Wikipedia talk:Manual of Style/Dates and numbers/Archive 66

The below discussion titled "Converting to SI prefixes" continues from Archive 65.

<< go to Archive 65

Converting to SI prefixes (continued)
This discussion continues from Archive 65.


 * PC manufacturer's specifications are not "marketing drivel." They are legal documents carefully reviewed for accuracy. They don't make "we're the best" claims. If you have more authoritative published sources for information about particular PC models covered in Wikipedia articles, please cite them. MOS is a guideline, WP:OR is official policy. It's not subject to a vote.--agr 12:41, 26 January 2007 (UTC)
 * Legal documents ? Reviewed for accuracy by who ? Sarenne 13:29, 26 January 2007 (UTC)
 * You're joking, right? When they publish that the computer has 2 MB of cache and 512 MB of RAM they are accurate? I thought we've been through this. It is not accurate and wikipedia is not accurate either when using MB instead of binary prefixes despite it's in WP:MS already.--Pethr 13:39, 26 January 2007 (UTC)
 * Err... I can't even count how many misleading statements, exaggerations, and outright lies I've seen in marketing documents, especially those relating to consumer products.  Computers are not immune to this by any means.  If you want an example of some of the most nauseatingly self-serving marketing in existence, browse Apple's site for a few minutes.  Most of it isn't exactly untrue, but anyone can see the spin that they put on things.  That's the job of marketing, to spin things in the way that is most likely to make a sale to their target audience.  Consumer product marketing is NOT, by and large, concerned with publishing the most accurate specifications. -- mattb


 * We can take it to a vote if you guys want, but that won't stop people from bringing this up. It's been voted on at least once before (twice, I think), and I still see the same debate with the same arguments every few months. -- mattb


 * OK, so there is consensus and guideline; still it can't be enforced in articles because there is edit war every time somebody tries to push it. So it's going to stay the same despite the guidelines. It seems pretty strange to me.--Pethr 02:16, 27 January 2007 (UTC)


 * To briefly interject... Matt, you're talking about marketing materials, while others (such as Arnold and Pethr) are talking about technical specifications. These are two distinct sets of documents.  Marketing materials are generally prepared by the marketing department and are used for advertising.  Technical specifications are factual documents which are not typically used in any sort of marketing capacity.  --ΨΦorg 21:38, 26 January 2007 (UTC)

In most consumer product manufacturing companies, published technical specifications go through an extensive engineering sign off and then are submitted for review by the legal department. They are contractual documents. Misrepresentations violate consumer protection laws and can result in civil suit. Apple, in particular, is a litigation magnet, so you can count on their review process being particularly thorough. If you can find one, just one, error in an Apple product specification, there are any number of class action law firms that would be glad to hear from you. Here is an example: http://www.apple.com/macbook/specs.html And the fact that you see the same debate with the same arguments every few months may be a clue that there is no consensus here.--agr 15:21, 26 January 2007 (UTC)
 * In these specs, Apple is incoherent. They use GB=1,000,000 B for hard drives and GB = 1024*1024*1024 bytes for RAM. Thus, there are at least 2 errors. Sarenne 15:37, 26 January 2007 (UTC)
 * This usage is standard practice in the industry and Apple clearly states that for hard drive capacity "1GB = 1 billion bytes; actual formatted capacity less." This is precisely the point. You want to substitute your view of how the specs should be written for what is the widely accepted standard in published sources. That is OR and POV. --agr 16:32, 26 January 2007 (UTC)
 * It is not "clear" at all. This usage is incoherent, Wikipedia should be coherent and accurate. We don't want to force Apple to be coherent, that's their choice. Sarenne 16:39, 26 January 2007 (UTC)


