesis of Wikipedia's policies on building consensus and respecting other contributors. The vast majority of reliable sources do not use IEC binary prefixes. Wikipedia is not the place to correct the world's use of binary units. SWTPC6800 00:45, 18 April 2007 (UTC)
    • I think that is far beyond the scope of the discussion at hand. You're addressing the much larger issue of whether style guides should be viewed as community consensus, and you're opening a giant can 'o worms if you want to bring that into this discussion. Can you kindly start another discussion on that matter (at the village pump or even on this talk page) and leave this section for discussing the core issues surrounding usage of the binary prefixes? -- mattb
      • The Manual of style is not a prescriptive document. It contains prescriptive elements, but it does not prescribe styles that are directly contrary to the style used on Wikipedia—and the larger world. Doing so is plain ineffective and is misleading to anyone using the styleguide as a reference. Wikipedia does not have a top-down structure. The new binary prefixes has never been the style used on Wikipedia and it is still not the style used on Wikipedia, even though it was added a year ago. The statements on this page are a lie. —Centrxtalk • 01:08, 18 April 2007 (UTC)
      • Thanks, that's extremely helpful to the discussion at hand in which there are numerous editors with valid opinions on both sides of the table. Classifying either point of view as "a lie" is top-notch, really, thanks for facilitating open discussion. -- mattb
        • I didn't classify any point of view as a lie. I classified the statements on this page a lie, which they are. —Centrxtalk • 01:29, 18 April 2007 (UTC)
          • Which statements? If you're going to accuse a person (or persons) of lying, don't beat around the bush. -- mattb 01:47, 18 April 2007 (UTC)
            • I think this discussion of lying should stop now. To everyone's credit, this discussion has stayed civil and I would hope it will return to that standard.--agr 13:33, 18 April 2007 (UTC)
  • I want the Manual of Style to be a useful reference for authors seeking for advice on best practice (to resolve an edit war for example). It should reject ambiguity and promote uniformity, because this is facts not minds. Wikipedia should look as consistent internally as possible, external consistency comes second in my opinion. In this particular case this leads to my support for binary prefixes, especially in symbols. The mention of non-reverting correct additions of binary prefixes just rephrases other guidelines (which are often ignored, though, because of the problem of perceived “article ownership” alien to my understanding of the idea behind Wikipedia). Christoph Päper 11:43, 18 April 2007 (UTC)

Final Phrasing

Please try to leave this section blank until we've established what we want. If you do wish to put your own phrasing here, realize that it may get replaced by another one later on.

Common usage

Remove binary prefix mandate There is a long tradition in the computer industry of using "kilobyte" and "megabyte" to denote binary powers of bytes in certain contexts. Despite efforts since 1999 to promote the IEC binary prefixes, they have not been adopted by major computer manufacturers, or publishers of newspapers, trade books and print encyclopedias. They do not appear in most, if not all, standard dictionaries. "Neologisms are words and terms that have recently been coined, generally do not appear in any dictionary, but may be used widely or within certain communities." (from WP:NEO) Wikipedia should not lead the way in introducing new terminology. Wikipedia should not be the first general reference work to adopt new terminology. Using binary prefixes often renders article text different from cited sources, making it difficult for readers to verify the information presented. "Wikipedia is a tertiary source that includes material on the basis of verifiability, not truth." (ibid.) Shifting "the interpretation of the exact quantity intended from the reader to the presumably more knowledgable editor," as supporters of the existing guideline propose, In many cases, using binary prefixes requires editors to use their expertise to interpret what original sources mean. This contradicts no original research and verifiability, two of Wikipedia's pillar policies. "Determining which meaning is the true meaning is original research — we don't do that here at Wikipedia." (ibid.) The ambiguities introduced by common usage are small, well known to readers who care and inconsequential to most other readers. The ambiguities can be dealt with in other ways, such as including explanatory text or links to an article describing the problem. Wikipedia articles should use terms consistent with widely available reliable sources pertinent to the article, with additional verifiable clarifications added as appropriate. --agr 21:40, 17 April 2007 (UTC) rev.--agr 02:35, 18 April 2007 (UTC)
I've received one reply from the JEDEC on the kilobyte/megabyte standards requirement issue and I'm waiting to hear back from their expert. I'd like to wait another few days for a reply from them as there may be some text to add regarding the standard use of these terms. Fnagaton 22:03, 17 April 2007 (UTC)
I also emailed them several days ago. Arnold, your proposed wording is far from dispassionate and mixes a lot of interpretation with fact. I thought most of us had agreed above that the neologism statement isn't appropriate. "Lead the way" is borderline POV since our usage is supported by several major standards organizations, so we have certainly not come up with this all on our own. What's more it's not appropriate to counter the other position in the poll option. These statements should be purely factual and free of argumentative language, otherwise we're right back to twenty pages of tit-for-tat debate. I tried my best to do that in the "keep" proposal, and I ask that you do the same (if you think there's any non-factual statement in the keep proposal, point it out and we'll try to come to an agreement). This is a major step backwards from all the previous proposed wordings and I wouldn't think endorsing it. -- mattb
I replaced the "lead the way" sentence and no longer quote the other position. And even though the new prefixes were coined by a recognized standards body, they are still neologisms, both by the ordinary meaning of the word and the definition in WP:NEO. I'm not aware that anyone on the remove side agreed otherwise.--agr 02:35, 18 April 2007 (UTC)
I've not received a reply from the JEDEC expert yet, however I'm out of the country without any internet access until Sunday evening my time. Since there may be a reply from the JEDEC during that time perhaps a delay until Sunday would be prudent to ensure the best possible information goes into this proposal? Fnagaton 08:20, 18 April 2007 (UTC)
I wouldn't worry; this won't proceed until everyone is happy, which could be some time yet. What's more, I'll post whatever response I get from JEDEC as well. -- mattb 14:09, 18 April 2007 (UTC)
Not received anythng yet. Fnagaton 13:06, 23 April 2007 (UTC)

