Talk:Kilobyte/Archive 1

Byte/bit/etc
Here, we recommend B for byte and b for bit, but Binary prefix recommends b for byte and bit for bit. Standard practice (from what I've seen) seems to be k and "bit" and "bps" for bits, and K and B and B/s for bytes, while M,G,T,etc apply to both. There is no 'correct symbol', since nobody ever defined a standard ("kibibyte" is icky and really isn't a 'standard', and I don't think they defined a symbol for bit or byte anyway).

Also, bits and decimal prefixes go together, while bytes and binary prefixes go together (the only exception to this is hard disk sizes).

There is no SI unit for bit, and there's no SI unit for byte. If there was, it would be "no unit" anyway, just like a radian isn't a unit (since an angle in radians is a distance divided by another distance). So "radians per second" is just "per second", and "cycles per second" is also just "per second" (because a cycle is a unit). This is icky, because then f = 1 Hz = 1 s-1, while &omega; = 2&pi; s-1. When I say "1 Hz = 2&pi; rad/s", everyone knows what I mean. When I say "1 s-1 = 2&pi; s-1", it loses its entire meaning.

There's also no standard size for a byte (which is presumably why POP servers call them 'octets'), but everyone uses 8-bit bytes because everyone else does.

If we did talk about bytes and bits in SI units, then I could quote my network speed as 100 Ms-1 (100 Mbps), or 12.5 Ms-1 (12.5 MB/s), or even use MHz. Until we have actual SI units for bits and bytes, inventing units and prefixes so we can use SI prefixes isn't a very good practice (it'd be like kmi or cin or mgal, and who's heard of a 'kilomile' or 'milligallon' anyway?).

We should just stick to common practice (which is arguably the only 'correct symbol'), or something non-ambiguous (KiB isn't ambiguous, but Kib looks like one symbol). And when we quote hard disk sizes, GB (decimal). And we shouldn't have inconsistencies around the wiki.

Thoughts? -- Elektron 13:43, 2004 May 20 (UTC)


 * A little off-topic, but doesn't "1 s-1 = 2&pi; s-1" simplify to "1 = 2&pi;"? --210.11.188.17 04:35, 3 March 2006 (UTC)
 * No. You must include the units. 1 Hz = 2&pi; rad/s which means 1 cycle/s = rad/(2&pi;.s) => 1 cycle = 2&pi; rad. 203.206.58.220 08:36, 3 March 2006 (UTC)
 * Exactly. The article is wrong in saying bytes are B and bits b. I know a few 1980s texts where the opposite was the case. The arguement was that kilo was always small 'k' and 'kb' was kilobyte and the odd-cass 'kB' was the less common kilobit. Carewolf 08:57, 6 September 2007 (UTC)

Kibibyte?
I feel that this and the related pages are being used to push an agenda. I have worked in IT for about 8 years and I have been using computers for 22 years, and I have never heard of a kibibyte until I read this article. I have asked many of the people I work with and not one of them had heard the term.

I think it is misleading to present the term as if it has any currency whatsoever. If the term would confuse industry professionals, it should not be listed as an alternative, merely as a curiosity.

The reality is that 1000-byte figure is very rarely used, and only the specialisations of networking (where it doesn't matter precisely how much is pushed through, and since figures are presented to humans in decimal, dividing by 1000 is easier) and by marketing agencies who are trying to make their products sound like they can store more data.

The industry standard is kilobyte = 1024 bytes, this is what is taught in most schools and universities. That it is inconsistent with SI doesn't matter because people dealing in bytes know what they're talking about.

Ben Arnold 01:18, 15 Jun 2005 (UTC)


 * You're not the first. Go read binary prefix.  It's been adopted by the IEC, IEEE, NIST, etc.

"That it is inconsistent with SI doesn't matter because people dealing in bytes know what they're talking about."
 * Yes it does. People who are dealing in bytes aren't the only people involved.  It's inconsistent with people developing clock circuits in hertz, for instance. - Omegatron 01:21, Jun 15, 2005 (UTC)
 * We aren't talking about clock hertz, but about kilobytes. People talking about kilobytes mean 1024 bytes unless they state otherwise.  Hard drive manufacturers have to put a footnote for their non-standard usage.  This is different from the SI meaning of the "kilo" prefix, but it's not clear what that matters.  I second Ben: this article is trying to improve usage, but an encyclopedia should describe usage.  - Lex Spoon, September 18, 2009

To continue this discussion. Ben, there is no agenda being pushed here. As stated by Omegatron, many engineering organizations have accepted it. Simply because people don't use the -ibi- form to mean powers of 2 is not sufficient grounds to rewrite an article stating the contrary to what IEC, IEEE, etc. have adopted. Your rewrite contradicts the table right next to it... Cburnett 01:28, Jun 15, 2005 (UTC)


 * Those organisations are, at the end of the day, lobby groups. They have a lot of prestige, and maybe with time they will influence general usage, but on this issue, at the present time, they are in the extreme minority. The UN has a lot of prestige but Wikipedia doesn't defer to them for definitions of countries (or we wouldn't have Taiwan on the list). I'd rewrite the table if I had time, but the whole group of articles is a big "what some Wikipedia users would like the world to be like", not "what the world is like" bias... and that's utterly unencyclopedic. It brings down the standard of Wikipedia as a whole. Ben Arnold 01:42, 15 Jun 2005 (UTC)


 * How is it biased?? - Omegatron 01:56, Jun 15, 2005 (UTC)


 * Ben is around enough to revert my deletion of {&shy;{POV}} but can't follow through, as expected when putting it on a page, with the discussion. Cburnett 04:06, Jun 15, 2005 (UTC)


 * Sorry I have a job, I can't be every hour or even every day... I've put the places that seem "point of view" below.


 * Additionally, I heard about it at least a couple years ago. Even the hard drive manufacturers agree with IEC, IEEE, etc. (though I do think they have an agenda since it makes their drives seem bigger by playing on people's (incorrectly) understanding that 1 kilobyte = 1024 bytes). Cburnett 01:30, Jun 15, 2005 (UTC)


 * A lot of people think this is some kind of marketing agenda, because hard drive manufacturers use the SI prefixes properly, which makes their storage sizes look bigger compared to the colloquial definition.
 * Your reversion is good. - Omegatron 01:33, Jun 15, 2005 (UTC)

Areas of the article that are POV:

equal to one thousand bytes.
 * this is POV Ben Arnold 06:59, 19 Jun 2005 (UTC)

The prefix K was used, to distinguish this quantity from the SI prefix k. However, the K prefix was never formally mandated and it is not used consistently.
 * the inference is that being distinct from the SI prefix and having a formal mandate are important; which is fine, but only if the counter-argument, that they don't matter, is noted Ben Arnold 06:59, 19 Jun 2005 (UTC)

the subtle upper-case / lower-case distinction between the SI prefix and this special use in Computing, was not available
 * the counter-argument is that since computing doesn't need fractional units, the SI case distinction could be dropped in favour of binary/decimal case-distinction Ben Arnold 06:59, 19 Jun 2005 (UTC)

significant errors of measurement [...] (which is about 47KB or 48kB)!
 * this statement isn't put into perspective (47KB out of 1024KB is 2.4%) and isn't worth an exclamation mark Ben Arnold 06:59, 19 Jun 2005 (UTC)
 * a big reason people use the decimal measure is precisely because they don't care about precision, but this isn't noted Ben Arnold 06:59, 19 Jun 2005 (UTC)

This problem, together with other abiguities eventually lead to the creation of the binary prefix standard.
 * calling it a standard, but not pointing out that it is very rarely used, is POV Ben Arnold 06:59, 19 Jun 2005 (UTC)

Because [...] continuing irregularities in using the binary prefix in the definition and usage of the kilobyte, the exact number in common practice could be either one of the following:
 * marginalising the most common useage as an "irregularity" is POV Ben Arnold 06:59, 19 Jun 2005 (UTC)


 * 1000 listed before 1024... even though 1024 is the more common usage Ben Arnold 06:59, 19 Jun 2005 (UTC)


 * Thank you for discussing before tagging as POV this time. Some of your complaints are genuine, some are very minor, and some have not been in the article since I condensed it four days ago. Either way, you should try and deal with them yourself, instead of complaining until someone else does.


 * But be more careful than you were in your one substantive edit to this article, some of which still remains, but most of which was redundant discussion not specific to kilobyte. This article is supposed to be quite short. Any detailed discussion of the binary/decimal confusion, including the hard-drive size thing, should go onto binary prefix.


 * In the meantime, since you do not appear to be very active on Wikipedia, I will remove the POV tag. – Smyth\talk 11:50, 19 Jun 2005 (UTC)


 * Active is a relative term. I'm in almost every day. That's been active enough for every other discussion I've been involved with. The point of the POV tag is to indicate that a discussion is occuring (this one). One party to that discussion should not unilaterally remove the tag unless the discussion has died. As for "you should try and deal with them yourself", I did try to deal with them myself, but my change was reverted and I was told to discuss the issues here. As a discussion is happening here, until this discussion is resolved, the page should remain tagged as POV.


 * As far as I'm concerned you're trying to have it both ways: I can't change the page because I haven't discussed it here and I shouldn't discuss it here I should just change the page. Catch 22 is a form of passive aggression. Rather than discussing the issues, you're tying me up in red tape of your own invention.


 * I'm genuinely interested in improving Wikipedia and making it a useful resource. I've come to the opinion that this article is one of those places on Wikipedia where a group of users lobby for a particular point of view and defend it religiously. That's called POV and is not the way things work on Wikipedia. In theory eventually you will get tired of defending your fiefdom and go away. Other people will recognise that what's said on these pages is incorrect and misleading and will change it. I hope the theory's true. Ben Arnold 03:30, 20 Jun 2005 (UTC)


 * Waiting for the other side to give up: sounds like a siege to me. Is this where I repeat your sentence: "That's called POV and is not the way things work on Wikipedia." <?> Cburnett 03:39, Jun 20, 2005 (UTC)


 * I'm not waiting for anything. I want to discuss how we can bring the article in to balance right now. My point was that in the fullness of time balance will supposedly come anyway. That's the theory. Not from anything I do, or anyone else, but by the same means that water carves a river through rock. Ben Arnold 04:00, 20 Jun 2005 (UTC)


 * For goodness sake, I didn't disagree with any of the complaints you had. I invited you to try and deal with them yourself. I pointed out what the problem was with your other edit. Yet all you seem to be interested in doing is causing trouble.


