Wikipedia talk:Manual of Style (dates and numbers)/Archive B1

From Wikipedia, the free encyclopedia
Jump to: navigation, search
Binary prefix archives

Binary prefixes[edit]

I have tolerated them until I read a (wikipedia) article about Flash RAM that used Mebibits. I am now completely against the use of the "IEC", units and converting every instance of xB to xiB (as some users such as Sarenne are doing). If voting were still open, it would go to The MoS should discourage the use of IEC prefixes anywhere. Otherwise you're just empowering the Bobblewiks of the world. - Diceman 09:22, 23 February 2007 (UTC)

Also, if official sources give the numbers in KB/MB/GB, using IEC prefixes is original research. jgpTC 09:28, 23 February 2007 (UTC)
Converting measurement units from one type to another is NOT original research. If you think it is, you need to reread that policy. Raul654 03:28, 25 February 2007 (UTC)
Trivial interpretation or paraphrasing of sources is allowed, we paraphrase sources all the time. If HP calls their new laptop a "mobile workstation", we can still call it a laptop. Similarly, when one is talking about certain things in computing, it means XiB, period. XB may be used in marketing materials, but many things may be used in marketing materials that we should paraphrase. If a news report says that the only two people present somewhere were "an older man and a younger woman", it is not OR for us to do trivial math and say that two people were there. Seraphimblade Talk to me Please review me! 09:50, 23 February 2007 (UTC)
But... that's correct. Flash RAM is in powers of two, no? — Omegatron 15:07, 23 February 2007 (UTC)
I don't see any reason why it shouldn't be revisted if the guidelines are clearly confusing, or result in confusing articles. Which they do. — Arthur Rubin | (talk) 15:56, 23 February 2007 (UTC)
No they don't. Can you provide some evidence? There have been many votes and discussions about this and the consensus is always the same. That's why we put it in the MoS. "When there are disagreements, they are resolved through polite discussion and negotiation, in an attempt to develop a consensus. If we find that a particular consensus happens often, we write it down as a guideline, to save people the time having to discuss the same principles over and over." — Omegatron 16:52, 23 February 2007 (UTC)
Have to agree with Omegatron. We've seen almost every possible argument for and against the IEC prefixes, so short of some new compelling evidence that they cause people undue duress, I still wholly support their use in appropriate situations. As far as I can tell, the main argument against the prefixes so far is that some editors don't like them and that they aren't common usage. The original research argument doesn't hold any water. This is trivial interpretation that any intelligent and competent editor can adequately perform. We are merely talking about the difference in the reader having to interpret what "MB" means in context, or the article editor having to discern where it is appropriate to use (the SI meaning of) "mega" and "mebi", respectively. I'd prefer the latter since the article editor will (usually) be more competent than the lay reader. -- mattb @ 2007-02-23T18:06Z
I've seen every possible argument for the IEC prefixes, and still see no reason why they should be used, either generally or on Wikipedia. The person reading the article typically does not know what a "mebibyte" is; Wikipedia is not for propagating and enforcing IEC standards. —Centrxtalk • 19:25, 23 February 2007 (UTC)
And I agree with Diceman, specifically when talking about articles on vintage technology. Especially when all original production material for the product in quesion, and all cited reference material in the entry don't use these "new standards". "Appropriate to use" is a very open ended term, especially the way its being written in the current MOS, which is seen by some as giving them carte blanche to go through every single article and change every single use of Mib, etc. As far as merely "the reader having to interpret", I think that statement doesn't hold water. I've seen entire passages in entries, having been cut or riddled with reference citation requests based on the fact that the reader here is seen as someone who *can't* easily discern and must be lead around. --Marty Goldberg 18:53, 23 February 2007 (UTC)
Okay, I'll put a fine point on it. "Appropriate to use" means "appropriate in any context where the IEC prefixes are more accurate than the common (incorrect) usage of SI prefixes." 4 GiB RAM is appropriate, 372.5 GiB hard drive (as opposed to 400 GB) is probably not appropriate. The argument of "stick with tradition" has no value in my opinion. To me this makes as much sense as saying that we shouldn't prefer the usage of SI standards in American English articles because imperial units are by far more common in American English. Anyway, all these arguments have been posed before. As Omegatron suggested, I'd like to see a handful of places where the usage of the IEC prefixes has caused undue confusion. I think you'll find that most of the conflict over this issue is in the form of editors who dislike the prefixes clashing with editors who like them. The MOSNUM currently suggests that the first usage of IEC binary prefixes be linked so the curious reader can quickly read about their definition and the reasons for their existence. -- mattb @ 2007-02-23T19:17Z
The common usage is not incorrect. —Centrxtalk • 19:31, 23 February 2007 (UTC)
It is just incoherent... When you use 512 MB for RAM and 500 MB for hard drive in the same article, the two "MB" don't have the same meaning so at least one of the "MB" is incorrect.Sarenne 19:43, 23 February 2007 (UTC)
Sorry, how about "non-standard use of SI" or "usage of SI prefixes not endorsed by BIPM" or "incorrect insofar as SI is concerned". However you wish to call it. -- mattb @ 2007-02-23T19:35Z
And this is where you'll find I'm still in disagreement. That's a strawman argument calling for examples of confusion as a basis for change, when Wikipedia has no such capacity to provide that in the first place. You'd have to poll all readers (as opposed to editors) of a page on their thoughts of confusion. As someone who writes in the video game and computer industry (and specifically in historical aspects), its my professional opinion that an entry that is on historical themed subject *is* made more confusing by these. Especially when references, paraphrases, and quotes within the entry all use "older" standard. It'd be a different matter if it was a more recently released product as was mentioned above - I totally agree with you on that context. Oh, but you'd find us on opposite sides on the SI standards issues as well. ;) --Marty Goldberg 19:38, 23 February 2007 (UTC)
The BIPM is not an authority on the English language. If you want to include a mention of mebibits in the International System of Units article, fine, but that does not mean that these prefixes will be understood by most readers, or are anything beyond alternative uses along the lines of American and British spelling. —Centrxtalk • 19:41, 23 February 2007 (UTC)
(edit conflict x many)
In the context of memory, it's likely to be more accurate; but the terms are still not in common use, even among technically literate people and organizations. And, "mebibyte" is unpronouncable by a large number of people (I can probably find a source for that, if you want, but anecdotal evidence suggests that 10-20% of my co-workers cannot pronounce it); even if the abbreviations is acceptable and appropriate, the term themselves may not be.
Furthermore, have any other standards organizations picked up on the terms? There's been enough time since introduction for some other standards organizations to pick up on it if they find it appropriate. If they don't, we probably shouldn't. If that's the case, it's technical jargon which should be avoided when possible; even if it may be more accurate. — Arthur Rubin | (talk) 19:43, 23 February 2007 (UTC)
I see it has been accepted by some standards organizations, but it's still unpronoucable. — Arthur Rubin | (talk) 19:49, 23 February 2007 (UTC)
It makes little difference if something is picked up by standards organizations but is not picked up by people or non-standards publication. The IEC making a decree is not going to change the fact that your co-worker or computer science PhD is going to rightly laugh in incredulity if you use the term. Though, "MiB" is a much more reasonable construction than the word "mebibyte", which appears to have been invented by someone with no concept of language or speech. I wonder if you could find even an IEC engineer who uses the word in his everyday speech. —Centrxtalk • 22:37, 23 February 2007 (UTC)

I'd also like to point out that patronizing users for applying MOS style guidelines is misdirected agression at best, harassment at worst. There's no need to pick on Sarenne when you know the proper place to take up your issue with a style guideline (here). Revert warring is also totally inappropriate behavior. The MOS currently endorses IEC binary prefixes and states that they are to be kept if added to an article. Until such a time that this guideline changes due to consensus here, you should not engage in edit warring or use reverting to make a point. [1], [2], [3], [4]-- mattb @ 2007-02-23T19:22Z
Actually, I think it's very fair to pick on Sarenne as he/she is intent on enforcing the use of Binary prefixes over the entire Wikipedia as far as I can see. - Diceman 15:44, 3 March 2007 (UTC)
I do what is recommanded, over the entire Wikipedia if it is necessary. I do not "enforce" anything, I just try to make Wikipedia more accurate and coherent and I'm tired to see that some people are reverting these changes. Sarenne 19:07, 3 March 2007 (UTC)
This isn't the user's talk page; you can leave your denunciations out of the Manual of Style talk page. —Centrxtalk • 19:30, 23 February 2007 (UTC)
I could, but I won't. I find it relevant enough to mention here, I am very sorry if you disagree. Since this has come up before and it seems to have come up again, in the context of semiconductor memory, the binary prefixes are nearly universally appropriate. It simply does not make sense to produce RAM arrays in multiples of non-binary powers due to the binary addressing used by microprocessors. If an oldish computer was advertised with 32 KB/kB RAM, you can bet your buttons that 32×210 bytes was the quantity meant, not 32×103 bytes. -- mattb @ 2007-02-23T19:35Z
Your justification has nothing to do with your above comment, which was about revert warring. —Centrxtalk • 19:38, 23 February 2007 (UTC)
You misunderstand, the latter comment wasn't intended as a justification, just an additional tidbit to add to the discussion. Frankly I don't feel the need to justify my revert warring comment to you. -- mattb @ 2007-02-23T19:39Z
Then you should remove it. —Centrxtalk • 19:42, 23 February 2007 (UTC)
No. I don't feel I said anything that is out of line with rules or policy. I could construct a lengthy response to the above objections, covering territory we've been over probably five times or more. Frankly, I'm tired of debating this every other month. If you feel that there's enough consensus to change the policy, I'm totally agreeable to another vote. The experienced editors here have already seen every significant argument in this debate and have their respective opinions. Another vote will be a lot quicker than rehashing all the arguments again. :) -- mattb @ 2007-02-23T19:46Z
I looked a some of the previous voting archives on the matter as well, and a lot of it seemed inconclusive. Even a majority of those that voted for the IEC use still mentioned their own personal opinions on exceptions or stipulations regarding concurrent useage of the two. --mattb @ 2007-02-23T19:46Z
The very fact that kibi-, mebi-, etc., are not widely accepted, as Omegatron's first link plainly shows, means that it's not the proper term. Made-up words aren't words just because people say they are (as is the case with made-up non-words like xe). dcandeto 09:07, 11 March 2007 (UTC)

I'm still not seeing any evidence that anything has changed. A few editors have a personal vendetta against it, as they always have. So what? These units are still to be used wherever appropriate. "Appropriate" means "where the actual size is important and the actual size is a power of two". Things like "hundreds of megabytes" don't need to be changed, but things like "a 512 MB Flash RAM chip" should be.

To me this makes as much sense as saying that we shouldn't prefer the usage of SI standards in American English articles because imperial units are by far more common in American English.