Quite a long rant here. If people vote for this, what are they actually voting for? — Omegatron 17:51, 20 April 2007 (UTC)

Exactly what it states in the first line - Remove binary prefix mandate. Same structure as your group's paragraph below, and I think its sums up the position pretty nicely though it could use some more work on the flow of positions. --Marty Goldberg 18:49, 20 April 2007 (UTC)
Omegatron has a point; there are far too many words. mattb also has a point; the wording is overly emotional and divisive. The direct quotations from WP:NEO are unneeded (and make it even longer) and should probably be paraphrased or removed instead. — Aluvus t/c 23:44, 20 April 2007 (UTC)
I think agr summed up this complex issue quite succinctly. Fnagaton 10:08, 24 April 2007 (UTC)
Ironic you say that since earlier you explicitly opposed some of the phrasing he used (as regards "neologism"). Again, I can't support this wording at all, it's far to argumentative and mixes a lot of interpretation with fact. -- mattb 14:28, 24 April 2007 (UTC)
No you're wrong because since I think the later arguments presented in WP:NEO are quite strong and do apply to this case, so it's not at all ironic. Fnagaton 14:35, 24 April 2007 (UTC)

Standard SI / IEC usage

Keep the current guideline. The IEC binary prefixes allow articles to express binary capacities unambiguously. Common usage of prefixes like "kilo-" in the computing context allows for several conflicting definitions. Recommending the usage of the binary prefixes in contexts where binary power multipliers are implied shifts the interpretation of the exact quantity intended from the reader to the presumably more knowledgable editor. The current guideline intends to keep unit convention as consistent as possible; using the SI definitions of "kilo", "mega", etc consistently across all articles, and using the binary prefixes where powers of two are needed. The guideline reflects the usage recommended by several major standards organizations: ANSI, BIPM, IEEE, IEC, NIST, etc. (refs). Changing "kB" to "KiB" or "MB" to "MiB" is less likely to confuse the reader than using several different meanings for kB or MB. The resulting unambiguity is an essential property for an encyclopedia.

Oppose poll

Refuse to participate in or recognize an allegedly inherently biased poll. The sole purpose of a poll is to get an idea on what would be the general consensus of the interested Wikipedia editors who would be eligible to vote if a full vote was conducted. For this, a representative sample is needed. While consensus can change, conducting a poll right after a copy edit campaign to change to MoS recommended usage and those wanting to change the previous consensus wording having advertised it where the changes have been specifically opposed is allegedly unlikely to get a sample which reflects the opinions of all interested editors. Allegedly, not a single case of someone who voted for the wording in the previous poll changing his mind has been demonstrated. Similarly no reason has been shown why the opinions on this issue among new Wikipedia editors on average would differ from those who edited when the previous decision was made. Accordingly, we consider the previous poll much more likely to still reflect the opinions of interested editors than any new poll conducted under these circumstances. This does not mean that it would never be possible to conduct a new poll on the issue given more neutral circumstances.