 * I will now go and deal with your complaints myself. – Smyth\talk 12:47, 20 Jun 2005 (UTC)


 * "In theory eventually you will get tired of defending your fiefdom and go away."
 * That's not the way it works. The way it works is that we edit the article together until it represents something we can both agree with.  If we can't do that politely or without revert warring, we talk about it on the talk page, and if we can't reach consensus on the talk page we pettily fight about it incessantly until someone starts dispute resolution.  :-)
 * "I can't change the page because I haven't discussed it here and I shouldn't discuss it here I should just change the page."
 * Of course you can change the article. Be bold.  "...but don't be reckless."  If other people disagree with your edits and you are unwilling to concede, you need to talk about it here instead of editing the article.
 * "That's called POV and is not the way things work on Wikipedia."
 * Honestly, I think you need to take a good long look in the NPOV mirror before making further contributions. A kilobyte is 1000 bytes in a lot of contexts.  Just because you haven't heard of it doesn't mean it's untrue.
 * Yes, a kilobyte is 1000 bytes in a number of contexts, I first heard of that usage years ago and I have never said anything to contrary. I agree that if I was presenting the opinion you attribute to me it would be quite firmly POV. However, I don't have that opinion. Perhaps you're reading what you want to read not what I say. Ben Arnold 00:28, 26 Jun 2005 (UTC)
 * "Not from anything I do, or anyone else"
 * Of course it's from what we do. The article's not going to edit itself.  We work on it together and eventually it converges to something we can both agree with.  - Omegatron 13:26, Jun 20, 2005 (UTC)
 * Yes that's the way it normally works. But that's not the way it worked on this page. On this page my changes were reverted the moment I made them without any explanation as to what was wrong with what I had written. Everytime I put up a NPOV message (because there was a genuine dispute going on) it was pulled down immediately. Put yourself in my shoes, you feel like you're up against a brick wall and it becomes too much of an effort to go forward.
 * However, I have to concede that there is obviously some reasonableness around here, because the article has now improved dramatically to the point where I only have two major complaints, I'll write about them below. Thanks to those who have listened to what I have had to say and moved forward rather than stonewalling. Ben Arnold 00:28, 26 Jun 2005 (UTC)
 * Omegatron, you edited the article to say that "kilobyte per second" always has a decimal meaning. This is surely false, as experimenting with any program that downloads anything will show. – Smyth\talk 13:49, 20 Jun 2005 (UTC)


 * Oh, right. For things like modems and ethernet it is decimal, but for software it is binary, because they are referencing to file sizes. - Omegatron 14:04, Jun 20, 2005 (UTC)

This discussion does not belong here but on talk:binary prefix since Ben's POV complaints are not unique to the kilobyte article. Cburnett 18:03, Jun 20, 2005 (UTC)

Two remaining neutrality issues
1. Unfortunately this abuse of the SI prefix got carried away from the slang of computer professionals into the mainstream lexicon by the marketing people, creating a lot of semantics problems.

This sentence should be revised or removed. I'm not exactly sure what it means, but it sounds like it requires some sort of supporting evidence, and the term "abuse" it clearly POV. I'd argue that rather than being an abuse of SI the "kilo-" prefix was orginally used by analogy with SI. Is the Windows term "desktop" an abuse of the term desktop because it's not exactly a desktop? Ben Arnold 00:28, 26 Jun 2005 (UTC)


 * Oh, c'mon. - Omegatron


 * I agree with Ben re POVness of the phrasing. But really, Ben, 1000 is not analogous to 1024. The kilo- prefix has meant 1000 since at least the beginning of the 19th century, long before anybody ever said "kilobyte"; therefor, IMO using kilo to mean something other than 1000 is misuse of the SI prefix. - Jrv 5 July 2005 21:14 (UTC)


 * I agree, the sentence is very bad. It would be nice if someone could find references for the origin of the binary prefixes, as there seems to be no actual historical information about it in Wikipedia. – Smyth\talk 5 July 2005 21:58 (UTC)


 * Do you mean the history of the IEC binary prefixes or the use of SI prefixes to mean powers of 2? - Omegatron July 5, 2005 23:28 (UTC)


 * The use of SI prefixes to mean powers of 2. – Smyth\talk 6 July 2005 18:47 (UTC)


 * Hmm.. I wonder if that was even a definite event.  "By 1970, Intel had produced a memory chip that could store one Kilobyte of information"  Someone should still be around who remembers... - Omegatron July 6, 2005 19:50 (UTC)


 * "In 1968 Intel released the first 1 kilobyte memory chip."
 * Thanks, Google Print. - Omegatron July 6, 2005 20:00 (UTC)

2. The table

Maybe this should be discussed elsewhere, but the table is misleading. It omits the fact that many (most?) people still use the term kilobyte to mean 1024 bytes. I'll edit the table now to "be bold" and show you what I think it should look like. Ben Arnold 00:28, 26 Jun 2005 (UTC)


 * Also Template:quantities of bits - Omegatron 03:55, Jun 26, 2005 (UTC)


 * See also Manual_of_Style_(dates_and_numbers) - Omegatron 04:02, Jun 26, 2005 (UTC)


 * Maybe put the standard values in bold or the colloquial values in parentheses or something, to set them apart?
 * By the way, this discussion should really be taking place at Talk:Binary prefix, since it pertains to more than just this article. - Omegatron 04:12, Jun 26, 2005 (UTC)

Where did the new second paragraph come from?
A few weeks ago, I was quite impressed that this page treated the issue reasonably well (as in, stating the facts without getting embroiled in one side or the other in the 1000/1024 discussion)

Now I come back to find this, right at the top of the article:

''The term "kilobyte" was first loosely used for a value of 1024 bytes (210), because 210 is roughly one thousand and powers of two are convenient for use with binary digital computers. This misuse of the SI prefix got carried away from the slang of computer professionals into the mainstream lexicon by the marketing people, creating a lot of semantics problems.''


 * Loosely used? What's loose about what was a common and well-defined unit of measurement?  It had an exact value which was understood; the definition only became indistinct (not 'loose') when a second unit of measurement appeared, sharing the same name and abbreviation
 * Misuse of the SI prefix - since when was SI involved in the measurement units of information?
 * Slang of computer professionals? - it's a unit of measurement, not slang!
 * By marketing people - that's rather a simplistic way of putting it isn't it? For an encyclopaedia article, I mean
 * Creating a lot of semantics problems - This page should probably explain the semantics, rather than vaguely refer to their existance

''Some suggested that the prefix K should be use to distinguish this quantity from the SI prefix k. However, the K prefix was never formally mandated and it is not used consistently. When larger units were needed for millions of bytes or more, the subtle upper-case / lower-case distinction between the SI prefix and this special use in computing, was not available (SI already uses the prefixes m and M to mean "thousandth" and "million" respectively). Higher-order SI prefixes are therefore used with either decimal (powers of 1000) or binary (powers of 1024) values, depending on context. See binary prefix for more details.''


 * Some suggested is not an accurate enough way to introduce encyclopaedic topics
 * However, the K prefix was never formally mandated - who exactly is in charge of 'mandating' the units of measurement we use? Were all the people who measured farms in acres thousands of years ago doing so illegally because ISO or ANSI hadn't given them some mandate?
 * Comparing it to ISO units is misleading anyway, because their value depends on complex physics experiments such as keeping the kilogram in a vacuum, counting atomic oscillations to measure the second, calibrating luminance meters, etc. Suggesting that calibration laborotories like those are required to count up to 1024 is just plain silly.
 * SI already uses the prefixes m and M to mean "thousandth" and "million" respectively - that's good, but again, why should users of the kilobyte care what SI is using for metres, kilograms, lumens, and seconds?

The new paragraphs do however, offer a clue as to some appropriate action


 * See binary prefix for more details

Couldn't the entire set of POV paragraphs be replaced by this one link? Ojw 12:15, 26 Jun 2005 (UTC)


 * Where from? That's not a fair question, unless you are claiming that the history section does not adequately reflect what has been done. You aren't claiming that, are you?


 * No, SI does not deal with units of information. However, the organizations which created those prefixes for use with SI and the rules for their use (BIPM, ISO, NIST, etc.) do sometimes assert an ownership of sorts, insisting that people use them the way they defined them.  They don't have any real enforcement powers for these claims of ownership, but that does not mean that they cannot make them.


 * "Some suggested" describes quite well a situation never officially pushed by any standards organization. The rest of the explanation following "Some suggested" describes the situation quite well, does it not?  I agree with you that there is no requirement that any of this usage be "mandated"; that is indeed one of the topics that many people do not understand about the various units of measurements and the rules for their use.   For example, the pound-force does not have an "officially mandated" definition, yet the definitions in actual use do not vary much.  There is also no guarantee that what is "mandated" by one metrological or professional organization will be the same as what is "mandated" by another, and the same goes for various other rules and recommendations that fall short of any mandate.


 * I agree that "slang" is not the proper terminology. So, rewrite it in some way that better reflects the fact that it is fairly well accepted jargon within that field.


 * I don't see how anybody could misunderstand the references to "semantics" to be referring to anything other than the ambiguity of the same prefixes being used for both the 103n and the 210n definitions.


 * Note also that the commonly used "kilobytes per second" is a measured quantity, not a counted quantity; it is just like a meter, a lumen, or a second in that regard. Bytes per second and bits per second take real number values, not integer values only. That's a partial answer to your question, "why should users of the kilobyte care what SI is using for metres, kilograms, lumens, and seconds?"  There are interrelationships involved here.  Gene Nygaard 13:20, 26 Jun 2005 (UTC)


 * Loosely used: This paragraph describes the history of the prefix. Noone has ever defined this unit, and when it was first used, it was used loosely.
 * Misuse of the SI prefix: It's rather the other way around. The units of information measurement got involved with SI.
 * Slang of computer professionals: When it was first used, this wasn't a unit of measurement, it was slang.
 * Marketing people: By all means feel free to improve the wording.
 * Semantics problems: Good point. I'll make it clearer.


 * Some suggested: Once again, feel free to improve the wording.
 * Formally mandated: BIPM. I don't see how ISO or ANSI are related to this?
 * ISO units: Once again, ISO has little to do with units.
 * SI: I don't see any reference to metres, kilograms, lumens, and seconds in this article. You imply that users "use" SI prefixes without caring about them? You just pointed out the problem. Delicates 13:53, 26 Jun 2005 (UTC)


 * The paragraphs describe the history of prefix. I don't see you offering any alternative history while claming this as a POV? Delicates 13:53, 26 Jun 2005 (UTC)


 * BIPM doesn't "formally mandate" anything; of the three organizations established under the Metre Convention, that is the province of the CGPM.
 * The CGPM, CIPM, and BIPM only provide the basic framework for the International System of Units. Many of the rules for its usage, especially in particular applications and in particular locations, are fleshed out by international standards organizations such as International Organization for Standardization, various other transnational organizations, national standards laboratories such as NIST and National Physical Laboratory, professional organizations such as IUPAP and IUPAC and WMO, etc. Let's not leave out the International Electrotechnical Commission, which set the standards for "kibi-", etc.  None of them has plenary power in this area.  See standards organization for many others.  Gene Nygaard 14:47, 26 Jun 2005 (UTC)

Which should we use in Wikipedia?
Discussion at Village pump (policy)/Archive I. - Omegatron July 8, 2005 13:25 (UTC)

A vote has been started on whether Wikipedia should use these prefixes all the time, only in highly technical contexts, or never. - Omegatron 14:50, July 12, 2005 (UTC)

DSL Line speed?
The article says: "For instance, a 512 kbit/s DSL line has a capacity of exactly 512000 bit/s. Dividing by 8, this is 64000 bytes/s, which is 62.5 KiB. However, the unpredictable levels of transmission overhead and error rate mean that the 2% error in referring to this as "64K" is fairly insignificant."

I'm not convinced that any of this is true! Most 512k ADSL lines actually sync at 576k. Not sure whether that's actually 576000 or 589824 in practice. Telecomms folk tend to be rather blase about all this - for example a 2 Meg E1 circuit is really 2048000 bit/s - a neat cross between the thousands and the 1024 way of doing things. I'd think this would be better deleted as a paragraph - it's not really helpful to the article itself.


 * I wrote those sentences, and my DSL modem also reports 576k. I assume the difference between this and 512k is to account for overhead below the IP layer, i.e. PPP and ATM framing. I don't know whether the IP layer itself is included in most bandwidth specifications, but yes, I can see how this would be confusing. – Smyth\talk 11:05, 27 August 2005 (UTC)


 * It should say what the hardware is actually communicating at first, ignoring layers and wrappers and compression. Then you can put typical file transfer rates after that. -  — Omegatron 14:14, August 27, 2005 (UTC)

As you say, the 576/512 difference is to do with ATM framing etc. However, as Omegatron says, this does confuse matters. Could I suggest a better example would be either a T1 or E1 line. An E1 is exactly 2048000 (and I'm sure of this - I work in telecomms and frequently attend the DSL Forum and have no idea what ADSL speeds are exactly). This is composed of 32 x 64000 bit/s (normally referred to as 64 kbit/s) streams. A T1 is 24 x 64000.