Exactly. Accuracy and uniformity trump jargon and tradition; this is an encyclopedia. — Omegatron 21:28, 23 February 2007 (UTC)

And I'm still not seeing any evidence to the contrary of myself and others posting their views here. This *is* an encyclopedia, not a technical reference manual. It requires research and revision based on content and subject, not blind application. If I have an article who all source materials, quotes, referenced materials, etc. uses a specific older standard, simply editing that out creates more confusion not accuracy. Simply brushing it off as a "personal vendetta" is irresponsible, and the same claim could be made for those who do blind edits to promote that useage. It would be much more accurate to allow for content that uses both but describes one as being more accurate to current standards and revisions. And I'm not talking across the board on this either, as I said, I agree with you for newer devices. But when I'm writing within the context of a historical device, the useage and conventions of the time are equally as important from a historical accuracy standpoint. You can't be accurate in one respect and inaccurate in another. --Marty Goldberg 21:35, 23 February 2007 (UTC)
The usage and conventions of the time were incoherentinconsistent. Like I said in your talk page, the same MB unit was used to mean 1000*1000 B (hard drive capacities) and 1024*1024 B (RAM, ROM, cache...). Even if you don't want to use MiB within the context of a "historical device", the articles are innacurate (or incoherent). The easiest way to make them accurate is to change the "memory" MB to MiB. Sarenne 22:22, 23 February 2007 (UTC)
That seems pretty coherent. And the articles would also be more precise if we changed words in quotations, or invented other words defined through some explicit logic, but this is the English-language Wikipedia, not the IEC-language Wikipedia or the Derrida-language Wikipedia. —Centrxtalk • 22:44, 23 February 2007 (UTC)
Once again, that has nothing to do with historical accuracy. Ancient cultures saying the sun god blotted out the sun, when its more "coherent" to say its an eclipse, doesn't make it any less of importance for useage or inclusion. You just denote the difference and way, but both still must be reported. It *was* the standard, and is historically accurate to represent it as such. Once again, when talking about the context of historical content you can't be accurate in one respects and innacurate in another. Circular reasoning again. --Marty Goldberg 22:28, 23 February 2007 (UTC)
And fact remains that the sun is blotted out. It is a vagueness, not an incoherence. —Centrxtalk • 22:39, 23 February 2007 (UTC)
Actually by "incoherent" I meant "inconsistent".
So, what *was* the standard : 1 MB = ....... B ? Sarenne 22:42, 23 February 2007 (UTC)
Actually, if everyone is using it at the time, that's pretty consistent. As stated below, 1 MB was stated as 1024KB or "About a 1000 kilobytes". --Marty Goldberg 22:56, 23 February 2007 (UTC)
"About a 1000 kilobytes" is not accurate. 1 MiB is. Sarenne 23:07, 23 February 2007 (UTC)
Again, being circular and repeating something you already stated above which has nothing to do with the current conversation. This is why I gave up with the talk page discussion. You just asked what the standard was then, and I gave it. It was accurate for the used definition at the time. Mib didn't exist, its a new standard. People agreed "its about 1000KB" or "1024KB" as the definition, that was the definition of accuracy at the time. The IEC decided recently to redefine it so that MB actually refered to as 1000 (and more closeland whipped up a new wording of "bibytes" to compensate. --Marty Goldberg 00:17, 24 February 2007 (UTC)
You didn't give *the* standard. "About a 1000 kilobytes" has never been accurate and has never been a standard. There were 2 different standards (one official, the other was just a (bad) convention) at the *same* time : 1 MB = 1000*1000 B and 1 MB = 1024*1024 B. These incompatible standards were used at the same time for different purpose. It is inconsistent to use both standards in the same article. MB cannot accurately mean 1024*1024 B and 1000*1000 B in the same article of an encyclopedia. By the way, IEC never redefined MB. Sarenne 00:40, 24 February 2007 (UTC)
Yes, I *did*. Per the IEEE-CS computer terms - "MB Megabyte, 1024 Kilobytes". Likewise, the IEC did redefine it. Per the Wiki entry for Megabyte: "In the past few years, standards and government authorities including IEC, IEEE, EU, and NIST, have addressed this ambiguity by promoting the use of megabyte to describe strictly 1000² bytes". Again, anything to run around in circles instead of keep the conversation in context. --Marty Goldberg 00:57, 24 February 2007 (UTC)
Can you provide an actual standard that defines the term megabyte thusly? Or are you just using common usage again (nobody here is trying to argue what common usage is). IEEE 1541 is IEEE's official endorsement of the IEC binary prefixes. Is there a similar prior endorsement of the <SI prefix> + "byte" construction to denote powers of two? -- mattb @ 2007-02-24T01:23Z
Huh? Nobody is stating anything contrary to 1541. The question asked was a) Was there an old standard (i.e. before that), and b) What was it. The IEEE-CS terminology I provided was directly from them, an actualy listing of their definition from the past. --Marty Goldberg 01:28, 24 February 2007 (UTC)
This is exactly what I meant to ask, can you provide a reference showing that IEEE ever officially endorsed the common usage of "megabyte"? Simply using it in that context doesn't imply endorsement, while we have IEEE 1541 to show us that the IEEE clearly endorses these new prefixes in appropriate computing contexts. IEEE-CS is just one of the many IEEE societies, not a standard or a statement of official endorsement of prefixes. You made the assertion that the binary power usage of "megabyte" et al was an "industry standard", which is something more than just common usage. I'm asking if you can show some support for this beyond saying "the IEEE has used the terms in this context before". I'm not meaning to be antagonistic, but merely to point out that there is a difference between common usage and a standard.-- mattb @ 2007-02-24T01:55Z'
I'm not going to bother looking because it doesn't really matter. Wikipedia is not an implementation of standards documents. It is the English-language Wikipedia, and the OED is quite fine for that purpose. —Centrxtalk • 18:18, 24 February 2007 (UTC)
IEEE-CS is not IEC, "promote" is not "redefine", 1024 is not 1000 and "About a 1000 kilobytes" is not a standard for a unit. You seem to have an issue with accuracy but if a contributor wants to make Wikipedia more accurate or more consistent, you shouldn't stop him just because you think that Atari articles must match exactly what used to be Atari brochures or what used to be the common usage 20 years ago. If you want a snapshot, quote. Sarenne 01:34, 24 February 2007 (UTC)
1)"IEEE-CS is not IEC", what does that have to do with anything? Once again, the question was a)Was there a standard before and b) what was it. 2) If there's a definition before and a newer definition is promoted, that's a redefinition. 3) That's correct, 1024 is not a 1000 which is why Megabyte or MB was defined as "1024KB or 'About a Thousand'". 3) You seem to have an issue with context, historical accuracy, and circular reasoning and etc. The simple fact that you call "historical accuracy" a "snapshot" betrays this, if not a bit condescending. And even more circular, if someone is interested in accurately portraying industry wide useage of an item (such as Atari consoles and computer, Amiga, etc. etc.) from 20 - 25 years ago in an entry that is about said product from 20-25 years ago, you shouldn't stop them. And once again, I'm not saying the SI standards shouldn't be used in any of these articles, I'm saying that the older standards should not be wiped out completely from the entries in favor of it. For someone who keeps pushing consistency, the idea of historical consistency with said brochures, articles, manuals, design documentation, and general references that are used in the entry, seems to escape you as well.--Marty Goldberg 01:48, 24 February 2007 (UTC)
What you call "historical accuracy" is an inaccuracy and inconsistency. Wikipedia doesn't need to be accurate with brochures, articles, manuals, design documentation if they are inconsistent or inaccurate. If you want to quote Atari brochures to "accurately portray industry wide useage", i will not stop you... but you should quote it, not paraphrase it. Sarenne 02:09, 24 February 2007 (UTC)
And here we go around in a circle again. It is inaccurate compared to the current standard, not inaccurate for the standard that was used at the time of these. And then you asked a) Was there a standard and b) What was it. And we go again...... --Marty Goldberg 02:14, 24 February 2007 (UTC)
"1024KB or 'About a Thousand'" has never been accurate nor "historicaly accurate", and never will be ! It is inaccurate by itself. Sarenne 02:24, 24 February 2007 (UTC)
Once again proving you're not understanding the context of "historicaly accurate" or even the context of the word "accurate" being presented. If it was used as the industry standard then, used in said material, said subject, etc. etc. etc., in order to maintain historical accuracy you must present it as such to maintain that accuracy. If I have an exhibit in a museum about a caveman, I don't give him a ferrari, that would be historically inaccurate and defeat the purpose of the display. If I'm presenting an entry on a historical topic (i.e. vintage computers, video game systems, etc.) I don't completely wipe out all reference to what was actually used then, otherwise its "historically inaccurate". Or is that context still not understood? --Marty Goldberg 02:32, 24 February 2007 (UTC)
You want to maintain an "historical accuracy" and not about Ataris but about industrial "standards". It is not the purpose of an encyclopedia. That's exactly what I said : you want a snapshot, so *quote*. Sarenne 02:45, 24 February 2007 (UTC)
Of course historical accuracy is a purpose of the encyclopedia. You're giving a complete and accurate (in all respects, including historical) representation of the subject matter. Its exactly what I said: you don't understand the context, so *quote*. But please, keep adding more and more accusations about what I mean and what's being said, so we can go around in circles even more. --Marty Goldberg 02:54, 24 February 2007 (UTC)
We don't have to reproduce errors, inconsistencies and inaccuracies just because it would be "historicaly accurate". Wikipedia is not an encyclopedia of 1980. EOT Sarenne 02:57, 24 February 2007 (UTC)
It is however an encyclopedia of subject matter from 1980, which has an appropriate context and historical accuracy. Wikipedia is an encyclopedia, not a technical manual. Its job is to discuss and present the topic/entry as a whole, even if the topics are now considered technicaly innacurate. But please, please mean the EOT. Or are you going to post something else and I can come back and edit out the EOT because its now an error and innacurate? (Though you did say it, so it would still be historically accurate) --Marty Goldberg 03:02, 24 February 2007 (UTC)
It *was* the standard
No. It wasn't. — Omegatron 22:31, 23 February 2007 (UTC)
Yes, it was the industry standard at the time. I'd be happy to provide technical document references, plans, brochures, and other material that used it across the board. The common explinations at the time in text books, the press, etc. etc. that MB was actually 1024k or "about a million kilobytes" as was also used. --Marty Goldberg 22:38, 23 February 2007 (UTC)
Then make sure the first use of MiB is blue-linked, and discuss the old usage in that article. There, now we've got our historical accuracy, without saying in our voice that the gods really are blotting out the sun. Seraphimblade Talk to me Please review me! 22:34, 23 February 2007 (UTC)
Not quite, because it still leaves the confusion when quotes and referenced material do not. And no need for emotional sarcasm. --Marty Goldberg 22:38, 23 February 2007 (UTC)
If the use of it is in the prose, then it would be "mebibyte", not "MiB", which should be confined only to infoboxes and other places where space is tight. On that, links should be made according to context, not for every dictionary word that readers might not know. Unless the article is talking about storage standards or the IEC, "mebibyte" should not be linked. If a word is so obscure and novel that it must be linked in order for even the most literate person to understand, then it does not belong there as a used word. Wikipedia is not a vehicle for propagating marginal jargon. —Centrxtalk • 22:51, 23 February 2007 (UTC)
That's a silly line of reasoning. By that logic the Newton (SI unit of force) is too obscure and jargony to use because it is often linked in articles for the benefit of the reader. Even the most literate person in the world may well have no idea what the Newton is if they've had no significant exposure to physics (or have simply forgotten). Likewise, I wouldn't expect someone unfamiliar with computers to know what "ISA" or "KB" or indeed "KiB" mean. We link practically anything relevant in Wikipedia articles. Am I mistaken in believing that your argument tends towards "megabyte is part of the English language and mebibyte isn't"? (per your earlier comment about this being the "English language Wikipedia). I encourage you to note how many SI units, which are accepted for usage on Wikipedia, are based on non-English names. The Ampere, the Becquerel, the Coulomb, the Ohm, the Sievert, the Volt, the Pascal... All from non-English names. I realize you simply don't like the sound of the IEC binary units, but frankly I've never particularly liked the nomenclature "one millisiemens" (ending 's' is correct usage), but it's blessed SI, and I'd rather use a consistent standard than make up my own mind about "what sounds better". -- mattb @ 2007-02-24T00:47Z
One major difference: any physicist, even anyone who has taken a single physics course, knows what the Newton is and uses it as such. Computer scientists, computer programmers, computer users, do not use "mebibyte"; i.e. it is not even the term used in the field. Ampere, ohm, etc. are all used in English and were used in English more than 100 years prior to the IEC decree. They are all standard in the field. The sound of these words merely shows that they were inventions of engineers rather ignorant of language and its usage, and that they will not gain acceptance. Even if they are going to be accepted, they are not currently and it is not Wikipedia's place to use them when they are not used in the normal language. You personally are free to decide that you will use words handed down to you from on high, but Wikipedia is not prescriptive. —Centrxtalk • 18:30, 24 February 2007 (UTC)
See also e.g. French Republican Calendar. —Centrxtalk • 18:48, 24 February 2007 (UTC)
Well, several rather relevant standards bodies and professional organizations have adopted them, and a significant number (though a minority) of computer users and experts also use them. There are a handful of SI units that are NOT in common usage, even by experts (especially consider the mess surrounding units associated with ionizing radiation), yet we recommend adherence to SI because it is a recognized standard and helps keep articles consistent and unambiguous. The IEC prefixes solve an abiguity problem and (in my opinion) work for Wikipedia's purposes. I still don't understand why you are so passionate about deriding the construction of the words themselves, but I don't personally consider a slight aesthetical displeasure even slightly relevant. I do, however, encourage you to submit a proposal to a major standards organization detailing your own preferred construction of unambiguous binary prefixes which are not "ignorant of language and its usage". -- mattb @ 2007-02-24T20:47Z
Actually, there isn't a single book on Google Books that uses "mebibyte" or "mebibytes" as a word. There are four that have it listed in dictionary form. In contrast, there are hundreds of books that use "megabyte" as word, many of them published after 1998. I found one book, published 2001, that uses "gibibytes", with reference to an appendix because the term is so unfamiliar. In contrast, pick any SI unit and you will find hundreds or thousands of books that use the term as words. Even "millisiemens" has 20 times more books that even include the word than any of the binary prefixes. If there is some standard unit other than the SI unit for measuring ionizing radiation, that also should be used in favor of or alongside an unused SI unit. —Centrxtalk • 23:46, 24 February 2007 (UTC)
I see no reason to state that we shouldn't use an accurate technical term rather than an inaccurate but commonly-misused one, just because some people might not understand it. I would venture a guess that most people don't know what a joule is and would be very confused to hear a person talk about a mole of hydrogen-what do small burrowing creatures have to do with hydrogen anyway?? But if we're referring to something measured in moles, or joules, we should use that term. (And if we use moles, or joules, we should link to the appropriate page, specifically since many readers may come across those terms and want more information about them.) Similarly, if we're referring to something measured in kibibytes, or mebibytes, that is the term we should use. Do you think people only come here to read about things they already know? Seraphimblade Talk to me Please review me! 01:45, 24 February 2007 (UTC)
The joule and the mole are all commonly used in their fields, and are known by anyone who has taken even a high school science course. Wikipedia is not the place for introducing new language. Wikipedia articles do consist of things that the person already knows; that is prerequisite for the person learning new things, and articles do not suddenly "teach" a person something totally unrelated to what the article is about. If someone wants to learn about these new unit names, they can read the article about SI or IEC, etc. but if someone is reading about the iMac they should not be confronted with terminology that even the computer-savvy layman is not familiar with. —Centrxtalk • 18:47, 24 February 2007 (UTC)
Something additional I noticed, though this may be a bit off-topic... It is extremely common usage to denote binary capacities with the value and unit not separated by a space. On Wikipedia, we endorse the SI standard of placing a space between value and unit. I'm rather perplexed why nobody ever seems to bring this up. If we're so terribly concerned about common usage, wouldn't we ignore SI in this regard and write data capacities without a space between the value and unit? You don't have to venture far to find 100GB, 512MB, etc used in various forms of computer-related documents. Anyway, as Omegatron pointed out, this issue has always had its staunch supporters and fierce critics. There are definitely valid points on both sides of the fence, but my opinion remains unchanged and would be reflected in any potential vote. -- mattb @ 2007-02-24T01:27Z
It's possible we should, but there is a major difference in the two issues: the space does not change the meaning at all; it is purely formal, there is no danger of misrepresenting any source material, and the reader is unlikely to be perplexed or to even notice it. —Centrxtalk • 18:55, 24 February 2007 (UTC)