I strongly oppose this third option because it contains parts that are not true, phrases that are PoV and not fact, phrases that try to second guess the minds of editors that have not actually posted. All this combined with the repeated use of "alledegly" is an attempt to push PoV into the poll and misrepresent the actual issues. However I've been told not to change it and SLi has been 3RR blocked, but instead I have been told that it is OK to ask someone else to change it. I suggest making the changes here: [[9]] These changes remove the sections that are not true, use PoV etc. Fnagaton 19:47, 29 April 2007 (UTC)
Just make sure it still reflects the original point and the original arguments after that. In my view the issue is quite simple: If you don't like the poll option, don't vote for it. Of course if it included something *trivially* untrue, I think it would be a fair game to remove it. However don't pretend there is no issue to be voted for just because you don't agree with the text.
As a side note, something you probably consider good news, it may be that I won't be editing Wikipedia too much for the next few weeks for reasons unrelated to this dispute. OTOH I think it may do good to me even wrt this dispute, as I edit Wikipedia simply for fun and disputes like this aren't mostly fun. Perhaps I just should leave controversial issues like these to those who love Wikipedia like their child and want to go to great lengths to protect it (to me it's just a hobby among others).
Why can't you keep the ---- between the poll options and the discussion? --SLi 08:44, 30 April 2007 (UTC)
Re "Just make sure it still reflects the original point" is not possible because your original point is that you write things where you supply no evidence and want to push that point of view. The issue is simple, you are writing a poll entry with dubious claims and using the word "allegedly" to avoid having to justify with hard facts. I could, if I were to follow your example, write something like "Allegedly the IEC, SI and IEEE are looking to scrap the use of binary prefixes". That follows your example because it is a statement without using any facts and uses the word "allegedly". That's why you should remove the text and replace it with something less dubious. Fnagaton 09:24, 30 April 2007 (UTC)
I've removed the sections where you have not provided evidence for your assumptions or where it's your point of view. You don't know the minds of those who voted previously, for example so you cannot assume. You're also making the assumption the circumstnaces are not neutral. Gathering the opinions of interested editors (those that contribute to the articles) is exactly what a neutral poll should encourage by the way. Forcing through changes behind the backs of the wider population with a vocal minority of editors would not be a neutral poll. Fnagaton 09:01, 27 April 2007 (UTC)
I edited those slightly and removed your strikeouts as they remove the entire point, which you just are not allowed to do. --SLi 02:12, 28 April 2007 (UTC)
Again I have removed the sections where you have not provided evidence for your assumptions or where it's your point of view, even though you wrote "allegedly" that still doesn't excuse you from misrepresentation and use of overly emotive language without any basis in fact. Fnagaton 09:13, 28 April 2007 (UTC)
SLi I have again removed the text since you seem intent on including phrases that are not true. For example you cannot include PoV where you claim the poll is biased, since you have no proof. You cannot include your PoV about the other parts because they are your unproven assumptions. Do not just revert the changes without discussion, if you want to add different text then you're going to have to stick to the facts instead of showing a lack of good faith by trying to second guess the minds of other posters. It is not good enough to simply put "allegedly" everywhere because that still makes it your PoV. Lastly your use of the wording "neutral" is your unrpoven assumption that you think the poll is not neutral, do not use such language. Fnagaton 19:25, 28 April 2007 (UTC)
For the record it now looks like SLi is removing talk page comments. [10] instead of talking about them. Fnagaton 19:57, 28 April 2007 (UTC)
Whoops, *that* wasn't intentional. Sorry, restoring. Also _please_ keep the ---- below BETWEEN the options and comments. --SLi 20:00, 28 April 2007 (UTC)
Here's the comment I mistakenly removed:
SLi I have again removed the text since you seem intent on including phrases that are not true. For example you cannot include PoV where you claim the poll is biased, since you have no proof. You cannot include your PoV about the other parts because they are your unproven assumptions. Do not just revert the changes without discussion, if you want to add different text then you're going to have to stick to the facts instead of showing a lack of good faith by trying to second guess the minds of other posters. It is not good enough to simply put "allegedly" everywhere because that still makes it your PoV. Lastly your use of the wording "neutral" is your unrpoven assumption that you think the poll is not neutral, do not use such language. Fnagaton 19:25, 28 April 2007 (UTC)
I'm striking out the statements added by SLi. I agree with them, but the point they convey is already covered in the first and second sentence. We should allow the potential voters to decide whether the ambiguity is acceptable or not and how confusing the IEC prefixes are likely to be. Again, this statement should be very terse, clear, factual, and not try to push an opinion. -- mattb
Um, I thought that's the point of the wordings, to give the voters opinions they can agree or disagree with. But ok, I'm fine with them including strictly facts only. --SLi 21:46, 18 April 2007 (UTC)