 * I've now taken the liberty of editing the DSL analogy out of the article. I've done this since it's not actually factually true (a 512kbit/s DSL line does not have a capacity of exactly 512000 bit/s) and I don't really think it adds a great deal to the argument. --Phil Holmes 13:47, 6 September 2005 (UTC)


 * Fair enough. – Smyth\talk 17:25, 7 September 2005 (UTC)


 * Thanks. :-) --Phil Holmes 21:00, 7 September 2005 (UTC)

Æ

Powers
Hold on... I made a valid edit, that the binary is powers of 2 and digital powers of 10, however the community seems to think that these are of 1024 and of 1000 instead...... How is 256 a power of 1024, please? Jezthepie 14:46, 7 October 2006 (UTC)


 * What does 256 have to do with anything? — Omegatron 16:35, 7 October 2006 (UTC)

Use of "kibibyte" in BitTorrent and the Linux Kernel?

 * "Although the word "kibibyte" is seldom seen in practice, it is starting to be adopted by software in which precision is important, such as BitTorrent or the Linux kernel. "

Three problems here:
 * How is "precision" more important in BitTorrent and the Linux kernel than in other software? All software must be bit-perfect in its internal use of memory (if a program can't remember where its variables are located, it certainly won't work!).
 * The Linux kernel is not written in English, and does not include a high-level user interface, so how does it "adopt" the word kibibyte? - in its documentation? I have just searched the entire source tree of the latest kernel release (2.6.18), and found only two occurrences of the word kibibyte (in a documentation file and a comment-line), compared to 70 occurrences of kilobyte.
 * The reference link points to a posting on a Linux kernel mailing list, expressing one individual's "esthetic distaste" for the word kibibyte. The context and outcome of the discussion are not clear. The subject line suggests it is something to do with the Configure.help file (which was replaced by a set of smaller Config.help files in kernel version 2.5.3). The current Config.help files do not contain the word kibibyte.

I propose removing the sentence, unless anyone has a better suggestion. Mtford 06:10, 5 November 2006 (UTC)


 * It has nothing at all to do with internal accuracy, but with the amount of accuracy needed by the user. In a disk partitioning application, for instance, it's very important to get the value exactly right, which is why GParted uses IEC prefixes and fdisk uses decimal SI prefixes.
 * There are two issues here: Use of SI prefixes in the correct decimal sense and use of IEC prefixes for a binary sense. Just because a program doesn't use the word "kibibyte" doesn't mean it's not using units correctly.  apt-get uses "kB" and "MB", for instance, but in the correct decimal way.  According to, the Linux kernel uses the correct units:
 * — Omegatron 16:19, 20 May 2007 (UTC)

K and k
The current phrasing is more clear (run-on sentences tend to be bad). The reason you can't extend it to M is because M is already capitalised — you could, of course, use mB to mean 10^6 bytes, and MB to mean 2^20 bytes, or vice-versa, but this could cause confusion since m is already an SI prefix. That's what the paragraph is getting at. The fact that millibytes don't exist for most practical purposes is largely irrelevant (it means that you can use mB, while being unclear as to what). I don't get what you mean by pros and cons — your edit adds neither. If anything, the bit about m should be removed; "M is already capital" is reason enough. ⇌Elektron 02:05, 16 September 2007 (UTC)

Not prefix: suffix
Not prefix: suffix. Suffix already. —Preceding unsigned comment added by 82.25.242.114 (talk) 23:46, 27 March 2008 (UTC)

Folderol (talk) 11:38, 12 November 2008 (UTC) I propose changing the order of abbreviations so that capitalized KB leads the list as a most commonly used

Citations required
Besides many other IT profesionals never hearing of the term "kibibyte" until this article, you make the statement: "most standards organizations instead recommend the term kibibyte (KiB)"

Citation is required for this. Which organizations?

Navywings (talk) 18:58, 28 April 2008 (UTC)


 * I copied over a couple of citations from the "main article: Binary prefix". Improvements are welcome! shreevatsa (talk) 02:41, 29 April 2008 (UTC)

Fabrication
I googled to check--- nowhere outside of NIST is the word "kilobyte" used to mean 1000 bytes. To say that it has two meanings when it really doesn't is a fabrication.Likebox (talk) 22:06, 15 November 2008 (UTC)


 * The definition of kilobyte, megabyte, and gigabyte has long been accepted to be powers of two. If people wish to redefine the concept using a standards body, the interpretation of the standards body has to become standard usage before an encyclopedia adopts it.Likebox (talk) 05:33, 16 November 2008 (UTC)


 * The concept of notability using 'secondary sources' can only be applied to terms that are not governed by standards organizations. This is precisely why we have standards bodies, to recommend, define, and promulgate standards. Fact is that standards bodies have redefined these terms. As it may be inconvenient, odd, or confusing (for a while) the new definitions make darn good sense. It doesn't make much sense that a technical discipline (IT) founded on the modern principles of science and technology should promote confusion in its use of units. Kbrose (talk) 17:54, 16 November 2008 (UTC)


 * In addition, it is not correct to state that "IT" ever universally used "kilo" to mean "1024". For example, "1 kbps" or "kilobit per second" has always meant "1000 bits per second", not 1024. And hard disks are also very much part of "IT", and the prefixes are used in their SI (decimal) senses not just by manufacturers but also by buyers :-) The simple fact is that unlike RAM, hard disks are not built on binary trees, and have no reason to use powers of 1024, so they use the normal meanings; there is no conspiracy involved. Also see 65k colors -- that's 65536 which is 64&times;1024, but it's still often called "65k". The Art of Computer Programming is one of the most respected computer science books, and it uses "kilobyte" to mean "1000 bytes" everywhere. Apart from RAM, there is no place in IT where powers of two arise naturally, and all other areas of IT have been using "kilo" to mean 1000. This is not a "fabrication". Shreevatsa (talk) 18:51, 16 November 2008 (UTC)

This is bollocks
I have to say I was quite amazed when I found that Wikipedia had decided to use metric values for kilobyte, megabyte, etc. Having worked professionally in IT for 25 years I have never, in all that time, not once, seen kilobyte refer to anything other than 1024 bytes. It's well known that disk drive manufacturers use "marketing units" to describe drive capacities but they are almost universally derided for doing so, and are very much the exception. When's the last time you saw a 4.19 megabyte RAM chip advertised, for example?

The term kilobyte isn't an SI unit name, it's a technical term in its own right, clearly inspired by the SI prefix for 1000, but not beholden to it. Or would you claim that a megalopolis is "1000 kiloloposis", that the millipede is misnamed and ought to be a "kilopede" if it has 1000 rather than 0.001 legs, etc.? This all sounds like the agenda of some pedant engineer getting off on how superior his or her knowledge is. It's the sort of argument I'd expect to hear from a spotty teenage nerd on the school playground and has absolutely no place in a work like Wikipedia if it wants to reflect actual usage of the term and retain any hint of credibility. —Preceding unsigned comment added by 24ten (talk • contribs) 16:54, 4 September 2009 (UTC)


 * 25 years and you never realized that your 10-base-T network runs at 10,000,000 bits/second (10 Mb/s), not 10,485,760 bits/second (10 Mib/s)? Amazing how those that know the least are the most arrogant.75.219.69.212 (talk) 23:20, 13 January 2011 (UTC)


 * Then how would you decide to solve the ambiguous problem in the general public?
 * There IS a problem with the misuse of the prefixes, the IEC-prefixes solves the ambiguity.
 * The misuse of this units caused companies and individuals a lot of money and problems.
 * One of my problems: When doing physics simulations,
 * trying to think the physical and computer context separately isn't a good idea.
 * (think of my physics simulation example)
 * The jargon of IT-specialists here has to make way for the general public.
 * Linux kernel (somewhere from 2001) and Mac OS 10.6 uses it.
 * Adapting to new different ways of saying stuff is weird.
 * Try to have some progressiveness. (< this is supposed to be encouraging)
 * Now a good joke about it:
 * It's hard, that extra i, really sucks all the ink out of my pen and productivity out of my mind.
 * Thelennonorth (talk) 14:39, 11 September 2009 (UTC)
 * Now a good joke about it:
 * It's hard, that extra i, really sucks all the ink out of my pen and productivity out of my mind.
 * Thelennonorth (talk) 14:39, 11 September 2009 (UTC)
 * Thelennonorth (talk) 14:39, 11 September 2009 (UTC)

The general public? They know that 8MB is twice as much as 4MB but are they really confused (or do they even care) by precisely how many actual bits of data that represents? There was no confusion until a few misguided pedants decided to apply strict SI terminology to something it was never intended for. And don't get me wrong - I'm a scientist by training and love SI units. Let's not get all revisionist though and pretend these terms mean something they don't.--24ten (talk) 22:36, 18 September 2009 (UTC)


 * @24ten, I'm afraid you're totally misguided and your arguments make no sense. 1 kilobyte is/was/will be 1000 byte, no matter how it is used.  the meaning of it is in it self, just because people like you are not able, or not willing to understand it, doesn't make it change its meaning. 84.72.61.221 (talk) 18:21, 4 November 2009 (UTC)


 * It wasn't always 1000 bytes... It isn't even universally 1000 bytes right now. —Preceding unsigned comment added by 67.22.211.139 (talk) 03:17, 25 February 2010 (UTC)


 * This is a great example of Wikipedia elites browbeating everyone else into their own prescriptive POV
 * http://www.askoxford.com/concise_oed/kilobyte?view=uk
 * http://www.merriam-webster.com/dictionary/kilobyte
 * http://dictionary.reference.com/browse/kilobyte (The only one that even mentions 1000 as a semi-valid meaning) Celmonas 02:21, 5 April 2010 (UTC)

It is not an example of Wikipedia elites browbeating everyone. The dispute has occurred because two different industries have met and they have traditionally used different terms for the same thing. In computer science it is far easier to do the maths in base 16 hexadecimal (or base 8 octal) than it is to use base 10. So computers science and the computer industry has developed using 1024 (2^10) and designating that as 1K. However the telecommunications industry has always measured quantity in so many bits per second. The baud rate used by the telecommunications industry was a feature of the technology and has always been measured in bits per second, when they managed speeds of 1000 bits per second they naturally called it 1 kiobit, telecommunications companies were interested in bits moved not in computer words (which anyway have not always been based on 8,16,32, or 64 bit lengths).

Historically the telecommunications industry (like Weather forecasting) have had a traditional interest in international standards, (AT&T my have had a toy called UNIX but it was never more than a toy) while the computer industry has traditionally been based on insular propitiatory technology. So it is almost inevitable that the telecommunications industry would drive the standards, and for them, as they measured bits per second it was sensible to to to 8 bits for a word and call 8,0000 bits a kilobyte, as to them the maths is based on bits moved and its easier to work in base 10 for billing purposes.

For computer scientists it is a real pain, to have to convert to base 10 as the maths is a mess (and after a few years as a programmer advances in their profession to become a real programmer they will be thinking in hex). However it is not all bad news for the industry because it means that the marketing departments in computer peripheral companies can sell smaller disk to other quiche eaters stating they are bigger than they really are, and with each move from kilobyte to megabyte to ... the difference gets bigger and bigger :-)

There is a fair bit of conversation above about disks (which seems to imply that somehow that accessing them is different from ram), all I can say to that is it depends at what level one is talking, but at OS and sub OS levels the maths are much easier if one uses hexadecimal. Again for most -- the quiche eaters -- it doesn't matter as a disk is a binary item: it works, it doesn't work; its not full, its full.