Yes, it was the industry standard at the time.

It certainly was not. Please provide evidence of this standardization. "MB" has meant 1,000×1,000, 1,000×1,024, and 1,024×1,024 bytes in different contexts and different times, and was never standardized. M- as a prefix has meant the same thing for hundreds of years, on the other hand.
Please read through the previous, endless, debates. This has already been discussed countless times, and you clearly have nothing new to bring to the discussion. Here's a start:

It most certainly was at one time. As stated, went through the archive already. I also already provided one instance of the IEEE using that as a 1024 standardization in its glossary. Likewise, as I stated it was an industry wide standard during the time period (1970's through 1980's), and most certainly for the historically entries items I was discussing about. The other inerpretations you mentioned came later, as the articles state as well. Once again, it was used across the board in documentation, promotional material, design specifications, etc. of the time. I also have several instructional text books from across that time spectrum where that definition is presented as the standard as well. Its not that there's nothing new being presented, its that its nothing you want to accept or listen to. Once again, very clearly, I'm not debating on its current use. I'm stating that previous definitions should be co-existing in entries on vintage hardware/topics so that historical accuracy is maintained. Especially within the context of cited references, reference material, or the like. As stated, I write in that area professionaly (historical analysis and exposition of vintage computer and video game topics), and I can't emphasize enough how important that is to maintain. --Marty Goldberg 03:45, 24 February 2007 (UTC)
As I tried to point out above, mention in a glossary does not constitute a standard in any way. It's acknowledgement of common usage at best. Also, I think it's even more confusing to have a policy endorsing the usage of IEC prefixes only for articles on "new" (by some arbitrary and as yet undefined distinction) topics. I don't see why we should have to use less accurate units in deference to historical usage. You don't see the article on the Great Sphinx of Giza reporting all its measures in cubits. -- mattb @ 2007-02-24T03:58Z
Once again, did not say only "new". I stated they should co-exist on "old". And yes, if I'm citing a passage, quotation, or reference of an original source that used cubits, cubits are indead mentioned. --Marty Goldberg 04:05, 24 February 2007 (UTC)
I don't believe anybody has suggested changing source texts. What does "co-existence" mean? Something like "The Nintendo included 2 KB (KiB) main memory"? That sort of thing is even worse than not using the IEC prefixes at all. I don't see any compelling reason at all to avoid usage of IEC prefixes in historical contexts; either the prefixes should or should not be used universally. The entire point of these things is for consistency and unambiguity. Picking and choosing historical contexts in which they should be used defeats the whole purpose. -- mattb @ 2007-02-24T04:11Z
Actually, yes, Sarenne had in his rants on my talk page, which is what prompted me to get involved in the topic here. He's also been going around making sweeping changes to every single appearance of MB/KB/etc. on every page, when the IEC/IEEE/Etc. is promoting those to still remain but cover 1000/etc. now. Unless he's gone and researched every single topic he's doing that in, the problem is there's the claimed confusion as to what a standard on MB meant. How does he know that he's not changing a 1000 item, that's supposed to remain as MB to an iB? And no, I wasn't suggesting the 2KB (KiB) thing, I agree that would look terrible throughout the text. But I think some kind of compromise can be worked out (which I've been more than willing to do), possibly using KiB in the text and the original specs in a more limited manner (infobox, or?) I'm sure there's something we can come up with working together. --Marty Goldberg 04:46, 24 February 2007 (UTC)
Yes, I must testify that people making mass formatting changes without reading or understanding the articles is liable to and does change the meaning. Separate from any issue of whichever form is most appropriate, this should not be done. Blind consistency produces error. —Centrxtalk • 18:59, 24 February 2007 (UTC)
As a rule of thumb, semiconductor memory contexts use multiples of binary powers and hard disks use multiples of decimal powers for reasons of addressing and marketing. Anyone familiar with computer architecture can verify this. Some other contexts aren't as cut-and-dry, but those two are pretty safe. -- mattb @ 2007-02-24T04:50Z
Why not present a short direct quote, say from a technical manual, if it's so absolutely imperative that "xB" be presented? Certainly, we must present direct quotes exactly as they were stated, as opposed to paraphrasing. Seraphimblade Talk to me Please review me! 05:08, 24 February 2007 (UTC)
That's kind of what I was thinking, possibly just in the system/machine specs box or initial presentation of system specs, having one xB/iB and using iB through the rest of the entry. It kind of prompts the reader (who is more commonly used to the xB reference) to question why the iB and click through on both of them (to learn about the new standards vs. old, which is what you want a reader here to do - learn), yet still keeps the historical accuracy and consistency with the support material present. --Marty Goldberg 17:09, 24 February 2007 (UTC)
How do you know that any {K,M,G}B reference, when used to refer to cache or memory sizes, isn't a 1024 item? Binary-coded decimal computers went away a long time ago, and it makes no sense whatsoever to cut short the memory capacity of a machine with binary addressing (I'm not sure how, for example, you'd limit a 1024*1024-byte binary machine to 1000*1000 bytes in hardware without adding a bunch of useless extra hardware or a pointless software hack). Can we please not waste any time whatsoever considering the possibility that any post-IBM 1400 series or Honeywell 200-series machine has a memory or cache size that's a multiple of 10? Guy Harris 08:18, 24 February 2007 (UTC)
Thanks for bringing up that example, it was kind of my point actually with the paragraph above. You have to be somewhat familiar with the hardware in question and what was actually being refered to before making a call on that sort of edit, especially with vintage hardware. The current set of "guidelines" and the way it was being interperpreted by some, would not have allowed for the BCD on some pre-IBM1400/Honeywell 200 series machines for example. Especially when doing almost robot like edits of every occurance of xB. --Marty Goldberg 17:09, 24 February 2007 (UTC)
Isn't there still a problem with the listed storage spaces not being in binary? If the box says "1 kilobyte", how do you know it's not really 1000 bytes or 1023 bytes without researching the issue rather than automatically changing it? Many times these terms are purposefully used in the vague sense, and where changing the unit is then simply wrong. —Centrxtalk • 23:50, 24 February 2007 (UTC)
If kilobyte/megabyte/gigabyte/terabyte is used to refer to a quantity of semiconductor memory (main memory, cache memory, flash, etc.), I know it's really 1024/1048576/1073741824/1099511627776 bytes because that's the way addressing hardware for that type of memory works. (Decimal computers were designed before semiconductor memory; machines such as the IBM 1400 series machines used core.)
For disk memory, the term is vague (for one thing, a sector typically has a power-of-2 number of bytes in it, so an "80 gigabyte" disk is unlikely to have 80000000000 bytes of data space on it). However, I don't think you'll find anybody advocating the use of the IEC terms for disk memory. Guy Harris 03:14, 25 February 2007 (UTC)
"The way it was being interpreted by some"? Could you put that in more concrete terms? I haven't seen Sarenne or others editing articles about computers from the 50s and 60s to use IEC binary prefixes. Your contentions have seemed to center around game consoles from the 80s, which as microcomputers, certainly use binary addressing. You're right that some familiarity with hardware is necessary to interpret this; are you familiar enough with all the hardware in question to mention what variety of microprocessor they used? As has been mentioned a few times before, most computers since the 1960s and very close to all microcomputers use binary number representation and binary memory addressing. -- mattb @ 2007-02-24T17:30Z
Actually, no, my contentions haven't seemed to focus around that. I said I deal in vintage computers *and* video game consoles, Sarenne kept bringing up and focusing on "Atari". And yes, I'm familiar with microprocessor varieties (or lack thereof for older computers). --Marty Goldberg 17:59, 24 February 2007 (UTC)
Okay, then where's the problem? I'm just asking for some specific examples for us to mull over. -- mattb @ 2007-02-24T18:07Z
I never said they did, you're missing the context again Matt, please reread. I simply said the way its being interpreted by some (and across the board robotic like edits) would not have allowed for for it, i.e. if they continue in the same manner, combined with the general way the guides are written, that would not allow for those older systems to retain their proper identifications. Unless there's some familiarity with the architecture, etc. Once again, when Sarene blindly states "Sources use decimal prefixes where the sizes are in fact binary. This has nothing to do with "vintage technology", yet that indeed isn't the case with vintage 50's and 60's computers, and then he's further calling for every single appearance of xB to be changed to SI and using the guidelines to justify it ("Yes, we *should* change every single 'old' reference... that's exactly why the MoS says that the use of the new standards should be accepted."), that's a pretty broad stroke and one has to assume he means it given the majority of his contributions listed have been iB edits. (And to clarify, I'm not knocking him for iB edits, so don't take it that way). I'm sure you'll come back with some response to brush this off, but this is important, or I wouldn't have brought it up. You think I like going through 2 full days of discussion? ;) --Marty Goldberg 18:38, 24 February 2007 (UTC)
I'm not calling for every single appearance of xB to be changed but, like the MoS states, I think we should accept every change xB => xiB when xB is beeing used in a binary sense, even if it is vintage technology, except if it is quoted material. If I change an xB that is actually used in a decimal sense then it's a mistake. Sarenne 18:53, 24 February 2007 (UTC)
You were previously calling for every single one to be changed, and stating it had nothing to do with vintage tech. I'm glad to see you revise your position then. It still needs to be clarified in the guideline then as far as this aspect. The other (single appearance of xB) is still being discussed above. --Marty Goldberg 19:02, 24 February 2007 (UTC)
You just misunderstood what I said. Yes, it has nothing to do with vintage technology and yes, every single 'old' xB (i.e. xB used in a binary sense) should be changed. That's exactly what the MoS says: "The use of the new binary prefix standards in the Wikipedia is not required, but is recommended for use in all articles where binary capacities are used." It's pretty clear, isn't it ? Sarenne 19:43, 24 February 2007 (UTC)
Serenne, I'm not picking up where we left off yesterday, I refuse to go around in circles like you like to. There was no missunderstanding, you said it in response to very specific statements by me that it shouldn't be across the board with vintage technology. Likewise in repsonse to my statement that (as we went through yesterday) for historical accuracy and continuity its not clear, and not accurate to do it across all references and documented material used in an article. And then I had to clarify for you what historical accuracy means, just as I previously clarified what "not required" and "should" vs. "must" means. Just as what's clear: "The following rules do not claim to be the last word on Wikipedia style." and "These are not rigid laws: they are principles that many editors have found to work well in most circumstances, but which should be applied with flexibility.". Hence the discussion a few paragraphs up about how to keep the historical accuracy for cited and referenced material while still using iB's as well. We're trying to reach a compromise here. If you have nothing concrete to add to coming up with a compromise and solution on that, then move on and please don't turn this in to another pissing contest full of sarcasm and claims about me, while repeating the same things over and over. --Marty Goldberg 20:14, 24 February 2007 (UTC)
I confirm that you misenderstood what I said, in a discussion you deleted. Your clarifications are useless, I know what you want and what you mean and I know what I said... you don't. You are going in circles by yourself.Sarenne 20:36, 24 February 2007 (UTC)
Sarenne, you can keep telling your self that if it makes you happy. And the discussion was here as well. Furthermore, it takes two to tango, even if one is just repeating the same thing over and over with sarcastic jibes while the other's trying to reason, discuss, and compromise. As I stated in the one on the talk page that I removed, you just can't keep from going on and on about the same thing, ignoring requests to stop, and adding more claims about me. Case proven. If you want to contribute, contribute. If you want to just keep posting sarcastic things to get the last word in, you'll find you're alone in this as I'm more interested in the discussions/debates I was having with the other above. EOD with you on this for me. But please, do another of your responses and prove it further. --Marty Goldberg 20:53, 24 February 2007 (UTC)
I don't know how you can find a compromise and discuss something if you don't understand what I'm saying. I'm forced to repeat myself because when you paraphrased what I supposedly said, you totaly altered what I had really meant. For example, you said that I revised my position... well I didn't... so you obviously misunderstood what I said. Sarenne 21:12, 24 February 2007 (UTC)
I agree that quoted texts should not be altered. However, I would not agree to applying this to specifications. That is, if your source says "32KB RAM", a Wikipedia article that mentions this figure should should read "32 KiB RAM". Only blocks of text that are directly quoted passages should be unaltered. -- mattb @ 2007-02-24T20:47Z
Yes, mine was more in to regards that if sources cited, referenced, etc. it should appear once in one of the previously suggested formats. Creating a citation with a link to a source is similar to a quote in functionality. You use quotes when its feasible due to size or overly important in the flow of the page, otherwise you use citations. I'm not suggesting litering the page with sometimes xB and sometimes xM (though that is what happens with the quote rule). Once again, "where it is important to do so" is pretty open ended and open to interpretation unless its further clarified. I believe my position on Hist. Ac. is important for example. --Marty Goldberg 21:01, 24 February 2007 (UTC)
Trying to get back on track, does anyone else want to propose additional or alternative phrasing to the current MOS? If it's no different from the original text proposed in this section, I'm afraid I still must wholeheartedly object. -- mattb @ 2007-02-24T20:47Z
Actually, I do have an alteration proposal: All instances of IEC units should be linked unless that unit is actually taken from a WP:RS for the article. The precise phrasing I suggest is to change from
However, because they are less familiar, binary unit prefixes such as MiB should be linked at least once per article to avoid confusion. Link as [[Mebibyte|MiB]] to avoid a disambiguation page.