 * It is not just marketing.It was what was used BEFORE personal computers.One thing is clear, the articles written in Wikipedia are about the computer of topic. ie.Mac/Pc. The citations and reference links contained herein about the product at hand DO NOT reference the new SI prefix. Therefore, the article should reflect those citations and references.   Planetary Chaos  Talk to me  18:27, 26 January 2007 (UTC)
 * I still challenge anyone to state how this is OR. WP:NOR certainly must allow some degree of latitude-if every article we had here was as sure as some people seem to want to use the "source's exact wording", every article we have would be a glaring copyvio! Trivial paraphrasing is allowed, provided it does not change the meaning of what was said. Finally, as it seems there is not any consensus to change the MOS at this time, I would hope that no one continues to use reverting good-faith changes supported by MOS in an attempt to make a point. Seraphimblade 02:47, 27 January 2007 (UTC)
 * It's OR because technical judgement or interpretation is required for each conversion. This is not a case of "trivial paraphrasing," or even ordinary unit conversion, where the conversion factor is always clear cut. One must determine in each case whether KB, MB or GB were meant as binary or decimal in the source. If published sources do not provide that information, we should not interpolate it. Verifiability is an absolute requirement on Wikipedia, not subject to consensus.--agr 00:05, 28 January 2007 (UTC)
 * That does, of course, leave it up to the readers to make the technical judgement. It also means that the article is "verifiable" in the sense that the article uses the same two letters as the primary source, not in the sense that the article and the primary source both unambiguously specify a particular amount of memory and the two amounts of memory are the same. Guy Harris 01:05, 28 January 2007 (UTC)
 * It's not Wikipedia's mission to "improve" on published sources. We can and do explain the ambiguity. It's described in megabyte, gigabyte and hard drive. An explanation should probably be added to random access memory. Technically knowledgeable readers are already familiar with this issue. Non-technical readers are better served by explanations in plain English rather than the systematic replacement of kB, MB and GB with units that are not used in other publications they are likely to encounter. --agr 13:09, 28 January 2007 (UTC)
 * So you are basically saying each and every reader, professional or layman, would have to know which interpretation of prefixes was meant in a specific context, but not one author may have certain enough knowledge to make the unambiguous change? Sorry, but that would be just silly (and in the real world it is indeed). Compare the issue to pound (rather ounce: troy or avoirdupois) or mile (statute or nautical), which are also ambiguous, at least in certain fields. Christoph Päper 09:56, 29 January 2007 (UTC)

Ok so let's change the wording.That means for every article that contains MB,GB and so on should be changed. Computers,games,phones,memory sticks, any and all devices.I don't know about you but i'm sure this would create chaos on Wikipedia.  Planetary Chaos  Talk to me  17:26, 27 January 2007 (UTC)


 * It hasn't in the year or so it has been a guideline. -- mattb

Only a few pages were changed to meet those guidlines and then changed back I have already counted more then 40 pages that would need to be changed.  Planetary Chaos  Talk to me  23:07, 27 January 2007 (UTC)


 * Do you honestly believe that the majority of pages on Wikipedia conform to this and the other style guides? We don't have to make decisions based on how many articles it would impact, just on the merits of those decisions alone. -- mattb


 * I think we need to change the wording, and have already said that in ambiguous cases we should leave things alone. However, there are many unambiguous cases, and really very few ambiguous ones. RAM, for example, is measured in binary. Period. HD's are measured in decimal. Period. Generally, across any one type of device, it is entirely possible to find which unit is always used. (If it would be strictly necessary, I suppose sources could be found stating that "RAM is always measured in MiB period", and the "MiB" designation could itself be footnoted, in which case sources are being followed.) I personally find the "OR" thing a bit ridiculous and pointy-it's like paraphrasing a news report saying that the only two people present at something were "an old man and a slightly younger one", and saying that there were two people present. Just because a three-year-old who can't do 1+1 wouldn't come to that conclusion doesn't make it OR if any reasonable person with a reasonable knowledge of the issues would. Seraphimblade 13:20, 28 January 2007 (UTC)

Imperial and Metric Auto Conversion like date format
I am noticing a lot of duplication of imperial and metric units. I notice that dates are automatically converted based on user preferences and I am curious if there is a similar capability for distance, area and volume. If one doesn't exist, how do we move to create such an option? Alan.ca 12:19, 23 January 2007 (UTC)