I think this article need to be rephrased, to recognise that there are two different meaning for the term with a lot less emphasis on which is correct. Currently it is pushing a POV and makes the article less useful for those readers who wish to read a balanced and impartial article. -- PBS (talk) 05:53, 15 April 2010 (UTC)

POV-pushing to promote the IEC prefixes
This version of the article was unacceptable. I deleted the table. My reasoning is clearly stated in the edit summary:" POV-pushing. The table showing the IEC prefixes discusses “kibibytes” and other terminology that is virtually unknown in the real world. If they are to be discussed, it is important to do so only in a footnote-like fashion that they failed.

MOSNUM is clear that it is improper to have tables and text that could lead the reader to conclude that the “correct and good way” way to denote binary capacity is to use IEC prefixes such as “kibibyte (KiB)” when the RSs far and wide ignore such terminology. Wikipedia follows the way the real world works and does not try to lead by example.

That table suggested that “KB” = 1000 bytes and “KiB” denotes 1024 bytes. In fact, “KB” can mean either depending upon context. I know: that sucks and isn’t the way it ought to be. But that’s the way the real world works and Wikipedia’s bedrock principal is that we always follow the RSs. Always. The vast majority of RSs do not use “KiB” and not one single periodical directed to a general-interest readership uses them.

As for uppercase v.s. lowercase K for “KB”, here is the reality: The lowercase “k” is the proper unit symbol for the prefix kilo. Uppercase “K” is properly the unit symbol for the unit of thermodynamic temperature kelvin. Moreover, the rule of the International System of Units (the SI) is that when describing the magnitude of a measure, a space always separates the numeric value of a quantity and its unit symbol, e.g. “1 kB”. Nonetheless, it is exceedingly common within the computing industry when denoting binary capacity—particularly in marketing literature and product packaging—to use uppercase K and no space (1KB), although “1 KB” is not incorrect and is often considered more suitable in technical writing.

When I have a moment, I will start a MfD on that table. Greg L (talk) 17:19, 2 May 2011 (UTC)

P.S. I completely revised the article so it has real facts without the extreme bias for talking at great length about the IEC prefixes, which was a violation of WP:WEIGHT and WP:SOAP. In particular, I removed the following text, all of which was utterly false:

1000 bytes (103): This definition is recommended for all uses by international standards organizations such as IEC, IEEE, and ISO,[6] with the abbreviation "kB". This unit is common for quantities which are not based on powers of two, such as bitrates. This definition is used for storage quantities in software such as the Linux kernel and Mac OS X 10.6.

First, OS X doesn’t use kilobytes in reporting hard disk capacity; it uses megabytes or gigabytes. The citation pointed that out. And clearly, OS X doesn’t use the IEC prefixes whatsoever. Secondly, the “citation” regarding Linux was not an RS; it was correspondence from one individual to another on a blog. Thirdly, the kernel for Linux was written before the IEC proposal. Linux uses terminology like KB and MB in the conventional fashion. The only part of Linux that doesn't work this way is a particular partitioning tool that uses a GUI interface (the built-in command-line tool works conventionally too) for a particular flavor of Linus (Ubuntu). Given that Linux, in total, has about a 2% share of the OS market, less than 0.2% of computer users will ever see an IEC prefix, and even then, it would only be when partitioning their drive (if they use the GUI-based tool). This article had been complete crap to promote the adoption of the IEC prefixes. That is absolutely verboten. The IEC prefixes are an asterisk on reality. Greg L (talk) 20:07, 2 May 2011 (UTC)


 * There has been much discussion about this on this page. Feel free to read the above threads.  What you did when you deleted all that stuff from the article was take a well laid out article and reduce it to a stub.  The labels here were listed as a "controversy" because everyone is well aware that there is some confusion as to what the proper way to use these names are.  If I was to transfer a kilobyte of information on a 1 kilobyte per second data transmission line would it take 1 second or 1.024 seconds?  If you fail to define the unit "kilobyte" then that is left ambiguous.  A desire to remove that ambiguity is what has led virtually every scientific body in the world to accept the definitions given in this article.


 * There are many places across the wikipedia where you will find articles where an educated professional in a field has provided correct data even if it is not what is "widely known" throughout the world. This is one such article.  It clearly stated that "kilobyte" is frequently used to mean either 1000 bytes or 1024 bytes based on the context and then it explained what the scientific world has decided to use.  NPOV is achieved by reviewing what the professionals in the filed do, not by what you read in the Best Buy ads.  Cavebear42 (talk) 00:15, 3 May 2011 (UTC)


 * "Profesionals in the field" don't use kilobyte in the way the article was trying to promote. I think Greg's version more accurately describes how the real world is. Glider87 (talk) 00:35, 3 May 2011 (UTC)


 * I support Greg L's change. Personally, I would consider adding a couple of examples of the usage of "kilobyte"—to help newbie readers. The binary prefixes article (which is still linked) explains the ambiguity, so it doesn't need to be reproduced here in its own section. GFHandel . 00:44, 3 May 2011 (UTC)


 * They do if they want their paper published. Neither IEC nor IEEE will accept a paper which uses KB to represent 1024 Bytes.  I guess we could continue to promote the misconstructed views and contibute to confusion.  Maybe we should go and edit the articles on List of common misconceptions to make sure they tell people the "common man's" view rather than what scientists have accepted?
 * When I first came across this article, I shared your view. It bothered me to think that I had been incorrectly using KB for so long.  However since all the major standards organizations (IEC, IEEE, SI) have been on board for over a decade, isn't it fair to say that KiB is the accepted standard?  Can you show me a standards organization who disagrees? Cavebear42 (talk) 21:31, 5 May 2011 (UTC)
 * It isn't an "accepted standard" when the majority of modern reliable sources don't use it. Also the majority of "scientists" don't use the -bi nonsense. The JEDEC disagrees and still defines kilo to be a power of two. Glider87 (talk) 12:49, 7 May 2011 (UTC)

I have the following comments: Regards Lightmouse (talk) 08:46, 3 May 2011 (UTC)
 * The article suggests 'KB' is ambiguous but probably means 1024 byte. The article discusses 'kB' in a footnote. It's only when I read the footnote that I understand the problem. Can it be within the body?
 * Discussions of ambiguity almost always make articles difficult. The space formatting issue ranks much lower than the explanation of meaning. It's not mentioned in the metre article and I'm sure it's absent in most unit articles. Could we simplify this complicated topic by leaving the space issue out?
 * The article doesn't appears to address things from the point of view of a writer. I might come to this article as a writer seeking advice on how to minimise ambiguity, possibly to avoid a legal dispute. Which symbol is most unambiguous?
 * From a reading-psychology point of view, I think the space is important. Greg L's version looks like a much better one. I don't see why it was reverted. Tony   (talk)  09:28, 3 May 2011 (UTC)


 * This is one of just a few unit articles that need to focus on helping readers and writers with the 1000 versus 1024 problem. It's simply a matter of taking the interesting footnote material and elevating it to the body.


 * If the space issue needs mentioning then put it in the body text of each unit article.


 * When the IT specialists come back online, perhaps they'll discuss my other questions. Lightmouse (talk) 10:07, 3 May 2011 (UTC)


 * Per Lightmouse’s suggestion, I expanded the body text by incorporating a point that previously had been in a footnote. It may appear to be a stub, but there isn’t much to say about what kilobyte means. What the article mustn’t do is give undo weight to a “dispute” and “controversy” over the IEC prefixes where none exists. About the only place on this pale blue dot where anyone is talking about the IEC prefixes are here on the talk pages of Wikipedia, where proponents of the the idea want *keep the dream alive* and have articles give great play to a standard the computer industry is soundly ignoring. With a whole paragraph now dedicated to the IEC prefixes, this article gives more air time to the IEC prefixes than the Microsoft Computer Dictionary, Fifth Edition, which doesn’t even mention terms like “kibibyte.” As I stated in my above post, the previous version of the article was using citations to buttress assertions when the citations were actually saying no such thing. In one case, the citation wasn’t to an RS but was instead a blog quoting an e-mail. The citations were beyond-weak for the simple reason that great effort was being made to put lipstick on a factual pig and pass it off as prom date. Greg L (talk) 14:19, 3 May 2011 (UTC)
 * Kibibibibyte: the English language can be so ugly, can't it. Tony   (talk)  15:20, 3 May 2011 (UTC) PS should there be a reference to the MOSNUM section on the symbol and its formatting? If it's frowned on to have it in the main text, at least in the See also section?  Tony   (talk)  15:22, 3 May 2011 (UTC)
 * A link in the article’s body text referencing non-articlespace like WP:MOSNUM? Or in a hidden editors’ note? I know what you mean about how “1 KB” is superior to “1KB” insofar as cross-project consistency. Note though that in many marine or missile practices, one can read “The North Korean missile has a range of 3500 nm” (a practice that drives SI-purists out of their minds because “nm” righteously ought to refer only to nanometers in a good and logical world). So I don’t think we can achieve cross-project consistency; the real world has varying practices by discipline (and, as we see with “kilobyte”, inconsistencies within a given discipline). There may be other Wikipedia articles that mention “1024KB”, such as DRAM or DIMM, where the majority of the RSs use language like “2GB DIMM.” I suggest we leave this one alone and let the individual editors on any given article strike a proper balance. Some may want to ensure the body text of articles are consistent with the practices of the RSs. For still another topic, where the nuances of the topic and the practices of RSs vary somewhat, editors might think it better to observe the rule of SI in body text (as you suggest) and use “1MB” or “1KB” only in direct quotes. Did you notice how I managed to entirely duck the issue of embracing one practice or the other in the body text? There is no instance of “1024KB” or “1024 KB” to imply “This is the way it is *properly* done that Wesley Crusher would approve of.” All that is in the article are the pure facts regarding the formatting issue in the form of that footnote—as ought to be the case. Greg L (talk) 16:02, 3 May 2011 (UTC)

Lowercase k
This article still doesn't clarify the primary issue for me. One of *the most* unusual aspects of kilobyte is that it can be seen with a capital 'K'. All other units (km, kg, kW, etc) use a lower case 'k'. I've asked this question twice in different ways and I can't see the answer so here's the third time of asking in yet another way. The body text explains what 'KB' *probably* means. Can anyone write body text saying what 'kB' *probably* means? Lightmouse (talk) 17:56, 3 May 2011 (UTC)


 * It’s a valid question, I suppose, but one would very seldom see “kB” in RSs (and therefore the real world). Perhaps it is being used on Wikipedia by SI purists who perceive little need to conform to our requirements to follow the RSs and who feel using Wikipedia as a venue to exercise “leadership by example” is somehow a good thing? If so, that’s not a good thing. In actual practice (the real world), *IF* one found an example of “kB”, its meaning would adhere to the same reality as with “KB”: it depends upon context. For most purposes, it means 1024 bytes, but with mass storage, it probably means 1000 bytes. Since “kB” isn’t used by the RSs, I don’t see how this issue can (or should) be directly discussed in the article since it seems exceedingly unlikely one would be able to find a citation to an RS discussing such a topic. On Wikipedia (an altogether different matter from the real world because advocacy of cool-beans ideas in contravention of WP:SOAP often occurs here), instances of “kB” might well be intentionally a construct of certain editors that is intended to uniquely denote 1000 bytes. This would be unconventional (a replacement for the already-failed IEC proposal?) and would essentially amount to a unique house style unfamiliar to a general-interest readership and unused by RSs. Greg L (talk) 20:09, 3 May 2011 (UTC)

This unit is very unusual in using upper case 'K'. The absence of an explanation of lower case 'k' is like the curious incident of the dog that didn't bark.