However, because they are less familiar, binary unit prefixes such as MiB should be linked at least once per article to avoid confusion. All instances of binary unit prefixes which are translated from incorrect usage of SI prefixes should be linked. Link as [[Mebibyte|MiB]] to avoid a disambiguation page.
Also, although this should not be part of this proposal, but the individual articles should note that Wikipedia may be the most common medium in which these units are used the units are not yet generally used. — Arthur Rubin | (talk) 21:09, 24 February 2007 (UTC)
Sounds like an interesting alternative as well. So I'm clear, are you saying to change all over to IEC but keep a link to SI or keep SI and like to IEC's with an explination? --Marty Goldberg 21:18, 24 February 2007 (UTC)
Change (incorrect usages of) SI units outside of quotes to IEC, but link all of them. If Kibibyte, etc., are merged to Binary prefixes, then I think we'd have to leave a soft redirect in place, rather than a hard redirect, or the link would be unnecessarily confusing. (I haven't checked when the merge tag was added, or if those proposals should be still active.) — Arthur Rubin | (talk) 21:34, 24 February 2007 (UTC)
I really dislike linking all of any term, and I think it goes against the recommendation of some policies. However, it's a reasonable compromise that I'm willing to agree to if you think it will significantly improve things. -- mattb @ 2007-02-24T23:08Z
That just exposes how backwards the use of these terms on Wikipedia is. Wikipedia is not the place to implement standards that are unknown to general readers. If and when this terminology because common use, even in the field, then Wikipedia should use these terms. Plain dictionary words should not be linked in Wikipedia articles, and this style guide should not contradict WP:MOSLINKS and WP:CONTEXT. If the linking is necessary in order for people to understand what is referred to, then these terms should not be used, or they should be used alongside their more common equivalents. —Centrxtalk • 00:00, 25 February 2007 (UTC)
So do you have another proposal? I think it's obvious that totally revoking the MOS endorsement of the IEC binary prefixes isn't likely to happen and I don't think any amount of dialog will sway anyone involved in this discussion. However, I'd like to come to a somewhat agreeable compromise since this issue keeps coming up. Forgive me if I seem terse at this point, but we're really just rehashing all the issues that have been discussed before and I'd like to see some proposed compromises that we can work on. -- mattb @ 2007-02-25T03:58Z
Coming in here without being involved in all the history of this, I have to say that the comment "I don't think any amount of dialog will sway anyone involved" means we have a real problem here. I think you'll find that there are a lot of editors who would feel that "MiB"/etc are weird, and would disagree with the current MoS on this if asked. I certainly would have if I'd had any idea such a decision was even being contemplated seriously. Note that the people here in the discussion on both sides have come here because they're involved (or were involved in the original decision). Yes, WIkipedia isn't a democracy - but it is based on consensus, and it's also has a higher goal which is to provide a usable encylopedia to the world. The goal of wikipedia is not to enforce or promote a particular standards body and its standards, and in this case, by pushing this standard so hard it is hurting the main goal of Wikipedia - providing an encyclopedia. To exaggerate, it's the equivalent of writing it in Esperanto "because it's a universal standard that everyone should use". (Yes, that used to be an argument for Esperanto.)

I would have no problem if these IEC prefixes were used in Wikipedia once even a significant minority of the computer industry started using them. I took a poll of (non-Wikipedia-editing) coworkers, mostly seasoned engineers, a few new PhD's and Master's. Out of around 20 (other than me), only one had ever even seen KiB/MiB/etc - and that was in an Wikipedia article (and I've only seen it here, too). And they had no idea what it meant when they saw it, and thought it was silly.

Can a group of editors who devoutly believe that everyone should use these new prefixes force the issue, and keep the requirement? Probably - Wikipedia rewards those who are persistent and care about an issue, whether it's a good idea or not, so long as there's a critical mass. But in the end I think this truly hurts Wikipedia. For the readers (remember them? The ones we're building this for?) it's like reading an article on land use where all the distances were in furlongs, and all the areas were in jugerum. — jesup 16:07, 26 February 2007 (UTC)

You can write in Esperanto, actually, if you'd like. But that's rather a nonissue here, this is the English Wikipedia. It is not the "non-binary-prefix" Wikipedia, so use of binary prefixes doesn't contradict, you know, our very name. The land comparison is also poor. Most people who had a reasonable base of knowledge (especially if they're on the metric system), but no special knowledge about computers, could be asked what a megabyte is. Well, they know what mega means, and they probably have some idea that a "byte" is a unit of something, so they'd say "A million bytes, of course!" If you're asking about their hard drive, they're right. But if you're asking about their RAM card, processor cache, any of that...they're wrong! Good guess, but wrong. For that reader (remember them? The ones we're building this for?), we owe it to them to be accurate and specific. We don't repeat urban legends, for example, just because it might be disconcerting to a reader to find out that some "common knowledge" thing is a total crock. If our verifiable information is that it's a crock, that's what goes in the article. And in this case, the idea that solid-state memory operates in powers-of-ten multiples such as kilo, mega, and giga is a crock. Those prefixes have an ancient, well-known meaning, and 2^10, 2^20, and 2^30 isn't it. So, we provide our readers accurate information. That is exactly why this is advocated. Your friends who are engineers already know that when you refer to a "gigabyte of RAM", you're not really referring to a billion bytes. But most of our readers do not. Using megabyte instead of mebibyte is not like using feet instead of furlongs, but having an accurate value either way. It's like saying "one yard" when the actual distance is really "one meter". It's close, but not quite accurate. Seraphimblade Talk to me Please review me! 16:21, 26 February 2007 (UTC)
I believe the proposal to link IEC terms has merit, because it attempts to address the central problems, unfamiliarity and ambiguity. Saying that all usages of one particular term should be linked is, of course, excessive. But the use of any term where it must be interpreted should be either disambiguated or the ambiguity referenced.
It seems obvious to me that, unless one wants to preserve ambiguity, ambiguous terms need to be defined by links. If a quote uses MB, even if 'we' know which measurement was intended, 'MB' should be linked to some article that notes that the historical usage of MB included at times a strained extension of 'mega-', and that depending on context (which we can leave up to the reader to discern), 'MB' may mean the SI 1000x1000 or the computerese 1024x1024 (which would be more correctly defined in modern usage as MiB or mebiByte, and which links to those articles as an introduction). (whew) Ambiguity is the enemy here, and in every resource material. At least give the reader evidence that a closer look is necessary. The first use of any of 'MB', 'MiB', 'megabyte', or 'mebiByte' should be linked.
This satisfies the objection that one should not change source material nor quotes. But we can certainly reference that old usages were merely conveniences when differentiating standards didn't exist. It also satisfies the objections to the use of new terms which themselves seem to be strained. Whether 'mebibyte' is pronounceable won't matter as the linked definition will be in text. ;-) But we can't expect the reader to replicate the thought processes of IEC and other standards groups without help.
Does an article already exist that clearly notes the conflicts between the stopgap terms and the intended meanings? Does the article already note the usage contexts (such as cache vs. disk)? Shenme 22:30, 25 February 2007 (UTC)

I took a poll of (non-Wikipedia-editing) coworkers, mostly seasoned engineers

Did you poll the readers? That's who you claim to care about. Most readers don't know (or care) that kilobyte sometimes means 1024 bytes.

it's like reading an article on land use where all the distances were in furlongs, and all the areas were in jugerum.