This isn't so much about promoting IEC prefixes as it is about using SI prefixes correctly. We need to maintain consistency from one article to the next. — Omegatron 17:40, 20 April 2007 (UTC)

Using your logic then you must edit articles to only use the metre and remove references to yards and inches. Until you do so then don't try to use the fallacious argument about consistency because the JEDEC is the standards organsiation that covers computer memory and they define KB (kilo-) and MB (mega-) in their standards documents. The consistency with reliable sources is actually more important in Wikipedia. Fnagaton 13:10, 23 April 2007 (UTC)
Yards and inches don't have inconsistent and conflicting definitions in modern usage. Not since 1958. The JEDEC argument requires no more discussion. We think it's a ridiculous over-interpretation of standard technical documentation practice, you think it's prescriptive. That will be put to rest one way or the other when/if the JEDEC people respond to our emails. -- mattb 14:39, 24 April 2007 (UTC)
So you admit yards have been inconsistent. Using kilobyte and megabyte and KB and MB as stated in the JEDEC standard is not inconsistent either and that is also the modern interpretation. So that means your argument is actually incorrect. The point is SI does say "use metre" (I checked, it does not allow yard) yet Wikipedia uses yards in some articles as already shown by others and myself on this page. So I have to ask why are you not following SI and changing those articles to only use the metre? Nobody who wants to use binary prefixes has answered that specific question when it has been put in various forms on this page. So I'll ask it directly to you instead. Also don't try to second guess and misrepresent what my actual position is by attempting to state what you think it is. Fnagaton 14:54, 24 April 2007 (UTC)
I'm done arguing with you. This is so cyclical it's nauseating. I'll be back if you ever garner enough support for a poll, which doesn't seem likely at this rate. -- mattb 15:06, 24 April 2007 (UTC)
I hope this means you are going to stay quiet instead of continuing your attempts to misrepresent what I write, all your "it's ironic" snide comments do not help. Why some people don't answer the SI question is an interesting problem. I think it's because those people know they are being inconsistent by only choosing to follow one SI "standard", which in turn means this means they know their argument is weak. I think deep down they know Wikipedia should reflect the terms used in the sources, as demonstrated by the American Football article for example. Fnagaton 15:20, 24 April 2007 (UTC)
Your yards/inches question has been answered. Perhaps nobody is bothering to answer it again because it's a loaded question and is worthless to this discussion. Take that for whatever its worth, because I've wasted far too much effort debating this. If you can get support for a poll, so be it, if not, move on. -- mattb 15:48, 24 April 2007 (UTC)
No it has not been answered on this page, not properly anyway. If you can show where it has actually been answered properly on this page without that answer ignoring various important aspects then I'd be interested in seeing it. I don't think it's wasted effort to debate a topic that needs to be discussed. Fnagaton 16:00, 24 April 2007 (UTC)
Added a new option. I believe the third option significantly differs from both previous ones presented, and is one I would myself be comfortable voting for. Although please don't take this as endorsing a poll on the issue. --SLi 18:53, 26 April 2007 (UTC)

The correct way to form a poll like this is to propose various wordings and let editors vote for the ones they like. Every editor has their own rationales for why they would prefer certain wordings, and those rationales should be their vote. That's the whole point of opinion polls. You don't ask people what their goals are first and then try to tailor the vote to address all of their concerns in 1 of 2 voting options. If you're capable of doing that, you should be capable of forming a consensus that takes everyone's concerns into account without resorting to a poll in the first place.

This voting proposal and the pages and pages of text that are going into creating some perfectly worded voting options are very misguided and a poor attempt to polarize the issue into two mutually exclusive camps.