Note Wikipedia isn't just for people who passively read non-WP stuff. It's also for writers of non-WP stuff. Many writers are taught to use lower case for km, kW, kV, kg. So it's not a surprise to for 'kB' to be out there. The format 'kB' may be rare but so is 'kibi' and it's hugely disproportionate to avoid explanation of the former and fight a holy war over the latter. Here are some examples of kB:


 * http://cnst.nist.gov/epg/Pubs/pub.html
 * http://cs110.wellesley.edu/lectures/compModel/how_computer_works.html
 * http://cxc.harvard.edu/mta_days/mta_db/pgsql/doc/html/kernel-resources.html
 * http://esh-docdb.fnal.gov/cgi-bin/ShowDocument?docid=1389&version=1
 * http://home.fnal.gov/~timur/srm/CMS.SRM.Profiling/All_Objects.html
 * http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=993110
 * http://uk-air.defra.gov.uk/reports/empire/ghg/ghg3.html
 * http://webarchive.nationalarchives.gov.uk/20090618140443/code.reddit.com/
 * http://www.carleton.edu/campus/ITS/staff/sfox/shrink.html
 * http://www.cs.colostate.edu/~cs155/wiki/HW4
 * http://www.cs.indiana.edu/classes/a216/factsheet.html
 * http://www.da.mod.uk/publications/sf-publications/atct_topic_view?b_start:int=300
 * http://www.luton.gov.uk/internet;internal&action=search?Submit=&startIndex=0&searchstring=data
 * http://www.nato.int/cps/en/natolive/audio.htm
 * http://www-oss.fnal.gov/projects/disksuite/tools-eval.html
 * https://buildsecurityin.us-cert.gov/bsi/articles/knowledge/coding/770-BSI.html
 * https://svn.pnl.gov/gcrm/wiki/tau

If 'kB' is incorrect, or is identical in meaning to 'KB', or is 'leadership' by 'SI purists', or all of the above that's fine by me. I'm sure I could add some text based on what Greg's written but I'd prefer it to be written by somebody with more specialist knowledge. Is anyone willing to provide body text that explains what kB means? Lightmouse (talk) 23:13, 3 May 2011 (UTC)


 * Lightmouse, if “KB” was used 24 million times to “kB” being used 23,000 times, and someone was inclined to spend the effort to add one hundred of the “kB”s here, that impressive display of blue doesn’t prove much of anything other than look like a widespread practice. The question of what the majority of the RSs practice isn’t too hard. Secondly it isn’t hard, all on your own, to figure out how some of these links calculate “kB”. Let’s take your fourth link: http://esh-docdb.fnal.gov/cgi-bin/ShowDocument?docid=1389&version=1, which is Fermilab (a bunch of scientists who would be loath to bastardize the prefix k by making it uppercase). See that first link at the top that is titled “Review of Safety Shoe Purchase.pdf”? It has a file size Fermilab declares to be “(225.2 kB)”. Download it and look at its properties. It comprises 230,630 bytes. Divide by 1024 and you get 225.2246 kilobytes—just like Fermilab says. None of this should come as a surprise because Merriam-Webster says it kilobyte is 1024 bytes and so too does Oxford, which also adds that it is properly uppercase K (though they dick up and say that it can be lowercase b, which RSs reserve for kilobit). This bit of evidence you dug up underlies the lawsuit against the manufacturers of hard drives because file sizes have historically used multiples of 1024 for M and G, but the hard drive manufacturers have stuck to decimal use for megabyte and gigabyte (and kilobyte by association). I fully expect someone will be able to find one of your above-mentioned Web sites using the decimal definition for lowercase k. When or if that is occurs, the observation will reinforce what we already know : the well-documented fact that “KB” (or “kB”) is ambiguous and means either 1024 or 1000 depending upon context. Why are you making this so complex? There is nothing unusual about the value of kilobyte if someone uses the symbol kB or KB for it, except that the majority of the RSs use “KB”. If you want something in the article about how “kB means the same thing as KB except that they use the lowercase k, I think we need to find an RS that states as much; my finding that Fermilab uses “kB” as 1024 bytes is anecdotal and can probably be easily refuted by someone running to some Ubuntu users’ group bulletin board (which would be equally anecdotal). So please, find a suitably authoritative RS and then add this to the article. It would be nice, though, to have some sort of stats on whether the lowercase k is worth even talking about . At what sort of percentage of using “kB” would be sufficiently significant that we don’t run afoul with WP:WEIGHT? I duknow… 10 percent? I suggest you pick a number that simply passes the *grin* test for most editors active on this subject. Whatever we do, let’s just stick to the core principals of Wikipedia, which is to follow the practices of the majority of most-reliable RSs, not violate WP:SOAP or WP:WEIGHT by talking at great length about a proposed standard that failed to gain hardly any traction in the industry (which I know you aren’t trying to do), and if we are tempted to add something that is challenged or is likely to be challenged, then cite it and cite it well. Greg L (talk) 00:36, 4 May 2011 (UTC)

P.S. Here is some great fun for the sort of stuff one can “prove” by finding Web sites. Here is Data Capacity Converter Online. It shows the capacity of various forms of media storage. Not surprisingly, a kilobyte is 1024 bytes. Interestingly, it uses TB, GB, MB, and kB (lowercase). For even more interesting results, we have this from WISE GEEK: How Much Text is in a Kilobyte or Megabyte? Again, it shows all powers of 2n10 (1024n&thinsp;) so that kilobyte is 1024 bytes. Not so interesting? It has this series: kB, mB, gB, and tB. All lowercase. The sweet Internet… Greg L (talk) 02:33, 4 May 2011 (UTC)

bps
In my experience, working in the telecom industry, bps normally denotes bits per second, not simply bits. It's most often used to show bandwidth or flow rate on telephony equipment. (10 Mbps - 10 megabits per second).

Steggall 15:52, 20 May 2004 (UTC)


 * I know, but I was referring to how you don't see b/s or bits/s in common usage, nor do you see Bps in common usage. — Preceding unsigned comment added by Elektron (talk • contribs) 17:17:54, 1 June 2004 (UTC)

My edits (June 2, 2004)

 * [] are brackets, not parentheses (And in any case, don't mix [] and ).
 * The SI kilo is no more 'correct' than the base-2 kilo, since byte isn't a SI unit (SI should only deal with bits, if anything, since 28 is completely arbitrary). Also, as stated in binary prefix, B is the symbol for bel.
 * It's not so much that the machine language is in binary, but that you can bitshift right and bitwise AND to deal with powers of two.

There are other tweaks that should probably be done as well (the article doesn't read very well). Elektron 16:49, 1 June 2004 (UTC)


 * Hi, Elektron - let's work on this together, as you seem to be knowledgable on the subject...
 * &bull; thanks for the this edit ("it becomes more apparent"), I fully agree, makes the sentence run smoother.
 * &bull; Where I come from, we use to denote units by [kB] brackets (but never mind). But what did you mean with "And in any case, don't mix [] and "?
 * &bull; 1st paragraph: I'd like to drop the mention of the 'telecommunication engineers' altogether, for when have they really used the unit kilo'byte'; I've thought they dealt with bit, kbit, Mbit, Gbit (and bit/s, kbit/s, ...).
 * &bull; "describing storage capacity and memory size of computers (as it is a power of 2, making easy for computers, which work in bianary, to manipulate)" I don't see a reason to delve into details about the working of registers, and that the mere mention of computers being 'binary' machines should suffice.
 * &bull; "The SI kilo is no more 'correct' than the base-2 kilo, since byte isn't a SI unit" To read "kilo" as "1,000" is the correct use of this SI prefix, that's what I meant, and it is a SI prefix, regardless whether or not the unit is SI or non-SI.
 * &bull; "There are other tweaks that should probably be done as well (the article doesn't read very well)" I didn't say it was perfect (far from that) but I did make an effort; if you think it reads not well, by all means go ahead and do make suggestions... :)

--Palapala 05:59, 3 June 2004 (UTC)


 * I was referring to "[Telecommunication engineers have used it all the way.]", which should definitely be (especially since  is used in other parts of the article. While [units] is probably a good idea, I think most people just use italics for variables and normal text for units. Unfortunately, it's very hard to write in italics (and actually look like you meant it to be in italics).
 * Certainly, G is a SI prefix, but when you call 109 bytes a GB, you adhere neither to common usage (where GB is 2 30 bytes) or SI usage (where GB is a gigabel). It's like calling 2 million pounds a kiloton &mdash; you adhere to neither standard.
 * We should also merge kilobyte, megabyte, gigabyte, terabyte, petabyte, exabyte, zettabyte, yottabyte, binary prefix (megabyte to yottabyte are very similar, and should never have been separate pages in the first place). I might get around to this sometime next week.
 * I also should get around to fixing integral data type. PowerPC assembly uses 'byte, halfword, word, doubleword' (even for 64-bit processors). "short" and "long" are more consistently 16 and 32 bits. But that's off topic. --Elektron 08:48, 5 June 2004 (UTC)

Binary Usage exponents
I'm thinking the exponents on the "Binary usage" column are wrong. I think the entry for kB should be 1024, not 10243, and the entry for MB should be 10242, not 10246, and so on. I think that whole column should be identical to the "Value" column in the "Binary" column on the page linked to via the "Binary usage" heading (the page: Binary_prefix). I didn't want to make this change immediately in case I'm missing something and since it looks like it's been that way for quite a while. — Preceding unsigned comment added by 50.65.34.248 (talk) 05:38, 24 June 2013 (UTC)

KB vs KiB
KiB is defined for binary use in all standards: NIST, IEC, CENELEC, JEDEC, ANSI, IEEE, DIN EN, ISO. KB is only defined by JEDEC. --DrSeehas (talk) 17:36, 5 September 2013 (UTC)

Modern Times?
First paragraph: Can someone who knows what "in modern times" is supposed to mean, in the context of this article, replace the phrase with something less subjective? Does it mean since the last ice age, since I had my last cup of coffee, included in a 2006 Bob Dylan album, or something else? Some dates would be nice. 86.140.5.154 (talk) 23:29, 9 November 2013 (UTC)

Size comparison
Is there any point to this section? --Ef80 (talk) 22:38, 8 November 2013 (UTC)
 * As it stands I don't think it's helpful. If you don't have any suggestions as to how to improve it (and I don't), go ahead and remove it.  Garamond Lethe t c  23:24, 8 November 2013 (UTC)


 * In general, I think it is useful to have an indication of what X kilo/mega/gigabytes means in "practical terms". Currently we have sections in various articles for the individual amounts, eg:
 * Kilobyte
 * Megabyte
 * Gigabyte
 * etc
 * Possibly they should all be replaced by a link in "See also" to Units of information. Mitch Ames (talk) 01:49, 9 November 2013 (UTC)


 * Done - for this article, anyway. 09:47, 12 November 2013 (UTC)

1000bits vs. 1024bits
At http://www.webopedia.com/TERM/K/kilobyte.html the naming convention is sugested that KB means 1024 bytes and kB means 1000 bytes.

At http://www.computerhope.com/jargon/k/kibibyte.htm they define kibibyte as meaning 1024 bytes.

At http://searchnetworking.techtarget.com/gDefinition/0,294236,sid7_gci499008,00.html the term kilobyte is also used to refer to 1000 bytes. — Preceding unsigned comment added by 198.29.2.119 (talk • contribs) 06:36:03, 21 April 2005 (UTC)

I'm sorry, but just right click on a file in windows, go to properties, look at size on disk. That's the standard. And it demonstrably shows the usage of 1024 bytes per kilobyte. Be consistent with the obvious if nothing else. — Preceding unsigned comment added by 72.205.235.66 (talk) 18:14, 24 April 2014 (UTC)

kilowords - "K" ambigious
Not sure if I should discuss this elsewhere (some Wikiproject Computing page or maybe the MOS pages?) or at the page I provide (but the issue is general and that is just one system).