Other way around. You're arguing in favor of furlongs while the rest of us are trying to use the metric system.

I believe the proposal to link IEC terms has merit

What proposal? It has always been part of the MoS to link the IEC terms.

Does an article already exist that clearly notes the conflicts between the stopgap terms and the intended meanings?

Binary prefix?

If and when this terminology because common use, even in the field, then Wikipedia should use these terms.

Then what the hell should we use in the meantime? Do you dispute that it's necessary to be accurate when referring to units? If you agree that it's necessary to be accurate, then how do we do it? Do you want to go around changing every instance of "100 MB" into "100 decimal MB"? That would be far more disruptive than just using standardized unambiguous units. So some readers aren't familiar with them. So what? They can read an article about it just as easily as they can read an article about the traditional units. Most readers aren't familiar with the ambiguous definitions of those, either. — Omegatron 16:38, 26 February 2007 (UTC)
What proposal? It has always been part of the MoS to link the IEC terms. Actually, that's not the case Omegatron. It became a part of the MoS when you put it there on 9 July 2005. And please don't claim what readers can and can't do, you can't speak for them any more than you're claiming we're not. --Marty Goldberg 18:35, 26 February 2007 (UTC)
I have to say that the comment "I don't think any amount of dialog will sway anyone involved" means we have a real problem here.
Sorry for speaking the truth. I only meant to say that the people who watch this page have seen this issue and all the arguments both ways several times now. Most of us see some validity in each others' points of view, but have our strong opinions regardless. Re-hashing the same arguments again won't change these opinions. That's all I intended to convey. -- mattb @ 2007-02-26T17:14Z
Matt, I understood what you meant as well, though it did come off a bit more authoritative as the other person mentioned. Either way, my intent in all this was and is still not to stop IEC from being used or to completely change the MoS entry (though I do appreciate the clarifications that were added during the couse of this discussion). It is to illustrate that striking IS from an article is not as cut and dry as the suggested MoS binary entry (and the MoS is only a suggestion by defintion) dictates. Just as the MoS states "should be applied with flexibility". Those statements are there to allow for differences of opinions, no matter how strong they are. --Marty Goldberg 18:35, 26 February 2007 (UTC)

I do not know why this repeated discussion this time turned so aggressive and sometimes outright silly; please calm down everybody! In most cases where it applies this really comes down to adding an i and a link to Binary prefix. This improves accuracy a bit, which is a good thing. Many readers will read over it without noticing it at all, because they are unaware of the issue and do not need to be bothered with it too much. Authors have in general more knowledge about the subject at hand than do readers, so they shall bear the duty to decide whether a source used decimal or binary values. (For an example compare the max. capacities quoted for 5¼" and 3½" diskettes, HDDs, CDs, DVDs and FMCs.) This has nothing to do with altering or augmenting quotes, which are very rare here anyway. In the English Wikipedia in particular we have to disambiguate use of units all the time, e.g. land vs. nautical miles, US vs. Imperial and wet vs. dry capacities, Avoirdupois vs. Troy weights, long vs. short vs. metric tons, pound-force vs. pound-mass, local vs. US dollars and so on; or, on a similar matter, Julian vs. Gregorian dates. Much of this could be overcome if we settled on the metric system and other international standards (like ISO 4217 for currencies), though. It is also questionable whether megabit is any more an English word than mebibit is. (Note the etymology of bit, which is similar to that of the binary prefixes.) Christoph Päper 09:30, 27 February 2007 (UTC)

Don't forget about kibinibbles! — Omegatron 00:39, 1 March 2007 (UTC)

Wikipedia is no place for the MiB religion to be practiced[edit]

Rman2000's concerns are articulated on the following pages: Wikipedia:Administrators' noticeboard#my edits keep going away or User talk:rman2000 and User talk:Sarenne#I am not a vandal. -- mattb @ 2007-02-28T23:57Z

So here we are now what? Is this a better forum for the topic? Rman2000 20:59, 28 February 2007 (UTC)

Well, I'm not sure we need the whole text dump, but yes this is a more appropriate forum. That being said, though, we really have been over all this again and again, and the consensus was to leave the MOS as it was, due to the fact that binary prefixes more accurately reflect the numerical values involved. Seraphimblade Talk to me Please review me! 21:07, 28 February 2007 (UTC)
Never mind that the terminology is obscure and NOT used by the technical community at large. Never mind that less information and more confusion results in the reader's mind. Never mind that you "enforce" a style that doesn't say must or can't with an iron fist. What you need is a pulpit and a congregation for this cause; it is a misuse of Wikipedia's openness and is nothing so much as censorship. Rman2000 21:27, 28 February 2007 (UTC)
Well...if you're going to use polemic, I strongly doubt that it will be more effective for change then the previous well-reasoned discussions. The MOS addresses the issue of possible confusion by stating that the first instance of XiB in any article should be linked to Binary prefixes. Most readers unfamiliar with the discrepancy will likely simply read XiB as XB and never notice the difference, for the astute ones that do, they can click the link and find out what's going on! Commonality of use similarly isn't a main consideration, "battery acid" may be a more common term for sulfuric acid, but we still call it sulfuric acid as that is the proper and precise technical term. Seraphimblade Talk to me Please review me! 21:34, 28 February 2007 (UTC)
Oh, dear. Promoting the standardized use of binary prefixes instead of the older and more wide-spread (albeit erroneous) use of regular prefixes is equal to censorship? Sounds a bit harsh. When will the correct use ever propagate out to "the masses" if not even encyclopedias and references in general can use it? --Strangnet 21:42, 28 February 2007 (UTC)
First, thank you for taking this discussion to the appropriate place. That being said, your concerns have been brought up many times before and don't change anything in my mind. If you want to read detailed responses to your reasoning, I encourage you to read the above discussion and the handful of previous discussions linked therein. Myself and others have provided our point of view on this many times, so I'm not going to simply copy/paste my responses here. Edit: Here are a couple of the previous discussions for your consideration: [5] [6] [7] [8] Please don't mistake my brevity as ignoring you, it's just that we've been over your exact arguments so many times that I'd rather let the archives speak for me. -- mattb @ 2007-02-28T23:40Z
Also, I would humbly suggest to Sarenne and others to be very slow to label people as vandals for changing IEC binary prefixes. Try to explain the situation, direct them to this page, do everything you can before using warning templates. This is a hot issue and should be treated with as much civility and good faith as possible. -- mattb @ 2007-02-28T23:57Z
I usually use templates once I advised the contributor to come here and to look at the MoS. Changing once these prefixes is of course not vandalism but I cannot find another word (maybe "anarchist", "rebels" ?) to label those people who keep reverting these changes knowing that they are acting against current consensus. For me, consensus is one of the major policy of wikipedia and must be followed. I would never make a change to the encyclopedia once I know it's against the consensus. If we let people acting against consensus, it's just anarchy and this anarchy begins with infinite edit wars.Sarenne 12:23, 1 March 2007 (UTC)
If we don't let people express themselves then consensus may as well be created by an elite and pushed down to the masses. If I could find enough people who agree that wikipedia should follow common usage of a term then I would become one of the "anarchists/rebels" that you refer to. And then your magic consensus stick would swing the other way leaving you as the "rebel". Wikipedia actually works so dont refer to infinite edit wars as if they occur because people do not think the same way as you do.
On an overall note, Why should wikipedia go against the common usage of the term just to save a few experts being confused. Non-computer science people have enough trouble differentiating their bytes from their gigabytes as it is without a new scheme. On the original research area, is wikipedia attempting to be the first major user of the term and is that not against the goal of simply being an aggregation of knowledge? Ansell 04:48, 26 March 2007 (UTC)
You misunderstood what I said. The "anarchists/rebels" are those who make changes in articles knowing that's against the consensus, not those who are against the consensus. Sarenne 12:13, 26 March 2007 (UTC)
Since binary prefixes have been standardized it's quite difficult to claim that the usage here is original research, don't you think? Since it is quite a difference between 1,000 and 1,024 and even worse between 1,000,000 and 1,073,741,824 don't you think it would be wrong for Wikipedia (and the general public) to not distinguish them? --Strangnet 05:03, 26 March 2007 (UTC)
Standards are made up by experts for their personal use. If non-experts think that experts use confusing terminology they simply ignore it. I think you mean 1,000,000,000 btw, but either way. I do not see it as a "conscience" issue, if the general public has not caught on to the new fad then why should it be used here. A rough idea of how many people actually use the term in common writing may be shown by the following which filters out the relevant "expert" usage of the term by about half. [9] Evidently it does not seem to have caught on. Much like this person says, the usage is "unlikely to catch on". Interestingly some fanatic is posting on a forum using the following [10] statement, which seems to imply a relationship between microsoft users and kilobyte users, which evidently shows how desperate a small group of people seem to be about having an "eternal wrong" righted.
In technical computer science articles which refer to a specific number, as opposed to the concept of a "gigabyte" of data, it seems reasonable to follow what the experts use, but the list of articles linking to kibibyte tells a completely different story. Why confuse a user looking at for example Celtic languages with a completely unfamiliar, and unuseful terminology? This prefix is not just being used for technical wordings, it is being used rampartly to enforce the POV of a small group about what everyone should use. Ansell 05:27, 26 March 2007 (UTC)
FYI: I took a look at several articles where kibibyte seemed to be "off-topic". In those articles, kibibyte was used in describing the size of a linked file. Often, this was done as part of the {{PDFlink}} template. --Kevinkor2 17:48, 26 March 2007 (UTC)

The great Wikipedia conspiracy to destroy Computer Science[edit]

From User_talk:Sarenne#I am not a vandal:

What I do know is that Wikipedia stands ALONE in a sea of billions of technical articles that use MB, KB etc by trying to use KiB and MiB etc.