 * I think that would be a bad idea. With dates, you're just reordering the parts of the date. Converting units would be a far more substantial change, and in my opinion requires human judgement. For example, to how many significant figures would you convert it? Which units do you choose anyway? &mdash; e.g., m2 or km2; m3 or litres; inches, feet, yards or miles? And can the engine parse the vast number of different units and different ways of writing them anyway? Quoting the source unit and providing a conversion in parentheses seems like the best solution to me. Stephen Turner (Talk) 14:08, 23 January 2007 (UTC)
 * I believe someone already wrote the patch, as noted either above, or in the bugzilla request for date syntax. Rich Farmbrough, 14:49 23 January 2007 (GMT).
 * Like Stephen, I'm wary of automatising this function. WPians just need to be skilled at judging the variables, for our readers' sake. Tony 15:07, 23 January 2007 (UTC)
 * I'm not wary. It's just a bad idea, plain and simple.  Gene Nygaard 17:30, 23 January 2007 (UTC)
 * Ok sure, you can go through all the City pages and do the calculations manually for Infobox City. Or maybe I should just automate this template, like the hurricane template and then everyone else who finds that it's possible will do the same so we can continue duplicating our work here and wasting time.  The key point here is that entering any figure twice is duplication which is prone to inconsistencies and typographical error.  Further, articles become cluttered with 3 different units stated for the same figure.  That's not to mention the conversion errors people make.  Alternatively, we can state everything in MKS causing regional or interest groups readers who seek to view the encyclopedia in a different measurement standard will find themselves doing the calculations.  Alan.ca


 * Date formats are simple in comparison, yet have caused an enormous amount of difficulty over the years to arrive at the current unsatisfactory situation. Automating units of measurement is an immeasurably more difficult task, just to get to the stage where registered users see what they want to see. And what happens to the majority of our users, who are unregistered and have no preferences to set? What units would they see? In this diverse world, we should cater for users from all cultures and backgrounds as best we can, and while it isn't too hard to work out that 5 June 2007 is the same day as June 5, 2007, we should avoid forcing readers to hunt for a calculator if they want to understand just how long, heavy, hot or bothered something is. --Pete 23:51, 25 January 2007 (UTC)


 * They would see the metric units or whatever units were most common in the world for that type of measurement. Alan.ca 01:27, 26 January 2007 (UTC)


 * Well, of course. But a sizable percentage would see unfamiliar units. It would be like us trying to work out dimensions given in cubits and leagues. We don't want to make life difficult for many of our readers as we pursue, with puritan zeal, a shining set of standards. The current system, where we quote Imperial and metric, and put the most relevant unit first, seems to work better than any alternative that has been suggested. --Pete 02:32, 26 January 2007 (UTC)
 * Who will determine the most relevant unit? Continuing in this fashion only promotes duplication and disputes over which unit is more relevant. Alan.ca 03:57, 27 January 2007 (UTC)
 * We'll work out disputes through consensus. In most cases, it is obvious which units are most relevant. --Pete 16:42, 27 January 2007 (UTC)
 * It's usually the one in the source. Stephen Turner (Talk) 17:12, 27 January 2007 (UTC)
 * Are the two of you suggesting that one article could use different units for the same figure type? Example: Toronto is 30 km from Mississauga, Ontario.  (see ref a)  However, the CN Tower is 1.2 miles tall. (see ref b)???? Alan.ca 10:15, 5 February 2007 (UTC)

Individual Years
I've put up the disputed tag because I think each year in this century, and for that matter, all centuries, are inherently notable. This morning someone else disagreed with this, and I wanted to come here to the talk page of this century where I began to remove redirects to the century as a whole for indvidual years yesterday. Hopefully we can gain a consensus in one way or another rather than have an edit war off the talk page. Just H 14:19, 23 January 2007 (UTC)
 * I don't think any individual year, beyond the 21st century, can be notable enough to have a page of its own. The most distant year to have a page of its own is 2065 and I think that's far enough for now. What information beyond anniversaries and fictional and astronomical events (all of which are on the century's page anyway) can be put on these suggested pages?Philip Stevens 15:14, 23 January 2007 (UTC)
 * The fact that there are anniversaries and astronomical events should be enough, but besides that there are also Non-Crystal Ball predictions from the present or past(just to clarify, even though the events may be crystal ball, the predictions themselves are not Crystal Ball and may be notable, ala Nostradamus or some other notable figure.)