See: Compatible Time-Sharing System: "32,768 36-bit word banks of core memory instead of the normal one; users had access to 27K of the total 32K". My first thought was to add just a (nonbreak) space (for the WP:MOS) "27K"->"27 K". From the context, and if you know computer history, you would know that K there doesn't mean KB (or Kelvin..). Linking to Kword is not possible as taken for KWord, change that? I assume changing to lower case k is also wrong as K means 1024, just not 1024 bytes. comp.arch (talk) 09:19, 15 January 2015 (UTC)


 * Hello! It's somewhat common to shorten "4 KB" to "4K", for example, using only "K" without the actual unit and leaving everything else to the context.  In other words, even if the example you've provided is left as-is it should be understandable enough, IMHO. &mdash; Dsimic (talk | contribs) 09:31, 15 January 2015 (UTC)


 * I disagree. As it stands, the example is both confusing and misleading, partly because "K" is sometimes used to mean 1024 bytes and sometimes just 1024.  In this case, it seems to mean 1024 36-bit words.  Adding a non-breaking space would help in my opinion, but only if it is explained somewhere that K is used as an abbreviation for 1024 words.  Dondervogel 2 (talk) 11:48, 15 January 2015 (UTC)


 * Sure thing, providing an additional explanation can never hurt. Though, I'm not sure how to do that in this exact example, so that the actual compaction achieved by using "K" isn't subdued by the additional explanation? &mdash; Dsimic (talk | contribs) 12:05, 15 January 2015 (UTC)


 * There is no compaction achieved by using "K" to mean 1024 unless you know in advance that K means 1024. Do all readers know that?  I think it is computer jargon that should be avoided, and if it cannot be avoided it must be explained.  Frankly, the example seems to me tailor-made for the kibi- prefix.  Once "word" is defined as 36 bits, then you would have 27 kibiwords and 32 kibiwords.  Problem solved.  If this obvious solution is ruled out, for whatever reason, perhaps follow the example of "186K words" used further down the article and say 32K words, 5K words and so on. Dondervogel 2 (talk) 12:47, 15 January 2015 (UTC)


 * ... something like ? &mdash; Dsimic (talk | contribs) 13:09, 15 January 2015 (UTC)


 * Yes, that's a big improvement. I added a link to Word (computer architecture) as well. Dondervogel 2 (talk) 13:22, 15 January 2015 (UTC)


 * You're right, with those changes it's much more readable for an encyclopedia article.  just gilds the lily. :) &mdash; Dsimic (talk | contribs) 13:30, 15 January 2015 (UTC)

See also at Wikipedia_talk:Manual_of_Style/Dates_and_numbers as I found another case I'm not sure what to do about.

IEC unambiguous ?
The claim that the IEC standard is unambiguous is not fully correct: it does provide an unambiguous prefix for 1024 (kibi), but it does not provide an unambiguous prefix for 1000. The choice 'kidi' spring to mind. 2A01:E35:2F45:9A0:BA76:3FFF:FEF7:E4D5 (talk) 19:37, 15 December 2014 (UTC)


 * If you're complying with the IEC rec then k is unambiguous for 1000. k should only be used for 1000, and ki of course only for 1024. The IEC rec says so explicitly, you don't even need to go to SI for that. Jeh (talk) 21:05, 24 February 2015 (UTC)

On how forceful to be with the binary/SI prefixes
Just as a reminder, Wikipedia actually doesn't generally prefer the strict interpretation of KB vs KiB (using alternate units, pretending that K solely means 1,000, etc.). By which I mean, according to the Manual of Style. That said, obviously both need to be addressed in the article, but it is not appropriate to say anything so bold as, ""In the SI prefix kilo- means 1000 (103); therefore one kilobyte is 1000 bytes." That's factually false. Depending on the context, one kilobyte is 1000 bytes or 1024 bytes. Maybe it should only be 1000, but it is either. If you want to change all of Wikipedia's policies on this (including the MoS; you'll have to make that agree as well if you want consistency across articles), then take it up at the village pump. Also, correcting the article to be descriptive, rather than prescriptive'' is clearly not vandalism. Don't pull that "RVV" crap again. Not a request. 108.161.119.154 (talk) 19:54, 24 February 2015 (UTC)
 * Hiding behind IP addresses is abuse of editing privilege, which is what apparently is practiced here. Independently, the article makes it perfectly clear what the situation is and this has been hashed over and over. The standards organization prescribe what is the accepted interpretation and to ignore it is senseless. All standards first define the proper meaning of a term, and not the confusion of it. Kbrose (talk) 20:16, 24 February 2015 (UTC)
 * Hiding behind IP addresses? Both false and unrelated to the improvement of the article.
 * Also, did you refresh your memory on the MoS? Would you like a link?
 * What is the precise problem with the previous phrasing? Because pretty much every (reputable) source both acknowledges the attempts of standards bodies to use less ambiguous terms, but also say that in practice it tends to be more contextual. 108.161.119.154 (talk) 20:27, 24 February 2015 (UTC)
 * Actually, for the sake of anyone else who might read this, Wikipedia has had a long history of disputes over the various prefixes to use (and meanings thereof). The policies are littered across the various guideline pages, etc. But the easiest one to keep track of (because it pretty much always nicely consolidates the themes from the others) is the Manual of Style.
 * In this case, because we're actually discussing the units themselves, it naturally makes sense to also explain kibibyte (in spite of it not being a widely-adopted unit, and Wikipedia's guidelines normally advising against its use). However, to phrase the article in such a way as "a kilobyte is 1000 bytes. fact." clearly violates that spirit. The previous (current?) phrasing gives a nuanced approach, which is preferable to saying, kilo means 1000, therefore a kilobyte is 1000 bytes. That latter sentiment is neither correct nor helpful.
 * It's true that this is constantly rehashed (similar to how UK vs US spellings are constantly rehashed), but it's really only a problem because people sometimes forget to address those wider-reaching policies on the appropriate guideline pages, rather than article pages. That is, until the MoS (or other guideline pages) changes, we can't just go ignoring them. 108.161.119.154 (talk) 20:42, 24 February 2015 (UTC)


 * "In the SI prefix kilo- means 1000 (103); therefore one kilobyte is 1000 bytes." The latter part of this sentence would be perfectly fine if "in this system", or similar, was appended. My interpretation of this version was that that was the intended meaning. (The first word of the sentence should have been deleted, however.) Jeh (talk) 20:59, 24 February 2015 (UTC)
 * That's a good phrasing. I'm personally partial to retaining the acknowledgement that it's highly contextual, but obviously that's subjective.
 * I'll leave it to you (or someone else) to decide what to do about that part. Onto the next part of the sweeping change: considering the "K" is being explained, I think it appropriate to still explain the B (as opposed to b: bit). Opinions? 108.161.119.154 (talk) 21:12, 24 February 2015 (UTC)
 * I changed it to your suggestion, Jeh. I also condensed that whole topic into a single paragraph. I'm not sure why those six sentences about what 'kilo' can mean needed to be spread across four paragraphs. Heck, upon rereading it, I'm not even convinced that the lede needs to have anything more than just that first sentence (the very next section dives straight into defining the unit, including the variant interpretations). But, as-is, I'd say this is still better. 139.57.240.18 (talk) 23:27, 4 March 2015 (UTC)
 * I put the graf breaks back. I feel that each break discusses a different aspect of the term and the breaks tell the reader "we're now talking about something else". The sentences did not flow into each other logically previously, therefore they should be separate grafs, however short. Perhaps it could be reworded to use fewer grafs, but I think at least four are required: The "definition" sentence, the SI meaning, the traditional binary meaning, and the IEC prefix version. Jeh (talk) 23:36, 4 March 2015 (UTC)
 * I think we both had the same basic goal. For my own part, I saw the entire lede as really being three parts: a basic definition that it's a unit of information, explanations of what the kilo/k meant, and then a note on the B. I guess I'm not entirely convinced it's even necessary, because (other than the first and last sentences) the entire lede is just explaining the definitions... before the first section which is also definitions. It has a feeling of, "before we can talk about definitions, we first need to talk about definitions." 139.57.240.18 (talk) 23:47, 4 March 2015 (UTC)
 * I see your point - not counting the "examples" section, the lede is nearly as long as the "Definitions" section. Suppose we shortened the lede to:
 * The kilobyte is a multiple of the unit byte for digital information. It refers to either 1000 bytes or 1024 bytes, depending on usage and context. The abbreviations for these are kB and KB, respectively, although this distinction is not always observed.
 * I think we can leave the info about the ISQ and the KiB to the definitions section. Anything else that the lede now says that isn't in the rest of the article can be moved there (and shouldn't have been in the lede in the first place). Jeh (talk) 00:53, 5 March 2015 (UTC)
 * That's pretty dang good. 139.57.240.18 (talk) 01:25, 5 March 2015 (UTC)

Did that to the lede, merged the stuff that needed to be merged into the body, etc. Please comment. Jeh (talk) 21:58, 5 March 2015 (UTC)
 * I like. I like very much!
 * It's clear, concise, and readable. Relatively small changes, but the article is substantially better for it. 139.57.240.18 (talk) 22:53, 6 March 2015 (UTC)
 * Thanks! Almost all of the current text was there already. For the most part I just rearranged things into a better sequence and added 'glue' where necessary. Jeh (talk) 12:30, 7 March 2015 (UTC)

Use of KiB in scientific publications
Who says their not widely used? Dondervogel 2 (talk) 21:52, 8 March 2015 (UTC)


 * You have it backwards. You have to provide a RS that says that they are. 1400 hits over 16 years of scientific papers is not "widely used". And anyway, raw Google hit counts cannot be used as a source; see WP:SET. And be careful of OR. You could run 16 of the above searches, one per year, and show a trend, but that could be considered OR. It would be far, far better if you could find a paper where someone had done that research, and cite the paper.


 * Also, this article is not about binary prefixes, so this is perhaps not the place to raise their banner.


 * You and I are both proponents of the IEC binary prefixes. (I think we need a userbox for that.) So we must be extra diligent about hewing to all of the relevant policies and guidelines, lest our efforts to legitimately document them here be reverted due to perceived bias. Besides WP:RS and WP:OR, there are WP:V, WP:NPOV, WP:UNDUE, etc. Jeh (talk) 22:58, 8 March 2015 (UTC)


 * I disagree. It is self-evident that the binary prefixes are not widely used in the lay literature.  There we agree.  It is not at all self-evident that they are not widely used where it matters.  In fact the opposite is the case: they are widely used when disambiguating, but I agree with you that a RS would be needed to support such a statement.  The best solution, if the article were about binary prefixes, would be to provide examples of where used and where not.  As it is not about binary prefixes the best solution is to delete the controversial statement. Dondervogel 2 (talk) 08:08, 9 March 2015 (UTC)
 * In fact I think the whole section entitled "The kibibyte" should go. It can be replaced with a single sentence at the end of "1024 bytes". Dondervogel 2 (talk) 08:10, 9 March 2015 (UTC)
 * I don't think a single sentence is enough, but we can start by deleting the section head, and putting the Binary prefix WL back in See Also. Jeh (talk) 08:37, 9 March 2015 (UTC)
 * Did that, and an experimental "shortening" - just try it on for size, don't mind the color :) Jeh (talk) 08:45, 9 March 2015 (UTC)
 * Yes, this is better. The Contents list looks weird though as a result. Dondervogel 2 (talk) 09:06, 9 March 2015 (UTC)
 * Fixed. Jeh (talk) 10:06, 9 March 2015 (UTC)

Why is K 1024 for TV measurements but 'now' supposedly not memory
If a K is 1000, then surely a 4K TV should be 4000 wide rather than 4096? If you think I'm referring to UHD, I'm not, I'm referring to the industry standard 4K from the Digital Cinema Initiatives. Zhulien (talk) 16:54, 1 October 2015 (UTC) — Preceding unsigned comment added by 60.242.142.84 (talk)