Although it may seem pointless or pedantic to people when they're first exposed to these units, they are in fact used in actual technical contexts by actual technical people. I'd just like to kill this "no one uses them except Wikipedia!1" argument once and for all. Here are a few examples of technical documents that use IEC terminology to avoid ambiguity, selected at random from Google Scholar searches:

Just because Microsoft doesn't use these units, and a few vocal editors haven't been personally exposed to them, doesn't mean that they're not in use. — Omegatron 01:14, 1 March 2007 (UTC)

Though it is totally true that they aren't yet in widespread usage. Obviously you aren't trying to imply otherwise, but I'd like to nail this complaint before someone brings it up. Yes. We know. Our reasons for wanting to use them have nothing to do with how (un)common they are. -- mattb @ 2007-03-01T01:28Z
Yes. Our reasons are the same as NASA's.  :-) — Omegatron 01:38, 1 March 2007 (UTC)
And in fact I can go to the same sites (such as the IBM research, etc.) and find examples of IS still in use. --Marty Goldberg 01:30, 1 March 2007 (UTC)
In fact, some of the very documents cited above use e.g. "gigabyte". Others do not have "bib" at all, they only use the abbreviations which is much less problematic. —Centrxtalk • 02:35, 1 March 2007 (UTC)
Nobody has ever argued otherwise. Also, not meaning to cause offense, but it's SI, and what you are referring to as SI is actually an incorrect usage of SI (yes, the word "incorrect" is totally appropriate here since the prefixes are rigidly defined within SI), so it might be better to refer to this as "conventional usage". -- mattb @ 2007-03-01T01:35Z
Matt, it was a typo. Obviously meant SI, but you did this response before I got a change to go back and revise it. Regardless if its considered "incorrect" here, the point is the same said resources also contain SI still being used in a scholarly context. If it wasn't still acceptable by these scholoarly journals (such as IBM), it wouldn't have been allowed by the editors. Hence in the same veign to Omega's statement on what's still being used and in what setting. --Marty Goldberg 01:43, 1 March 2007 (UTC)
Who is in a better position to define acceptable usage of SI than the body who created and administers SI? Granted, plenty of people don't properly use SI, but there is no argument that the conventional usage of "KB, MB, GB", etc is incorrect inasmuch as SI is concerned. In other words, I find it ironic for you to refer to this as "SI" when it is actually something else. -- mattb @ 2007-03-01T01:53Z
Not only are they rigidly defined, the organization in charge of administrating SI explicitly says that it is incorrect usage. "These SI prefixes refer strictly to powers of 10. They should not be used to indicate powers of 2 (for example, one kilobit represents 1000 bits and not 1024 bits)." — Omegatron 01:38, 1 March 2007 (UTC)
In particular, they say it here. Guy Harris 02:24, 1 March 2007 (UTC)
Indeed. Now I feel I must spearhead another complaint before someone brings it up: Yes, we know that even BIPM's denial of conventional usage and its endorsement of IEC prefixes in a footnote of the latest SI standard don't magically invalidate conventional usage. Again, our position is based solely upon the merits of using the IEC prefixes, not what is more popular. The fact of actual popularity of the IEC units vs conventional usage has never been contested. -- mattb @ 2007-03-01T01:43Z
The correctness is in doubt, and the common usage prevails. —Centrxtalk • 02:34, 1 March 2007 (UTC)
Could you elaborate on "correctness"? Insofar as major standards organizations are concerned, there's no question as to correctness. Common usage I don't contest. -- mattb @ 2007-03-01T02:42Z
"Major standards organizations" do not dictate the English language. —Centrxtalk • 04:36, 1 March 2007 (UTC)
Nobody can dictate any language, but everyone influences it, some more and some less. (If you like you may compare the works of the Académie française with actual spoken French.) How did kilo and mega get into the English language by the way? Anyway, technical or scientific units are not much a matter of linguistics, but they are a matter of communication. Christoph Päper 11:33, 1 March 2007 (UTC)
I presume you don't mean that it's in doubt whether, for example, any binary machine claimed to have some number of "KB"/"MB"/"GB" of main memory or cache memory has, in fact, that number *1024/*1024*1024/*1024*1024*1024 bytes of main or cache memory, as nobody familiar with binary computer hardware would doubt that. Maybe an IBM 1401 with 2KB of main memory had 2*1000 bytes of main memory, but an IBM System/360 with 16KB of main memory had 16*1024 bytes of main memory. Guy Harris 02:44, 1 March 2007 (UTC)
No (though there will be problems with people semi-automatically changing the units on IBM 1401). The correctness or superior correctness of the new terminology is in doubt. The meaning of the common terms used for the past 50 years is quite fine, and superior to the novel inventions. Those common terms remain the most common terms, even with experts in the field, and that is therefore what Wikipedia naturally uses. The entire manual of style is merge of common usage on Wikipedia with the professional usage that can be found in formal printed publications. In this case, the common usage by the vulgar masses is the same as the common usage by the published experts, so that's what belongs in the manual of style. The use of this new terminology is rather pointless if they are not going to prevail and be established throughout the language. It is quite possible ten years hence that these new terms will still not be commonly used, and that the standards bodies will alter them, in which case we find that Wikipedia was not even nobly spearheading some revision of the English language, it was just being confusing to its readers during a historical anomaly. We cannot make that determination now; these are new standards and they are not common usage anywhere. —Centrxtalk • 04:50, 1 March 2007 (UTC)
Ah yes, because an extra 'i' is extremely confusing to the lay man... Okay, I see what you're after, but we aren't covering new ground here. It's just the same old "we shouldn't use the IEC prefixes because they aren't commonly used" argument. I understand the point of view, but will always disagree with it. -- mattb @ 2007-03-01T04:58Z
Well that's the point of view found in the Manual of Style. Wikipedia does not prescribe the use of the English language, and the Manual of Style cannot be at odds with what everyone will add to the encyclopedia. If layman, industry programmers, and computer science professors are all going to add text with "megabyte" instead of "mebibyte", if most changes to make an article conform to the MoS are going to be reversed as either gibberish or peculiar and novel, if the reader ultimately is either confused or thinks it's a spelling error (and perhaps fixes it), then the intended purpose of the style guide contradicts itself and will nevertheless not be achieved. The Manual of Style has no force except insofar as editors read it, agree with it, and implement it, and the Manual cannot counter-act the general tendency of average and expert editors alike. —Centrxtalk • 05:10, 1 March 2007 (UTC)
So we should just accept crappy spelling and incorrect grammar because it's "common usage" and the typical style of the people adding text to our articles? Of course not. Editors should be bold and encouraged to add their sloppy writing, as long as it's verifiable and encyclopedic, but if someone else wants to fix it and make it more accurate, the improvements are also welcomed and accepted. — Omegatron 23:42, 1 March 2007 (UTC)

Random aside musing: my own anecdotal observation leads me to believe that engineers are often a lot more inclined to embrace IEC prefixes than computer programmers. I would venture to guess that this has something to do with having had a lot of academic exposure to SI (through physics, electronics, etc), and often being required to follow it to the letter. Anyway, it's not extremely relevant to the discussion, just something I've noticed. -- mattb @ 2007-03-01T01:58Z

Most of the technical documentation I see these days uses mebibytes. Gwen Gale 12:36, 1 March 2007 (UTC)

Most of it that I see as well. (The conclusion above that using MiB is somehow not using mebibytes is ludicrous, it's the accepted abbreviation for that term, and I see it widely.) Even the SI, linked above, says Don't use these prefixes to represent binary, use the binary prefixes! There's your source. Use of "kilo", "mega", and the like, when one is representing binary values, is demonstrably incorrect. And no matter how widespread something is, if it's verifiably incorrect, we don't use it. If an urban legend is widely believed, but we have sources conclusively showing it's wrong, we put that it is wrong, despite how few people already know that. In this case, the SI prefixes are verifiably wrong. SI even says so! The binary prefixes are demonstrably correct. Therefore, urban legends and "common knowledge" aside, we go with verifiably correct information. Seraphimblade Talk to me Please review me! 13:24, 1 March 2007 (UTC)
I think that is nice that G Gale sees only XiB. And I read lots of technical articles and I never see XiB used....NEVER. The only place I have ever seen xB used where there is potential of confusion in the computing industry is in the size of mass storage devices in marketing literature--once known, it is hardly a point of confusion for the vast majority of folks. Rman2000 05:15, 3 March 2007 (UTC)
Again "...the accepted abbreviation..." by whom? Well, put it another way. It is also NOT the accepted abbreviation used by, well lets see: Intel, Microsoft, Oracle, Linksys, Micron (and a hundred more memory manufacturers), IBM, HP, Dell, Gateway, Tiawan motherboard manufacturers, PC Magazine, Anandtech, every hardware engineering company I know of, Circuit Cellar, etc etc etc. Someone listed a handfull of documents above as a counter example but the exceptions don't make the rule...I stand by my observation that Wikipedia stands alone in a sea of billions of technical articles, marketing statements, product descriptions, product packaging... Rman2000 05:15, 3 March 2007 (UTC)
I'm clueless as to what fuss is (I mean I've read the discussion). The unit's already changed in the industry so far as I can see, the mainstream'll follow and I thought WP was known for its helpful IT articles. Gwen Gale 15:53, 1 March 2007 (UTC)
It hasn't change. The IEC issued a decree that this is the new standard, and few follow it; i.e. the industry has to use it before the mainstream and Wikipedia do. —Centrxtalk • 23:26, 1 March 2007 (UTC)
I meant and shoulda said usage had changed. Sorry I wasn't clear. :) Gwen Gale 06:37, 2 March 2007 (UTC)
Agreeing with Gwen Gale, this whole thing looks like a sequel to the IAC's redefinition of the term "planet" with Pluto not passing the new criteria. How many people protested that? --Stratadrake 02:28, 2 March 2007 (UTC)
The problem I have with using kibibyte and other similar terms is when it is inconsistent with the original documentation describing a computer. For example the Commodore 64 documentation and magazines of the time when the computer was widely popular will all use 64KB and 64kilobyte terminology. As a test you should try using Google to search for "64 kilobyte commodore" and "64 kibibyte commodore". The search using kilobyte returns about twenty times as many references and most of the kibibyte references are copied from Wikipedia. So naturally the person reading their old manuals or remembering the terminology used and doing a web search will search for "64 kilobyte" and not find the Wikipedia article. This means Wikipedia is a less effective resource. There should be a style guide in Wikipedia regarding this issue and I think the style should be to keep using the terms that are mostly used in the industry, meaning the kilobyte etc. —The preceding unsigned comment was added by Fnagaton (talkcontribs) 17:56, 4 April 2007 (UTC).

"Should be kept"[edit]

When the current phrasing with regards to usage of IEC prefixes was adopted, we added the text "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 reasoning for adding this was to help prevent the IEC prefix debate from erupting on every single page where an editor disagreed with the prefixes (which was happening), instead being able to defer to the MOS and this talk page. I see nothing advantageous about re-hashing all of our arguments put forth here on every page that might use the IEC prefixes (except in those rare cases where there really is a valid debate over whether a quantity is correctly 2n or 10n). The objection has been used recently that the use of the word "should" allows for the possibility of removing IEC prefixes simply because the editor doesn't like their usage (again, this is not referring to the aforementioned cases where usage of IEC prefixes would actually indicate an incorrect quantity). This has led some (especially anonymous IP editors) to justify revert war behavior with comments like "the MOS doesn't require use of these prefixes". This is a bit of a sticky situation. Clearly the MOS cannot require any style, but it does represent the result of consensus and in my experience, reverting warring against MOS consensus is often grounds enough for a temporary block. For example, anyone reverting SI articles to Imperial or CGS units based purely on stylistic preferences would be asked to stop.