Please read false dilemma, Polling discourages consensus, Wikipedia:Polling is not a substitute for discussion, m:Polling is evil, etc. — Omegatron 17:58, 20 April 2007 (UTC)

Wikipedia is also not a place to force opinions onto people. I'm getting sick and tired of the recent anonymous edits to Commodore 64. Basically this comment is directed to the anonymous editor (I have no doubt that it is one person doing this) who is having to use a proxy to make their edits to make them realise that since they are having to go to such lengths to force their opinion then hopefully they would stop and realise their opinion is not as correct and they seem to think it is. Fnagaton 14:30, 23 April 2007 (UTC)
No offense intended, but I find this comment slightly ironic when your primary contribution to Wikipedia thus far has been to try and change the binary prefixes guideline. I take with a grain of salt your expectations of what Wikipedia is and is not. -- mattb 14:53, 23 April 2007 (UTC)
That is misrepresentation. Fnagaton 15:02, 23 April 2007 (UTC)
Wait. You just started editing on April 2nd, and were editing this talk page by April 4th? — Omegatron 03:50, 24 April 2007 (UTC)
You are using ad hominem and that is not a valid argument, i.e. you are attempting to attack the person instead of tackling the real issues. I suggest you retract your comments and concentrate on the actual topic. Fnagaton 10:06, 24 April 2007 (UTC)
Wikipedia seeks to encourage new editors. I think their opinion on a matter like this is particularly valuable. They have not been jaded by years of Wiki-politics and posturing.--agr 11:18, 24 April 2007 (UTC)
That's fine. But the issue has already been debated to death and settled. The whole point of guidelines is to prevent endless debate like this.
A strong majority agrees that we should use SI prefixes consistently across articles, and, although they aren't perfect, the IEC prefixes should be used where appropriate. A smaller number of editors thinks the IEC prefixes are too new and not widespread enough for us to use, so we compromised and said that the ambiguous units are allowed, but if they're changed to be more accurate in certain circumstances where it's important, the change should be accepted.
That's that. The decision has already been made.
We all agree that consensus can change, but no new arguments are being introduced, and the original neologism argument is only becoming less valid as time goes by. — Omegatron 18:26, 24 April 2007 (UTC)
Not correct. One of the new arguments is that the "guideline" has been tested in the wild and found to cause more problems than it was meant to solve. It has caused new debate and new information to come to light, the JEDEC for example is important. Fnagaton 18:56, 24 April 2007 (UTC)
Since the this guideline was adopted, as far as I know, not a single major computer manufacture, print encyclopedia, print dictionary, newspaper or book publisher has adopted the IEC prefixes. Microsoft and Apple have both released major new operating systems that do not use them. If anything, the neologism argument has gotten stronger. --agr 03:38, 25 April 2007 (UTC)
Hm... If it's operating systems you like, how about GNU coretools and the Linux kernel? Two major components of a major OS. In any case, this is still the common usage argument; it's nothing new. -- mattb 03:48, 25 April 2007 (UTC)
As far as you know, eh? Well that's certainly a convincing argument. — Omegatron 22:09, 28 April 2007 (UTC)
agr is correct and I note the lack of hard evidence to refute those claims, I do however note a snide comment which isn't at all persuasive. Also GNU and Linux are far from being standardised to only use binary prefixes. Fnagaton 09:02, 29 April 2007 (UTC)

More binary prefix changes

Well it looks like Sarenne has gone back to making binary prefix changes where none of the sources use binary prefixes. Unless the user stops and reverts the changes until this can be discussed and a proper vote can be taken I think it is fair to put a comment on each article talk page referencing the change and pointing people to this page so that it can be discussed in a central place and a vote can be worked towards. I'll wait a few hours though... Fnagaton 16:51, 24 April 2007 (UTC)