Because unfortunately the IEC agenda of SI prefix definitions is being pushed by a minority against the opinion of most to all of the digital/computer/IT/electronics professionals along with current educational practice and common parlance. Unfortunately I think it reflects badly on WP when an article fails to be balanced with respect to majority opinion and public use, and instead reflects the most persistant editing group. IEC invents standards all the time that fall by the wayside. In this case these prefixes (kibi etc), which I never encountered in 30 years of microcontrollers, electronics, and engineering, will likely never gain traction because binary electronics is based on powers of two. Your 4K TV is an example of such implementation. 12-bits of address map to 4096 pixels. What it amounts to is a pedantic view as to the "correct" use of 'kilo' as a prefix, and/or that the IEC standard somehow trumps the accepted and widely used JEDEC standard from the industry that invented the term 'kilobyte'. What the article should say first is that kilobyte is a digital electronics term refering to 2^10 = 1024 bytes, and second that there is a (rarely used) IEC standard that wishes to rename this to kibibtye to avoid confusion with the SI prefix of kilo=1000. I'm just glad I'm not a millipede. Mike163 (talk) 23:33, 12 January 2016 (UTC)


 * Except that "kilo" does not always mean 4096 even in the computer industry. See the Binary prefix article. Jeh (talk) 02:06, 13 January 2016 (UTC)


 * I'm not saying it never happens. Occasionally a company will think it can boost its apparent capacity using SI not powers of 2.  Just that the vast majority of public and industry usage remains on the 2^x they know.  Try some of the alledged linked evidence from Binary prefix: none of the HDD links are still valid.  Show me a single semiconductor data sheet using the terms.  WP should not reflect the future desires of a niche: it is here to educate and inform about today. Mike163 (talk) 21:08, 15 January 2016 (UTC)


 * Your claim of "occasionally" is erroneous. (And common to many people who rail against claims that "K" should ever mean 1000 in the computer industry; they are often unaware that cites of RAM capacity are much more the exception than the rule.) The hard drive industry has nearly always used SI prefixes for drive capacities, going back to the very first disk drive, the IBM 350 RAMAC which held "5 million characters" (which meant 5,000,000). Same with tape storage capacities. And DVD and Blu-ray disk capacity (but, oddly, not CD). And storage transfer rates. And data transfer rates over network interfaces, over buses, and even to and from semiconductor memory. Even on devices like USB storage "keys" the "GB" means 10 to the 9th. (Sorry about the stale links that back this up. Many of them were to product data sheets and those do tend to get replaced as product lineups change. I'll see about updating them.)
 * A significant number of editors here, myself among them, would prefer to use the IEC prefixes in many articles about e.g. operating system memory usage and layout because they're unambiguous. However after much debate (some of it quite acrimonious) we arrived at the present wording of WP:MOSNUM, which precludes such usage unless most of the sources for a particular article use them. So in the articles about e.g. RAM (and UHD TV) our articles reflect our sources (like JEDEC documents) and so we do use the traditional binary prefixes. Correspondingly, articles about hard drives and buses use the SI prefixes because that's what the sources for those articles use.
 * But the sources for this article, since it is about the unit, are the documents from the standards bodies that define the unit. We would be completely remiss in ignoring them, and they do include completely sufficient support for the IEC prefixes. As for the traditional binary prefixes, this article doesn't pretend that K never means 1024 in the real world; in fact we note here that it often does. Same for the articles on "Megabyte", etc.
 * This is, as I said, the state of WP consensus on the matter. If you want to revisit it the place to do it is at the MOSNUM talk page. I doubt you'll get a warm welcome; as I said, the previous discussions on these matters were quite heated at times and I don't think anyone wants to re-open them. Jeh (talk) 06:14, 16 January 2016 (UTC)
 * Thank you for proving my point. WP doesn't reach Consensus decision-making on acrimonious points, one side wears the other down.  Look at the structure of those RAM chips: it's binary addressed blocks in 2^n format, as are the micros that access it.  Magnetic data stores are the exception.  I gave up on the argument because a minority of IEC supporters are not going to change the opinion of the public, the binary physics of digital electronics, the acceptance of the older JDEC standards and the size of products offered to the market (ipad, TV etc).   Good luck with your view.  I fear you can expect to be correcting people indefinitely.  Mike163 (talk) 18:08, 2 October 2016 (UTC)

"4K TV" is not a measurement, but a marketing term. They would have called them the same even if the screens had 3840 pixels, or 4098 (1366 x 3). Get real. Kbrose (talk) 21:11, 2 October 2016 (UTC)
 * by "get real" do you mean "agree with me"? It's another example that the majority are not pendants to the extent that they need 4k to be exactly 4000.  Which is exactly where this thread started, so I'm out.  Mike163 (talk) 07:30, 3 October 2016 (UTC)

1024 v 1000
I really think it should be clear to us that there are two alternative definitions of what a kilobyte is. Different firms adopt different definitions. I suspect most people think a kilobyte is 1024 bytes. It also the definition used in most dictionaries. More recently there's been a trend toward 1000 bytes. The article should not seek to declare one view to be correct and the other wrong. — Blue-Haired Lawyer t 17:19, 1 October 2016 (UTC)

Even Wikipedia's manual of style says that terms like kilobyte, megabyte etc... are ambiguous. See WP:COMPUNITS. — Blue-Haired Lawyer t 17:51, 1 October 2016 (UTC)


 * The article already makes the definitions clear. WP articles also recognize the ambiguities, and this has been discussed extensively. The definitions of the standards organizations, however, should be the prevalent definitions. Kbrose (talk) 18:07, 1 October 2016 (UTC)


 * The article never says that one view is correct and the other is wrong. The article carefully gives examples of KB = 1024 bytes, notes that most common operating systems (with the sole exception of Mac OS X) display disk and file sizes using that convention, etc. But kB = 1000 bytes does have the weight of SI behind it and is used by a number of classes of products. . We would be remiss in reporting or implying otherwise. We could however reference the JEDEC (a trade organization of semiconductor, incl. semiconductor memory, makers) documents which do define KB = 1024 bytes. However we will need to note that these JEDEC standards disclaim applicability to any other class of products. Jeh (talk) 18:43, 1 October 2016 (UTC)


 * That my view has been "rehashed many times" just goes to show that there is little or no consensus for the current lead and content. A poor opening statement followed by unreliable sources that (even at that) barely justify the statements being justified by then. Please don't remove the NPOV template. As is quite clear from the template itself when it says:
 * "Please do not remove this message until conditions to do so are met."
 * They clearly haven't! — Blue-Haired Lawyer t 23:57, 1 October 2016 (UTC)


 * The point is that the present wording has been stable for some time, indicating consensus for it despite the controversy that is acknowledged by all. Consensus can change and if you consider a change to be necessary, please indicate here the precise change you wish to make, with justification, so it can be discussed and a new wording agreed ... or not. Dondervogel 2 (talk) 10:54, 2 October 2016 (UTC)


 * I don't agree that lack of change means consensus. It means greater stamina for the 1000 opinion on edits, I gave up.  The physics of binary electronics means 2^n is the actual implementation.  Speaking from my decades career semiconductor experience, the large cost pressure on semiconductors means that memory semiconductors now can quote 10-base but the structure of the chips is either uneven blocks or use of the spare memory for hidden functions.  Micros and digital electronics have binary address lines.  IMO 1000 can be considered universal when you can buy a tablet or phone with 10Gb, 50Gb, 100Gb and a micro with 10-base memory map.  Until then there is a place for 1024 still.  Mike163 (talk) 13:28, 2 October 2016 (UTC)
 * Nobody has ever disputed that memories and other hardware works on a binary basis, where powers of two are not only convenient, but necessary. That view is much beside the point. But digital memories are only one branch of digital technology, and in many other areas the metric prefixes have always designated their original decimal meaning. Transmission system engineers would perhaps be equally upset if someone came along and changed their units to a binary interpretation, as they for decades have understood kilo to mean 1000, and not 1024. In an ever more intertwined marketplace these separate realms of definitions are not adequate anymore and cause confusion. Even the JEDEC consortium acknowledges the problem and unambiguously states that the traditional usage is confusing. They also endorse the new units, but clearly cannot deny the volume of precedence of using metric prefixes (with binary meaning) for binary quantities. JEDEC does not define kilo as being 1000, or being 1024. They simply cite common usage, but their recommendation appears to favor the new binary prefixes as well. ALL international standards organization have done so. It is not just an isolated IEC ego-trip. The world of technology and commerce has always let the standards organizations lead the way to unification in measures. Refusing this development is simply silly. No physicist, for example, still creates the same noise wanting to use cgs units, rather than SI units. Frankly, WP is one of the few media where both sides of the issue are still contrasted with more objectivity than anywhere else. But the future seem pretty clear, it's just that most people are slow in changing, partly out of fear of becoming obsolete. If some old-timer engineer has never seen the new binary prefixes, it only means they don't follow modern technology outside of their immediate area of expertise. In reality the binary prefixes are rather common these days; when involved in modern software development one can find a new application using them, almost daily. Kbrose (talk) 14:16, 2 October 2016 (UTC)

The lead describes the world how you want it to be. It should describe it how it is. The lawyer with blue hair is right. --89.217.33.79 (talk) 11:36, 12 October 2016 (UTC)

Synthesis

 * "The International System of Units (SI) defines the prefix kilo as 1000 (103); therefore one kilobyte is 1000 bytes."

This may be what the BIPM would like to be and obviously many editors on this page agree, but its still synthesis. Just because the kilo- prefix is a 1000 is the SI does not automatically mean that a kilobyte is 1000 bytes. Neither byte nor kilobyte are SI units. — Blue-Haired Lawyer t 15:01, 27 October 2016 (UTC)


 * That would indeed be synthesis if it were based solely on the opinion of WP editors, but it's not. It is an inference made by national and international standards bodies around the world. (I just added a reference to support the inference, but one could add many others, among them BIPM, IEEE and NIST) Dondervogel 2 (talk) 17:22, 27 October 2016 (UTC)


 * It is true that SI does not define units for quantities of information, but SI does note that the SI prefixes (k, M, etc.) may be applied outside the contexts where SI's base units or derived units would be used. But where that is done, the SI-defined prefixes must retain their 1000-based meaning, and cannot be used as placeholders for other numbers, like 1024. In fact BIPM says literally that "for example, one kilobit represents 1000 bits and not 1024 bits"—see, p. 121. So there's no synthesis involved at all in the above-quoted claim, no doubt or ambiguity about it. Jeh (talk) 19:10, 27 October 2016 (UTC)


 * Yet for quite some time people thought of a kilobyte as being 1024 bytes and didn't think it as given that a the prefix kilo- must in all circumstances means 1000 and nothing else. Indeed many still think this. Just because a prefix has a certain meaning does not mean that there aren't some exceptions:
 * 'In-' is a prefix inverting the meaning of a word. Something is flammable when it can easily catch fire. Therefore something is inflammable when it is unlikely to catch fire.
 * The International System of Units (SI) defines the prefix centi as 100 (102); therefore a centipede has exactly 100 legs.
 * What the article ought say is as follows, because we have the sources to back it up:
 * "The International Electrotechnical Commission (IEC) define a kilobyte is 103 or 1000 bytes. This is consistent with the International System of Units (SI) which defines the prefix kilo- as 1000 (103)."
 * — Blue-Haired Lawyer t 00:58, 28 October 2016 (UTC)
 * We also have a "horse's mouth" source in BIPM, the organization that is responsible for SI, saying exactly this: "for example, one kilobit represents 1000 bits and not 1024 bits". You can argue forever that some parts of the computer industry use the SI-defined prefixes to mean powers of 1024, and no one is disputing that that's the case. But you can't argue that SI can be interpreted as allowing it, or that the claim "SI doesn't allow it" is synthesis... not unless you think you are free to ignore what BIPM says about SI. You're not.
 * btw, your "exceptions exist" analogies are very far of the mark. "in" is not an SI-defined prefix for a multiple of a unit at all. And "flammable" and "pede" are not units of measure of anything. And "centi" means 1/100, not 100. Haw! Jeh (talk) 01:55, 28 October 2016 (UTC)
 * Hectopede doesn't really have the same ring to it does it. All I am saying it that you can't claim it as a fact, that because SI defines the prefix kilo as 1000 that one kilobyte is 1000 bytes. It's not automatic! My proposed wording doesn't ignore the IEC or the SI. It makes direct reference to both. — Blue-Haired Lawyer t 13:08, 28 October 2016 (UTC)
 * It is accurate to state that the IEC defines 1 kB := 1000 B, but it is also misleading because mentioning only the IEC gives the impression this organization is out on a limb. I can support a statement along the lines "The main national international standards organizations all define the kilobyte as one thousand bytes", backed up by appropriate references to IEEE, NIST, IEC, BIPM. Dondervogel 2 (talk) 18:21, 28 October 2016 (UTC)

Reliable sources

 * 1) Conversion of Data Transfer Rate Units
 * 2) 1977 Disk/Trend Report Rigid Disk Drives, published June 1977
 * 3) SanDisk USB Flash Drive "Note: 1 megabyte (MB) = 1 million bytes; 1 gigabyte (GB) = 1 billion bytes."