Well there is a point. How do you see "purely on stylistic preferences"? I have given more than sufficient argument for keeping xB in articles where the article was originally so written or where is makes logical sense based on the original source of the information presented. But those arguments are brushed aside with iron-fisted references to the MOS and they are reverted in the blink of an eye. Restore them again, and Sarenne and the XiB police are there in a flash to "correct" the binary sinners. Rman2000 06:10, 3 March 2007 (UTC)
You have not made a single point that hasn't been made here many times. While the reasoning may seem novel to you and you feel that it constitutes a "sufficient argument" for getting your way, the IEC policy still stands despite these same objections being brought up several times. I would apologize for being terse at times due to having to devote inordinate amounts of time to debating this, but I'd venture to say that this isn't your first foray into editing on Wikipedia nor your first exposure to this debate. Honestly, drop the act. -- mattb @ 2007-03-03T06:52Z

The reason I'm pointing this out is so we have something to refer to in the future when the argument is made that it's okay to revert valid changes to IEC prefixes simply because an editor objects to their usage on principle and the MOS uses the word "should". If an editor has an issue with this policy, they have every right to take it up here, but I don't believe it is acceptible to revert war or try and debate this on article talk pages and user talk pages. Putting our differences aside on whether the IEC prefixes should be used on Wikipedia, can we agree that debates over purely stylistic issues that are directly addressed by the MOS should be taken up only on the respective MOS talk page? It's enough work to debate this particular issue every couple of months on this page without having to see edit wars erupt because an editor feels that their opinion trumps the MOS. I really don't want to try and use stronger wording in this section because that could equally be misapplied by someone who doesn't understand the issues, but I think we need something concrete to point to when people wish to use the "MOS doesn't require" argument. Thoughts? -- mattb @ 2007-03-01T21:39Z

See my comment above. I disagree. I don't know whether it is an all article or none argument; I simply haven't read the entire Wikipedia. I think it is reasonable to make an argument on a specific page or pages that may or may not apply elsewhere. So, bringing the discussion here, to some degree, puts the discussion in a vacuum. You admit that the MOS is only a guideline; but you support its enforcement as iron-clad law. I think that obscures the discussion from the very readers of the pages that, more than likely, have some input to content. Rman2000 06:10, 3 March 2007 (UTC)
And I completely disagree with the wording "result of consesus". Looking back it looks like the same x ammount of people (or same group of you) voting for it, and new people every time on the other side, which by actual consensus would add to a lot more people against it. The simple fact that new people bring it up every month shows that its not an overall consensus. Just a consensus of who happened to know there was an argument and discussion going on at that time and just happened to be here for a vote. Likewise, its just as much work for "us editors" to see someone come breezing through making robot like edits just because they feel they can use the current version of the MoS as "binding law" that everyone must bow to. The MoS clearly states it is purely suggestive, and likewise should be applied flexibly. When its the "consesnus" of the regular editors of these entries not to do it, there is nothing in the MoS that states either one trumps the other. I see no flexibility in the areas I mentioned previously, which once again was not calling for non-use of the IEC. All I see is the same bunch using their own binary reasoning, which in its own accord violates what the MoS states on this issue. --Marty Goldberg 22:59, 1 March 2007 (UTC)
Looking back it looks like the same x ammount of people (or same group of you) voting for it, and new people every time on the other side
A good number of the "new people" you mention who are categorically against this don't actually have any clue about the issue involved and are simply annoyed that we've adopted something different than what they're used to seeing (disclaimer: no insult intended, but I can't think of any way to phrase this in a direct yet PC manner). Forgive me if I sound arrogant, but I don't give a lot of thought to comments from editors who can't demonstrate that they know the first thing about the nature of this issue, the reasons there are ambiguitities, or indeed, simple arithmetic and power notation. I'm certainly not classifying the entire opposition into this category, and there are valid points against using the IEC prefixes, but if it came to a vote again, I have no doubts that the policy would stand. In fact, in every discussion we've had, there have been "new" editors commenting in support the IEC policy as well, so I don't think that the appearance of different people on both sides of this issue is meaningful in itself.
WOW. "A good number of the "new poeple" you mention who are categorically against this don't actually have any clue..." Yes, those who dare to voice another opinion...they are the unwashed, the unbelievers. And this person already knows what a future vote would hold. This is so dismissive and incredibly unprofessional; does it sound in the least bit religious? Rman2000 06:10, 3 March 2007 (UTC)
Likewise, its just as much work for "us editors" to see someone come breezing through making robot like edits just because they feel they can use the current version of the MoS as "binding law" that everyone must bow to.
You have yet to give a concrete example of where this policy has been used to to apply the IEC prefixes incorrectly. If the "robot edits" are correct in their nature, what's the problem other than the fact that you personally dislike the IEC prefixes? What is the "work" it causes you? Debating here? That can hardly be helped... Of course the MOS is applied with flexibility, but only where there's a good reason to do so. If the justification for applying the IEC prefix policy "with flexibility" is against the entire reason for having the policy, this is in contention with the consensus that created the policy. Do you not see a problem with allowing this very debate to play out on every page where some anonymous editor just has a vendetta against the prefixes or is totally ignorant of the issue? (don't get me wrong, Marty, I'm not in any way lumping you into this category) I wouldn't even bring this issue up if it weren't for the recent rash of IP editors (quite likely only one or two people) who stage mass reverts justified by the same logic and goading remarks ("we can do this all day, you won't win", etc). -- mattb @ 2007-03-01T23:21Z
"...apply ... incorrectly". I have given several examples of where it is inconsistent with the very source of the information reflected in the WP page. The "work" is trying to make the page reflect the technical world from which is is sprung. I would think that it is the antithesis of the very nature of the "openness" of Wikipedia to revert every instantance of xB to XiB when the author feels technically justified in its use. But it is clear that you view anyone who uses xB, in place of Xib, as perverted, especially if they don't log on (even though, again, Wikipedia is supposed to be "open" to all; it is clear the IP logged changers are second class citizens). Rman2000 06:10, 3 March 2007 (UTC)
Just to put a finer point on the issue of "new editors", of the people involved in this discussion, only myself, Omegatron, and Christoph were involved in the original discussion of this policy (unless I've missed someone). As I said, there are "new" supporters on both sides. -- mattb @ 2007-03-01T23:31Z
I've known about "this issue" (confusion over 1000's vs 1024's) for around, oh, 19 years now. I used to code operating systems where we had to deal with the occasional confusion over this, and I used to maintain disk drivers when disk makers started advertising using SI units. Would things have been better if people had thought of this problem before co-opting K and M for 1024-based values? Yes. Does it matter now? Not really. Is the IEC tilting at windmills? Pretty much yes - in 6 years since they proposed this, about the only significant place to pick it up has been Wikipedia. As I'm sure has been mentioned before, it's a lot like metric time.
Your request above is a back-door way to try to present this (very heavily argued) guideline as policy, and give you and the others who troll Wikipedia looking for violating pages a touchstone to enforce your edits. While I don't want to start an edit-war, I and a lot of others obviously strongly disagree with what is on WP:MOSNUM. You present this as consensus; I can't see that from the discussions here now or in the past archives. I'm not saying that there isn't a group here that agree with you, or that you are the only ones arguing to keep this rule, but I don't think you can honestly say that there is consensus on this point. I think everything we see here shows that there is a pretty strong lack of consensus - lots of people pointedly on one side or the other, with very few switching sides.

Radical proposal - would you be ok with a page retaining K for 1024's if it was linked something like this: [[Binary prefixes#Binary prefixes using SI symbols|K]]? — jesup 04:50, 2 March 2007 (UTC)
You know, I absolutely hate it when people patronize others by citing policies they are obviously already aware of, but give me a break. There's nothing hidden or insidious in my comments, I merely ask for opinions with regards to something I percieve as a problem. I don't have a hidden agenda to promote little 'i's that are designed to cause the demise of the civilized world as we know it, I'm really just trying to find a way to cut down on these edit wars with IPs. After discussing this endlessly, I fully understand the various valid reasons for rejecting usage of the IEC prefixes, but thus far there have been significantly more people who support them. The characterization of myself "and others" as persons who "troll Wikipedia" is unnecessary and rather offensive, not to mention incorrect, and it would be lovely if you could make your point without ugly labels. -- mattb @ 2007-03-02T06:50Z

Six years is nothing. I began hearing talk about Pluto not being a planet at least ten years ago when I casually said there were nine planets and was gently admonished by a relative who works at CERN who said the "planet count" would sooner or later be dropped to eight. Meanwhile anyone who calls an editor a WP:TROLL for using standard IT notation should have a gander at WP:No personal attacks, for starters. Gwen Gale 08:00, 2 March 2007 (UTC)

My apologies for overstepping in my comments - The dangers of commenting when sleepy. The 'troll' comment was just because I see editors scanning wikipedia for pages that violate the guideline and fixing them, then (as you note) sometimes fighting the normal page editors to keep the KiB/etc edits in. This probably has been discussed before, but would it cause less consternation and conflict if you started by inserting a boilerplate block/tag/etc on the talk page and/or main page to notify the editors there of the issue and of the reasoning, before changing the page. Then come back to it a week or month later. If you're lucky and they agree with you and the reasoning, they'll have fixed it already. They may come here to dicuss it. And if they haven't at least they won't feel submarined by the change. In terms of social engineering, this is a method far less likely to provoke conflict, I think (and more likely to get people to accept the change, in fact).
I still think something like this ([[Binary prefixes#Binary prefixes using SI symbols|K]]) should be considered as an alternative (non-preferred if you like) on pages where the active editors disagree strongly, or where for historical reasons there are reasons to keep the original usage.
Hopefully this has been a more productive comment. — jesup 14:05, 2 March 2007 (UTC)
Trolling's a blockable thing, so it's a bit rash to use it lightly (or whatever) around here :) Anyway I do agree about the social engineering side. Putting a boilerplate tag on affected articles as a "warning" of things to come is a helpful notion IMHO, given this would give editors time to bone up on the definition and adapt their thinking to it. Gwen Gale 14:16, 2 March 2007 (UTC)
Actually "trolling isn't a bad word and I didn't read it as offensive -- just the truth. Make a change from XiB to XB and see how many "seconds" go by before the Sarenne bunch pounce on the page and change it back. And they will go edit for edit until the author tires of trying to see the page as they think it should be. You will get threaten with the MOS and immediate blockage if you don't leave the page as XiB. You can pretend in your own world that it isn't going on; but it is. And it makes no sense to the person writing the page. The MOS as WRITTEN does not say MUST or CAN'T. But the self appointed XiB police are there to "enforce" the guideline. Look at the Sarenne edits in . They are solely on enforcement of XiB. And Matt Britt's only edit of was to enforce XiB terminology. If you didn't have XiB police, you wouldn't be having XiB edit wars. Is this really so hard for everyone to understand? Rman2000 06:10, 3 March 2007 (UTC)
Ok but it's standard notation, after all. Gwen Gale 06:17, 3 March 2007 (UTC)
"Standard" - yes. Generally used - no (outside of Wikipedia). So "it's standard notation, after all" doesn't really answer the comments from Rman2000. His point (that there is an effective XiB police force enforcing a guideline) is factually correct. Have people considered my comment about reducing conflict over this issue using a tag/boilerplate first and see if the editors of a page agree to change it on their own (perhaps avoiding edit-wars and seriously ruffled feathers over this issue)? The people making these edits also need to realize it is not a policy, and it is not a MUST. It's a guideline, and right now people here are trying to enforce it as if it were a policy over the consensus of editors of some of these pages. — jesup 22:07, 17 March 2007 (UTC)
You know, the religious overtones you're throwing into this are doing nothing for me. It's impossible to take you seriously when you blow a content dispute to such a ridiculous proportion as to use religious imagery. I also find it ironic that you are chastizing others for reverting to their preferred page versions when most of the edits you have made with this account have done just that. In any case, this isn't about any individual editor, and I'd appreciate it if you'd stop with the persecution act. You can make your point without playing up the victim part; this issue has many supporters on both sides. -- mattb @ 2007-03-03T06:49Z

Another section break[edit]

Okay, we've gone around in circles with the same arguments being rehashed many, many times. The debate has gone past ad nauseum for me, and this is beginning to break down into a shouting war. Most of you have made your points well and have valid viewpoints that I respect, even if I disagree with them. Kudos to you, let's be friends despite our differences and all that jazz. However, there's really nothing I can say to Rman2000 that hasn't already been said, and his accusatory tone and persecution complex are two things that I have no desire whatsoever to deal with.