If you want to put a comment on each page where there are binary prefixes, you have some work to do... Sarenne 17:11, 24 April 2007 (UTC)
Fnagaton, you are mistaken if you believe that you can demand for others to stop invoking this guideline until a poll is taken. Like it or not, no consensus for change has (yet) been demonstrated. -- mattb 17:34, 24 April 2007 (UTC)
I think you're wrong about the consensus as demonstrated by "SWTPC6800 05:16, 10 April 2007 (UTC)". You're also wrong to say it's a demand since I didn't use the word "demand" anywhere, however what it actually is is fair warning of what will happen because this topic needs to be discussed and not just by a minority but by the contributing editors of those articles that have been changed. Fnagaton 18:00, 24 April 2007 (UTC)
Following an editor around warning others about their changes is borderline stalking, and I personally think it's talk page spamming. However, I'll let others weigh in on the appropriateness of what you propose; I won't interfere either way. -- mattb 18:07, 24 April 2007 (UTC)
Stalking? What a load of rubbish. Large scale changes are being made to articles and they should be discussed and since there is little to be gained from going around and around in multiple talk pages then the proper place for it to be discussed is here. You had better direct those comments to the idiot "person" who started spamming my inbox and calling my phone in the middle of the night just because they disagreed with me and wanted to use binary prefixes. It's probably the same "person" who started using a proxy to make anonymous changes to the Commodore 64 article. Fnagaton 18:17, 24 April 2007 (UTC)
If someone is calling your phone in the middle of the night to harass you, especially if they've been threatening, that's probably gone beyond the point of what we can deal with here-I'd frankly advise you to call the police. However, that doesn't necessarily mean what you're doing is acceptable, either, just because what someone else is doing is orders of magnitude worse. Seraphimblade Talk to me 19:39, 24 April 2007 (UTC)
You're right, it does over step the mark but it wasn't threatening, more like a squeaky voiced rant from someone who needs to grow up, I actually laughed which seemed to make them splutter even more. However I'm pretty sure someone writing here knows who has been using such tactics or at least suspects who has been using the anonymous proxy. I urge anyone who knows anything to come forward. Fnagaton 09:02, 25 April 2007 (UTC)

Look, there is a definite lack of consensus here. It isn't a matter of "consensus to change". It's a matter of a lack of consensus on what it should say. Keeping that in mind, deciding to unilaterally start changing all instances to binary was very much an act in bad faith. The discussion is currently ongoing. There is no consensus either way. Bladestorm 19:46, 24 April 2007 (UTC)

That's really not how that works. The existing guideline was overwhelmingly supported at the time of its inception. That doesn't mean we can never change it, indeed, such are always open to change. However, before doing that, there should be an equally strong consensus that changing it is the way to go. Especially in a case like this, where the issue is not just stylistic but also of accuracy, it's very important that a uniform style be used throughout the articles the guideline covers. That's exactly the purpose of MOS guidelines, to encourage and promote consistency. Seraphimblade Talk to me 19:56, 24 April 2007 (UTC)
The consensus was doubtful on January 29, 2007. User_talk:Sjenkins7000 ---- SWTPC6800 22:03, 24 April 2007 (UTC)
I cannot see where the bad faith would be. The current guidline is clear. Acting on it cannot be bad faith. That some discussion is ongoing is no reason to suspend the guideline. −Woodstone 19:53, 24 April 2007 (UTC)
Edit: Seraphimblade and Woodstone beat me to the point... I'm afraid I must disagree (with Bladestorm). It does take a consensus to change a guideline that was originally adopted by obvious consensus. You can't just pick any random controversial guideline (there are many) and assert that it shouldn't be followed until it is affirmed to everybody's satisfaction. Consensus does not mean that everybody agrees with the end result, and this sort of guideline will NEVER become uncontroversial no matter how many circular discussions and polls are held.
I acknowledge that consensus may change on this issue, but lack of consensus to change the guideline does not invalidate it, even while discussions are still in progress. On the contrary, for the time being, this is still a guideline and should be respected until it is changed. I do, however, submit the suggestion to all parties (Sarenne included) to refrain from making changes related to or revert war over this guideline right now. However, on the flip side, this discussion has gone on for weeks (months, perhaps) without any progress being made, and I don't find it reasonable to assert that the guideline is eternally in limbo just because we haven't been able to come to an agreement as regards a poll. -- mattb 20:04, 24 April 2007 (UTC)
If that's the common belief, then I retract my accusation (anyone is free to 'strike' it out if you like). However, I would still say that it's "bad form" to make sweeping edits based on a disputed issue, when you're an active participant of that conflict. Still, I suppose I wasn't really being very helpful, eh? So I think I may have a bit of a stop-gap solution. Bladestorm 20:13, 24 April 2007 (UTC)
Looking at the most recent figures there does seem to be a consensus to change though. Fnagaton 09:05, 25 April 2007 (UTC)
I agree that now is probably not a good time to be enacting changes based on this guideline en masse and have offered this advice several times. -- mattb 20:17, 24 April 2007 (UTC)
Now is precisely the best time because the guideline has been tried and serious issues with it have been shown. Fnagaton 09:05, 25 April 2007 (UTC)
I see no current consensus to change. (I see no current consensus at all, but no consensus always defaults to status quo.) Seraphimblade Talk to me 09:29, 25 April 2007 (UTC)