The above are three of the five sources from the "1000 bytes" section. We have:

— Blue-Haired Lawyer t 15:01, 27 October 2016 (UTC)
 * A vague reference to a forty year old (!!) report to justify a statement that modern harddrive capacity is reported in decimal bytes;
 * A link to a Hobbyist's website to justify the prevalence of decimal bytes in data transmission; and
 * A link to the website of a single manufacturer being used to justify a statement that all flash drive manufacturers use the same standard.


 * We could update all of these and the results would still be the same: We'd find overwhelming evidence supporting these same positions. But you're the one challenging the current content - Why don't you go find evidence that modern hard drive, SSD, or flash drive capacity is quoted by the manufacturers and the trade and academic press using binary prefixes? Or that use of binary prefixes for transmission and other clock rates is common? Jeh (talk) 19:14, 27 October 2016 (UTC)


 * That's not how Wikipedia works, as you well know. — Blue-Haired Lawyer t 00:14, 28 October 2016 (UTC)


 * Really? You can just make whatever changes you want without sources to back up what you want the article to say? Jeh (talk) 01:37, 28 October 2016 (UTC)


 * I haven't made any changes which were't sourced. It's the original article that is poorly sourced. What you're basically saying is that I have to find sources to change what the article says but that you don't need any (or any good ones at least) to keep it as is. The policy is fairly clear: statements on articles should be sourced with reliable sources. If you manage to find the "overwhelming evidence" supporting the existing statements, all well and good. The article will be greatly improved because of it. — Blue-Haired Lawyer t 13:08, 28 October 2016 (UTC)


 * WP:SOFIXIT. You're the one complaining - you go update the references. Who knows, you might learn something. Jeh (talk) 16:58, 28 October 2016 (UTC)
 * I think BHL has a point here. The article could be improved by better references, especially for the flash drive.  When the computer industry refers to a memory of 512 MB, "everyone" knows that what they really mean is 512 MiB.  Everyone in the know, it seems, except flash drive manufacturers, who seem to take a perverse pleasure in confusing consumers.  I don't know of a better reference but I'll dig around.  If I find one I'll report back here.  Dondervogel 2 (talk) 18:38, 28 October 2016 (UTC)
 * All I found was one example that appears to use binary, not decimal arithmetic for flash memory°: "A typical 1 GByte device, the Samsung K9W8G08U1M [9], consists of two 512 MByte dies in the same package." Dondervogel 2 (talk) 19:09, 28 October 2016 (UTC)
 * That's a chip. A flash drive using such a chip would likely present a capacity of 1,000,000,000 at the interface. The remainder is used for overprovisioning. I believe that flash drive makers are thinking that since their devices show up in the OS as if they were hard drives, they should be advertised the same way. SSDs are also marketed in "decimal GB" (or TB). As for newer references, here's one. The language is a bit tangled but their statements "2. Disk Drive and Flash Memory Card Manufacturers commonly define a MB as one million bytes (exactly 1,000,000 bytes) and a GB as one billion bytes (exactly 1,000,000,000 bytes)." and "SanDisk defines 1 GB as 1,000,000,000 bytes. Operating Systems define 1 GB as 1,073,741,824 BYTES." are clear. Jeh (talk) 19:53, 28 October 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 2 external links on Kilobyte. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20080513155718/http://apac.sandisk.com/Products/Catalog%281349%29-SanDisk_Extreme_Ducati_Edition_USB_Flash_Drive.aspx to http://apac.sandisk.com/Products/Catalog(1349)-SanDisk_Extreme_Ducati_Edition_USB_Flash_Drive.aspx
 * Added tag to http://www.att.com/support_static_files/KB/KB24648.html
 * Added archive https://web.archive.org/web/20161026234911/https://www.orange.ro/recharge/internet.html to https://www.orange.ro/recharge/internet.html

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 09:01, 10 December 2017 (UTC)

Debate
The article opens with "The International System of Units (SI) defines the prefix kilo as 1000 (103); therefore one kilobyte is 1000 bytes.", which sounds like it was added by someone with strong opinions about the value of a kilobyte. Since the programmer community still largely insists on the value being 1024 bytes (whether or not the SI agrees with them), I argue the article should start by explaining the variance in definitions to resolve the ambiguity. Thoughts? Hppavilion1 (talk) 21:31, 9 March 2017 (UTC)


 * The fourth sentence in the lede isn't soon enough? I think it is. I don't think the article should start with "teach the controversy".
 * Did you know that data and clock rates very commonly use the decimal meanings of k, M, etc.? K=1024 is nowhere near as universal as many believe.
 * btw, you don't have to end a post here with "thoughts?" That's implicit. Jeh (talk) 11:55, 10 March 2017 (UTC)
 * The lede seems fine to me. It opens with the international standard definition of kilobyte, followed by one specialist use that uses the term in a different way. Two other remarks
 * The kilobyte is not an SI unit. It is defined by major standards organizations (ISO, IEC, IEEE, etc) as 1000 kB
 * I cannot agree with "the programmer community still largely insists on the value being 1024 bytes" because programming goes far beyond the realm of random access memory.
 * Dondervogel 2 (talk) 17:40, 11 March 2017 (UTC)
 * It's not really Wikipedia's job to teach (nor prescribe or otherwise soapbox) though; it's supposed to document, and a KB as 1024 bytes is too commonplace to start off with the somewhat contentious saying that actually it's 1,000. --Vometia (talk) 04:43, 4 December 2017 (UTC)
 * Nowhere do I see a statement that KB is 1000, or even 1000 bytes. I guess we must be reading different articles. Dondervogel 2 (talk) 08:46, 4 December 2017 (UTC)
 * per this definition, one kilobyte is 1000 bytes
 * Second line of this article. --Vometia (talk) 11:28, 4 December 2017 (UTC)
 * I have changed that wording a bit so it's less "this is the word from on high". Jeh (talk) 12:55, 4 December 2017 (UTC)
 * Thanks, that was my issue in the first place: "this respected organization defines it as this, therefor that is what it must mean" when actual usage very often disagrees. Hppavilion1 (talk) 23:52, 27 January 2018 (UTC)
 * That's the compromise version; the original was (as mentioned) "The International System of Units (SI) defines the prefix kilo as 1000 (103); therefore one kilobyte is 1000 bytes.", which seemed like a rather adversarial way to deal with the different definitions- akin to starting the article for 'child' with "Under US law, a minor is someone under 18 years of age; therefor, anyone less than 18 is a child", when different cultures and countries use different standards. This version at least says "per this definition", making it clear that this is the case when following one particular definition. Hppavilion1 (talk) 23:52, 27 January 2018 (UTC)

Hppavilion1, can you please provide an official "different standards" that supports kilo as 1024? 2A02:AA12:3141:1780:5163:3641:47F1:EA04 (talk) 15:15, 31 January 2018 (UTC)
 * The typically cited industry-specific source for that is one of the JEDEC publications. It's referenced at the Binary prefix article. Jeh (talk) 08:00, 23 February 2018 (UTC)


 * Merriam-Webster, among dozens of sites which confirm this standard computing usage. If I understand this aright, this older usage is now deprecated, but denying that it ever existed does seem foolish. Chiswick Chap (talk) 22:03, 22 February 2018 (UTC)
 * No one is denying that anything ever existed. "kilobyte = 1024 bytes" is a current use in some contexts, not at all "deprecated". 1000 bytes is also current usage, generally in different contexts. This is all thoroughly documented at the Binary prefix article. Jeh (talk) 00:40, 23 February 2018 (UTC)
 * Um, you are missing the point that has clearly been made in this discussion, not by me, that the text at the top of the article boldly and incorrectly implies that kilobyte is, was, and ever shall be 1000.000, a simple untruth. Chiswick Chap (talk) 07:44, 23 February 2018 (UTC)
 * Uh, no. I'm denying the point. I see no way to read the "text at the top of the article", however you define that, and derive the meaning you ascribe to it. The previous wording was problematic in that way but it's been changed, to the apparent satisfaction of User:Hppavilion1, who opened this discussion - at least, they did not raise further objections. Jeh (talk) 08:00, 23 February 2018 (UTC)

Mention of Microsoft Windows market share....
In this article about a fairly agnostic unit found in computing, is the mention of Microsoft's market share really of any importance? I have no problem with "The binary interpretation of metric prefixes is still prominently used by the Microsoft Windows operating system," but I don't see that their market share adds anything to the discussion. Furthermore, their market share might change over time, maybe quickly, maybe slowly, so eventually the article might need to be updated to keep that number current.

But, since this article is about an agnostic unit of measurement, perhaps that phrase can be eliminated without loss of clarity or relevant information? — Preceding unsigned comment added by Zonker.in.geneva (talk • contribs) 11:25, 17 January 2019 (UTC)


 * I did not find what you were referencing, so I believe it was removed at some point. I'm not about to dig through the article history just to add Template:Done. Your insight is appreciated, though; thank you. — Christopher, Sheridan, OR (talk) 03:58, 19 January 2021 (UTC)

Initialism mixed case
This article contains a mix of kB (lowercase k, uppercase B), and KB (all uppercase). Why? I've never seen kilobytes represented as anything but all the same letter case, and the floppy disk article matches this convention. Is this a typo, or a mix of conventions and inconsistent usage? — Christopher, Sheridan, OR (talk) 03:53, 19 January 2021 (UTC)
 * They mean different things. One kilobyte (1000 B) is represented by kB. One kibibyte (1024 B) is represented sometimes by KiB and sometimes by KB. Dondervogel 2 (talk) 07:10, 19 January 2021 (UTC)
 * In Kilobyte, the first three paragraphs gives all the examples that I'm familiar with for kilobyte. I guess I missed the 1998 defining of the kibibyte (KiB) and redefining of kilobyte ( KB kB). Thank you for explaining this, much appreciated. — Christopher, Sheridan, OR (talk) 09:29, 19 January 2021 (UTC)
 * I believe that it is an error. I often see people, software, companies write mhz, MHZ, mHz when they mean MHz. Sometimes I write Kg for kilogram. Lowercase k and uppercase k look similar when it is handwritten. When it comes to temperature, people often write 20 C and just forget to put the degree symbol. Some write 300 °K when the correct way to write is 300 K. People make mistakes. Don't be surprised. There is no such thing as KB. Sometimes they don't write the B and just write 100 K. Again, I would consider these as mistakes. Humans are sometimes inconsistent and sometimes make mistakes. Personally, I go with the 1024 B = 1 kB definition. Vmelkon (talk) 02:47, 15 May 2021 (UTC)