There's also...
 * Dates mentioned in notable fiction.
 * Showing different calendar years for differing calendar systems
 * Pecularities of a year (such as 2002 being a palindrome and so on.)

and probably many others that I can't think of here, but others definately could if we just give them a chance. Just H 18:40, 23 January 2007 (UTC)
 * Yes but all of these can be found on the respective century pages and I see no reason to repeat this information. Also, there is very little of this information per year and creating a page for every individual year would lead to many hundreds of little articles with only "such a thing happen in Star Trek in this year" on them. Would it not be better to have all this information on one or two century pages where it can be easily accessed? Philip Stevens 19:30, 23 January 2007 (UTC)


 * Then let's start pages for George Bush's right foot, Goerge Bush's left foot, Goerge Bush's right thumb,.... Tony 23:30, 23 January 2007 (UTC)
 * That analogy is a bit 'over the top'. I'd like to see a statement on this page that states no pages should be created for individual years beyond the 21st century (not for the next 30 or 40 years at least) and any that are created should be instantly redirected to their respective century. Philip Stevens 06:16, 24 January 2007 (UTC)

Binary prefixes
I would like to question the guideline that states "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." There is currently something of an edit war going on about such a change on the pages dealing with several Macintosh personal computers, e.g. MacBook, Mac Mini. There are several good reasons not to use binary prefixes for articles dealing with consumer products:
 * These prefixes have not received general acceptance. Almost no one uses them. Their use on Wikipedia only serves to confuse readers.
 * While the binary prefixes have official recognition, their limited acceptance arguably makes them neologisms. We should not take the lead in popularizing new terms. See WP:NEO.
 * Use of the binary prefixes makes it harder to verify the accuracy of specifications contained in these articles as one cannot directly compare what the article says with the manufacturer's specifications. Typically there is a mix of binary and decimal prefixes in those specs and sorting this out comes close to original research. See the complex guidelines included in this MOS page.

I suggest that the sentence I quoted above be deleted and that the use of binary prefixes be left to the consensus of the editors working on particular articles or projects.--agr 10:45, 25 January 2007 (UTC)


 * See the "Converting to SI prefixes" section of this talk page for further discussion of that issue.


 * Note also that one of the reasons offered by people converting to binary prefixes is that the decimal prefixes are, in fact, inaccurate, so that when the vendor says that a processor has 4 MB of cache or a machine has 2 GB of memory, that vendor statement is inaccurate as the machine actually has 4 MiB of cache or the machine actually has 2 GiB of memory. Some others appear to have thought, at least at some point, that a processor claimed to have 4 MB of cache has, in fact, 4*1000*1000 bytes of cache, or that a machine claimed to have 2 GB of memory has, in fact, 2*1000*1000*1000 bytes of memory; hopefully, nobody here would argue that, as it's wrong.  Given that, verifying, in any meaningful sense, the accuracy of specifications in an article requires that you somehow determine whether the vendor, when using K/M/G/T, meant 1000/1000*1000/1000*1000*1000/1000*1000*1000*1000 or meant 1024/1024*1024/1024*1024*1024/1024*1024*1024*1024, which could well mean that verifying a specification requires original research. Guy Harris 11:16, 25 January 2007 (UTC)