Therefore, I am personally finished with this round of debate. No new evidence or ideas have been brought forth in awhile. From what I can see, this issue is fairly split between established editors, and I don't think there's a consensus to change the current wording of the MOSNUM. If you disagree, that's fine, let's have a vote (though I really don't think the outcome of such a vote would satisfy anyone involved). I'm just totally sick of reiterating the same points over and over, so I'm done for now. -- mattb @ 2007-03-03T07:06Z

P.S. - Jesup, I'm sorry to have ignored your proposal. It got a little obscured in all the background noise. The entire point of this discussion centers around whether IEC prefixes should be used in all appropriate contexts. Personally, I don't see any reason that we should selectively apply the IEC style guideline whenever an editor on some page objects to using them. It's counter-productive and if I were to agree to that I might as well agree to removing the IEC style recommendation altogether. So while I appreciate your attempt to provide an amiable middle-ground solution, I can't say that I would view that as a "blessed alternative" to the current style recommendations. What I can agree with is that it's a good idea to link the first usage of any non-obvious unit in most articles, and I think there is sufficient confusion surrounding the exact definition of a "kilobyte" to merit its linking. My opinion is that no significant change to the current wording of the MOSNUM on the matter of binary prefixes is necessary, and for now I'm done explaining my reasons why I maintain this opinion. -- mattb @ 2007-03-03T07:15Z

Binary prefixes aren't used![edit]

The bottom line for me is that very few people reading articles about computers will have any idea what a mebibyte is. It isn't used even among computer people. The fact that thousands of usages of the prefixes exist is yet another example of the wiki model failing us. Don't bother arguing though--it would just be rehash of all the above. --PSzalapski 19:42, 28 March 2007 (UTC)

There are far better reasons that the wiki model fails than a few little 'i's. -- mattb @ 2007-03-28T19:43Z

A cunning plan[edit]

This problem, and the need for the IEC prefixes, arises because those experienced in computing "know" that k sometimes means 1024, and sometimes 1000, and it's part of their expertise to know when which usage is appropriate, so they see no need for KiB. The general public don't understand this, as was shown by lawsuit(s) complaining that hard drive disc sizes were shown by the computer as smaller than the number on the box. This could be addressed by two additional guidance points: firstly, have a standard infobox which should be added in articles where this issue arises, giving a very brief statement along the lines of "As units such as MB and GB can confusingly refer to binary or decimal numbers depending on context, the less common Binary prefixes are used for clarity". Secondly, wherever decimal usage of the units is shown, show the binary equivalent in brackets. Thus, "160 GB HDD (158.69 GiB)". Since I've not been assiduously following the debate, this may have been discussed earlier: if so, my apologies for repetition. .. dave souza, talk 09:08, 3 March 2007 (UTC)

I think this is a helpful notion, like the one about throwing in a boilerplate tag alerting readers to the difference. Truth is I have to read technical documentation almost every day and the standard notation is binary. Manufacturers have continued using decimal notation on packaging only because that's what consumers are familiar with. Hence MBs and MiBs are often muddled and I'd say WP articles should be helpful and in some standard, brief and pedagogical way, teach it. Gwen Gale 09:27, 3 March 2007 (UTC)
Such infoboxes as useful as they might seem on first thought are quickly removed, because they do not serve their goal well and they disturb the article. This happened for example to most if not all notifications for articles incorporating non-roman characters. I think normal links are enough.
You usually do not give more significant numbers for a converted value than for the source, so your example would be 159 GiB (or 149 GiB if calculated correctly). Anyhow, I think such conversions are not necessary in general. Most of the time there is one nice and round value in either unit that one should use.
My final words on the whole issue: The Manual of Style lists best practices rules. (At lest it should, some passages do not express this view.) Editors do not have to follow it slavishly, but all articles should comply to it after some time of maturation. That is what copy editing is for. One should be grateful for the people (wonks?) who do this boring work that improves Wikipedia rather than harassing them. Christoph Päper 17:41, 3 March 2007 (UTC)
Showing the binary equivalent in parentheses after the decimal form sounds useful to me. It is much like the successful practice of showing both English/American and metric measurements, with one in parentheses. I think that would be much better than forcing an obscure convention onto our readers without explanation. -- Donald Albury 12:29, 5 March 2007 (UTC)
The decimal form is used only for some storage capacities (Hard drives...) and nobody has forced or want to force only binary units in this case. Sarenne 12:43, 5 March 2007 (UTC)
I have been researching the use of the "new binary prefix" for three days now and I can't seem to find a mainstream web site (other then IEC and Wikipedia) that uses the xiB terminology.

It would seem to me that Wikipedia should be the best it can be and use old style until the new format can catch up to the currant times. Purgatory Fubar 22:34, 5 March 2007 (UTC)

Please accept my advance apologies if I have ventured inappropriately. I am a casual Wikipedia user, not a contributor, and this is my first post. If my comments are out of bounds, please immediately delete the entirety of my comments. PaulYShimada 10:35, 8 March 2007 (UTC)
Since its inception, I have admired what I assumed was the primary purpose of Wikipedia: "TO SERVE THE USERS OF WIKIPEDIA BY INFORMING, EDUCATING, AND DISPELLING MISINFORMATION." After reading the discussions regarding distinctions between KB<>KiB, MB<>MiB, GB<>GiB, and so on ("Talk:Kilobyte" and "Wikipedia talk:Manual of Style (dates and numbers), Binary prefixes"), I felt an urgent need to speak up for those who I believe make up the vast majority of typical USERS of Wikipedia. Many (if not most of us) want to LEARN and DISPEL MISINFORMATION! To a much lesser degree are we interested in learning the jargon, usage, conventions, and history of the myriad of disciplines touched upon by Wikipedia. Most of us are even less interested in the protocols or nuances of being encyclopediasts.
It IS IMPORTANT to reveal to common Wikipedia users the distinctions between KB<>KiB, MB<>MiB, GB<>GiB, and so on. While using Google to seek practical solutions to a common PC problem, "Delayed Write Failed," I encountered numerous uses of KiB, MiB, GiB, and so on; it is not true that "they are not currently and ... they are not used in the normal language." I also learned that it is likely many computer users are NOT AWARE of the MB<>MiB distinctions, and that lack of knowledge was a likely cause for misunderstanding and the resulted in several accusations regarding the dishonesty and insincerity of hard disk manufactures and the related lawsuit(s).
"I took a poll of (non-Wikipedia-editing) coworkers, mostly seasoned engineers... Did you poll the readers? That's who you claim to care about. Most readers don't know (or care) that kilobyte sometimes means 1024 bytes." Since I got DSL last month, I have been browsing the Web at least 12 hours daily, and I agree that "most ... don't know ..." However, I would suggest that if many intelligent posters did know about KB<>KiB, such knowledge would contribute to dispelling confusion and misinformation. Is that not high among the goals of Wikipedia? Wikipedia may "not be prescriptive," but should Wikipedia contribute to perpetuation of a long-standing confusion that can now be dispelled by recent nomenclature not available when the original ambiguity was born? "Wikipedia is not the place to implement standards that are unknown to general readers," but is it not an accepted Wikipedia function to "DISAMBIGUATE?"
"Appropriate use" is not justification for apparently misinforming those who are less knowledgeable than the learned Wikipedia contributors who wrote in "Talk:Kilobyte" and "Wikipedia talk:Manual of Style (dates and numbers), 10. Binary prefixes." I would guess that many typical computer and Wikipedia users are not similarly knowledgeable of so-called "appropriate use." One "Wikipedia:Talk" writer stated, "4 GiB RAM is appropriate, 372.5 GiB hard drive (as opposed to 400 GB) is probably not appropriate." But many typical PC users would have asked, "Why does Windows recognize only 372.5 GB? What happened to the missing 27.5 GB on my new 400 GB hard disk? Most common among explanations is CMOS BIOS incompatibility. Hopefully, a friend may have suggested that the answer could be found in Wikipedia. A final thrust at appropriate use--30 years ago, I bought static RAM memory boards for my S-100 computer that were advertised as having 65.5 K, instead of the normal 64 K of dynamic RAM. Was that advertisement a lie? It may have been inappropriate, and for a less knowledgeable buyer, the advertisement was certainly misleading. (For example, was the additional 1.5 K RAM due to it being static rather than dynamic?) I feel that 30 years is long enough to wait for widespread clarification of this ambiguity in nomenclature that has lead to numerous instances of bewilderment.
The primary issue for Wikipedia should be "TO SERVE THE USERS OF WIKIPEDIA BY INFORMING, EDUCATING, AND DISPELLING MISINFORMATION." It should NOT be an issue of who or which group is RIGHT(?), or has more standards organizations on their side, or has more bibliographic citations, or has the side of historical legacies, or who can better frame their side of the debate, or who has "personal vendetta against it," or who can better shame others with pettiness such as "which appears to have been invented by someone with no concept of language or speech," and so on. The ONLY ISSUE should be "TO SERVE THE WIKIPEDIA USERS!" To ignore the needs or to assume specific expertise among Wikipedia users would be illogical. Experts would not need to consult Wikipedia on this topic and could discern the contextual distinctions for interpreting GB<>GiB. The bibyte (XiB) nominclature may be recent and difficult to pronounce (try Wikipedia's "DISAMBIGUATION"), but should Wikipedia minimize or ignore new terminologies, new technologies, or new discoveries because they are unfamiliar or disagreeable?
As human beings trying to cope in an increasingly complex world, we sometimes rely on our personal preferences (prejudices) to ignore some among the myriad changes in our lives and the world. Galileo may have taught us about the scientific method and the value of "truth," but I feel that Aristotle set the tone for Wikipedia, "Mine is the first step and therefore a small one, though worked out with much thought and hard labor. You, my readers or hearers of my lectures, if you think I have done as much as can fairly be expected of an initial start. . . will acknowledge what I have achieved and will pardon what I have left for others to accomplish" []. Wikipedia's highest calling is to be a teacher, regardless of what is taught by others!
On the brighter side, I did find several threads among the Wikipedia "Talks" cited that sought to formalize approaches to convey clearer and more complete information (GB<>GiB). In that endeavor, I would leave the technique(s) to the expert encyclopediasts and technical editors. A list of techniques, such as blue hyperlinks, information boxes, parenthetical equivalents, and whatever else could only help to dispel misinformation and confusion. I hope that your good hearts and minds will soon focus on HOW, and not on IF. Some consistency demonstrates professionalism; lock-step betrays lack of understanding. However, the ONLY criterion for HOW should be, "TO SERVE THE USERS OF WIKIPEDIA BY INFORMING, EDUCATING, AND DISPELLING MISINFORMATION FOR THE INCREASED BENEFIT OF THE WIKIPEDIA USERS." PaulYShimada 10:35, 8 March 2007 (UTC)
I object to changing descriptions of pre-2000 computers, and quantities of memory less than one gigabyte, to the binary prefixes. This is inconsistent with how these machines were advertised and used and inconsistent with original references. The fact that fractional quantities are (in my experience) never used with kilobytes or megabytes shows the extra precision is spurious. --Wtshymanski 17:53, 26 March 2007 (UTC)
In hingsight it seems that such objections count for nothing given that rewriting articles for old system continues in earnest. I have to say, it's this sort of dogmatic pursuit of agendas over the conveyance of useful information that caused me to quit as a Wikipedia editor, and seeing the same people still forcing through the same old agendas on every other page I visit is likely to make me quit reading it too. I'm sure those pushing the agendas think what they're doing is terribly important and that tne ends justify the means, but from where I'm standing it's all rather pointless and annoying and ultimately it's just disruptive. I'm sorry, but I do view it as vandalism, and like any other vandals, I wish those responsible would turn their energies to something more constructive instead.-- 16:27, 26 April 2007 (UTC)