 * Memory and processor cache operate with the binary specifications. It does not require "original research" to state that a machine with 2 RAM cards at 512 each has a total of 1024 MiB of memory, any more than it requires "original research" to state that the machine has two cards because 1+1=2. On the other hand, hard drive capacities typically are measured in "true" decimal specifications-an HD specified at 200 GB really does have (200*1000*1000*1000) bytes of available storage. Also, we shouldn't shy away from using a specific term over a general one just because it's a "big word" that some people may not initially understand. The XiB prefixes are international standards with unambiguous meanings, they are not a neologism. I would venture a guess that many of our users may not know exactly what sulfuric acid is, but that doesn't mean we should use the more general term "battery acid"-instead, we should use the specific term, and link to its page, so that our readers may learn what the specific term means if it's not clear to them. (Some people refer to the electrolyte in alkaline batteries as being "battery acid"!) In the same vein, it is our job to be specific when referring to technical specifications. The XiB prefixes are unambiguous terms with clearly defined meanings, and we do a disservice to our readers by not being as accurate as we possibly can. Seraphimblade 11:31, 25 January 2007 (UTC)

Wikipedia is a tertiary source and verifiability is one of its bedrock principles. Making manufacturers' specifications "more accurate" goes beyond our charter. Converting to binary prefixes requires some technical judgement. If that judgement cannot be independently verified by a published source, it's a problem. We are talking about dozens of articles here. And I don't agree that it is a service to our readers to use terminology that is not widely adopted--and which may never be. We don't call water "dihydrogen oxide," even though that term may more closely conform to internationally accepted chemical nomenclature. Note I am not suggesting the XiB prefixes be banned, merely that their use be subject to editorial consensus. --agr


 * In practice, vendors use K/M/G/T as binary prefixes; nobody should dispute that. A Core 2 Duo that Intel says has "4 MB" of level 2 cache, for example, does not have 4 000 000 bytes of L2 cache, it has 4 194 304 bytes of L2 cache.  Trust me on this one - it'd just be too much of a pain to get rid of those extra 194304 bytes, given that addressing in modern processors work with binary addresses, not BCD addresses.  Thus, clearly, K/M/G/T are ambiguous, as, when you're not counting bits or groups of bits, they're decimal prefixes, not binary prefixes.  The same technical judgement that would be required to convert those prefixes to binary prefixes in a Wikipedia article would also be required to interpret a Wikipedia article not using binary prefixes. Guy Harris 19:23, 25 January 2007 (UTC)


 * Wikipedia does not derive its value from blindly copying data from manufacturers. Making the data unambiguous and comparable is added value of wikipedia. Note for example that although disks are usually measured in decimal prefixes (e.g. 100 GB), Windows reports with the same visible prefix, but digital values (e.g. 93.1 GB, actually 93.1 GiB). Clearing up that kind of differences is useful. Therefore we should perhaps adopt a similar approach as to other units of measure and indicate both the source value and a conversion or re-interpretation to internationally standardised units. So me might have:
 * the disk size is 100 GB (93.1 GiB)
 * the memory size is 2 GB (2 GiB)
 * This approach allows both verification against sources and verification of interpretation. &minus;Woodstone 14:01, 25 January 2007 (UTC)


 * This isn't a matter of verifiability at all, it's simply which prefix is more accurate. It's not making vendors' specifications "more accurate", it's correctly interpreting the context and removing the ambiguity, something that many readers could not do own their own.  I support the current wording exactly because whenever I've seen this issue play out on a per-article basis it often gets overrun by people who are totally ignorant of the issue and just stubbornly cling to their own uninformed and confused ideas.  A strong guideline is necessary to prevent pages and pages of debate from developing on every article where this comes up; we've discussed it enough times here as it is. -- mattb
 * Referring to editors who disagree with your position as "totally ignorant of the issue and just stubbornly cling to their own uninformed and confused ideas." is unacceptable. See WP:Civility. We are trying to write articles appropriate to our audience. The personal computer industry has thus far not accepted the IEC binary prefix recommendation. A reader who looks at an article converted to use them will be confused when comparing it to other sources, not aided. We allow articles about cars to cite top speeds in mi/hr and km/hr, not m/s and we don't expect fuel economy in meters/mole, even though the later terms are what SI specifies. Nor should we be "interpreting" manufacturer's specifications. From WP:OR "the only way to demonstrate that you are not doing original research is to cite reliable sources that provide information directly related to the topic of the article, and to adhere to what those sources say." I have less problem with Woodstone suggestion of giving XiB units parenthetically, as it is adding info, not interpreting, though, again, I think this can be left to project editors who know their audience and subject area.--agr 16:10, 25 January 2007 (UTC)


 * I'm sorry, but I don't see identifying a problem for what it is as uncivil. I'm not going to water down my words for the sake of some silly ideal of political correctness, but I'm certainly not mentioning names or trying to pick a fight or demeaning anybody.  Nor did I imply that everyone who doesn't agree with me is ignorant; if you bother to read, I have at several points acknowledged the validity of opposing points of view on this matter.  However, the fact is that many (not all) people are either ignorant or confused about this issue.  If you look at all the previous discussions you can clearly see this element popping up in the form of bad arithmetic or just a total lack of understanding of what these prefixes are and intend to do.  If you think it's uncivil for me to say so explicitly, I'm sorry to say that I can't please you with candy-coated words or a runaround suitable for a career in diplomacy.  Anyway, my position is clear, I feel the current verbiage should stay. -- mattb

I fully support the current wording. These prefixes don't cause confusion; they cause accuracy. Just make sure they link to the appropriate articles as specified in the MoS. — Omegatron 15:46, 25 January 2007 (UTC)

what i think? since the pc and mac manufactures and the computers them selves do not use the new si. the article for that computer should not either.i support the new prefixes but only if the computer and the manufacture start using them my mac does not so the article for it should not reflect that change.when a computer or manufacture starts using the new si prefix then yes i think it should go in to the article the way i see it mb may be old not not fully correct but it is what the computer and manufactures say wikipedia should not try and change this as it is not wikipedias or its editors place to do so. —Preceding unsigned comment added by 63.215.27.55 (talk • contribs)

The argument on popular understanding is a straman: nobody really cares about the added i. Christoph Päper 15:25, 26 January 2007 (UTC)


 * I tend to agree. It's an arguable point which is difficult, if not impossible to back up with any hard facts. -- mattb

Unicode
Should unicode character numbers be expressed as 0x1234, U+1234, &amp;#x1234, or &amp;#4660; ? I don't think the html versions are appropriate in contexts where html entities aren't relevant (and, often, when discussing html entities it's better to talk about a named entity) I just changed 12-hour clock because I thought the html entities were jarring, but I'd like to see a style guideline on this issue. —Random832 2007-01-28 18:55 UTC (01/28 13:55 EST)


 * I presume you're asking about a direct textual reference to Unicode entities, not usage of unicode characters (which shouldn't be escaped since Mediawiki directly supports unicode). I'd say that's a good question, and I'd personally prefer to see the U+1234 representation since it is the most common usage I've previously encountered. -- mattb

Units of energy in food articles
When talking about kilocalories/"food calories", which abbreviation should be used, kcal or cal? —Random832 2007-01-29 13:46 UTC (01/29 08:46 EST)
 * Are (kilo)calories backed up by the MoS? Anyway, if you think you have to use them, (all of the various) calories are abbreviated cal, and kilocalories accordingly kcal. I think there was a strange convention to use Cal (uppercase c) for kilocalories somewhere somewhen (which is employed in some mocked calculation that proved drinking cold beer actually required more energy than it provides). That convention should of course not be used on Wikipedia (metric prefixes work just fine here) and even less so the implied one that said use cal (lowercase c) for kilocalories, which I have not heard yet about, but I am only consumingly involved in the business (and hope that (parenthesised) values in kcal will vanish from product labels (in the EU) soon&mdash;also from an aesthetic point of view, because they distort the tables of nutrition facts). Christoph Päper 17:42, 29 January 2007 (UTC)