Jump to content

Talk:Hard disk drive/Archive 8

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 5Archive 6Archive 7Archive 8Archive 9Archive 10→Archive 15

Let's not lie to the readers

Could we at least have a version of the table that doesn't say something that is on the face of it stupid, like 1 terabyte = 0.905 terabytes? Or are we so used to clicking on "Start" to shut down the computer that we no longer care abuot the lies on the screen? Logical machines, these. --Wtshymanski (talk) 14:03, 11 April 2011 (UTC)

Nobody on this page has suggested using "1 terabyte = 0.905 terabytes". A table was proposed that was less confusing than the IEC prefixes version but was rejected by someone because it didn't use IEC prefixes. 220.255.2.59 (talk) 14:16, 11 April 2011 (UTC)
Sorry but I think that table is horrible. I think the original table is horrible too. "Here are the different ways of expressing one terabyte" is not a particularly useful thing to present here, as this article is not about "the terabyte". It's about hard drives. Furthermore, just as very few users will ever see one terabyte expressed as "931 KiB", very few will ever see it expressed as "1000000 megabytes" or "1000000000 kilobytes". Jeh (talk) 16:34, 11 April 2011 (UTC)

I also think both tables don't work. Either we design a new table that's acceptable to the different camps, and is actually useful for our readers, or do with text only in the capacity section of the article, using RS in either case. Despite my position that the table is MOSNUM compliant, since IEC Units of Measure (UoM) isn't the primary usage, my major concern is that the table has the logical absurdity of saying that 1000 GB = 931 GB, using exactly the same UoM. If we can get around that somehow, without using IEC units, I'm happy. When I first saw the table, I thought that using IEC units was a neat way of avoiding the 1000 = 931 problem, but I'm not an advocate for the usage of IEC units in WP. I think there is potentially some merit in the position that most general readers probably won't know what they mean. In any case that's an issue to revisit at MOSNUM, not here. However, most general readers will probably also not intuitively understand scientific notation or powers of ten notation either. I think Jeh is absolutely right in saying that ... this article is not about "the terabyte". An IEC free section that actually helps reader understanding also has the absolutely wonderful effect of making this whole brouhaha become moot, at least here. On a side note, I still don't quite understand how this became a hot button issue here. Using IEC units just isn't worth the drama, angst, user blocks, gallons of virtual ink, and edit wars. Nice postings, Jeh. — Becksguy (talk) 18:30, 11 April 2011 (UTC)

The only mathematically correct way to write this equation would be use distinct dimensional units, e.g. , where and . —Ruud 20:23, 11 April 2011 (UTC)
Well, yes, that's the advantage of saying "One terabyte =0.909 tebibytes", rather like saying "10 eggs = 0.8333 dozen eggs". Jimbo forbid we should have anything this straightforward in an article with this many editors. --Wtshymanski (talk) 20:38, 11 April 2011 (UTC)
Mathematically correct, fine. But the fact is that the OS that's on 90% of desktops does say "931 GB" for a "1 terabyte" hard drive. Jeh (talk) 22:03, 11 April 2011 (UTC)
Syntax versus semantics. Now if the almighty manual of style and their infallible interpreters would only allow us to explain to the readers why this is the case. —Ruud 22:09, 11 April 2011 (UTC)
Can this explanation really not be done without using IEC prefixes? If you do use IEC prefixes, then since literally 99% of the OSs on desktops and laptops do not display using IEC prefixes, how is this helpful? My opinion is that anyone who understands IEC prefixes won't be in need of any such explanation. I agree that the IEC prefixes would be helpful to the explanation if they were understood by the general readership. The case has not at all been made that they are so understood. Jeh (talk) 22:35, 11 April 2011 (UTC)
I don't see how you would be able to explain the two semantically different meanings of "gigabyte" without denoting them by two syntactically different symbols. Now we could make up our own set of symbols, as I did above, and explain their meaning. We could also use the IEC symbols and explain their meaning. —Ruud 22:45, 11 April 2011 (UTC)
The different number of bytes is already explained in other articles dedicated to this topic. As Jeh said just punt it to those articles. I don't see any benefit in trying to explain it here again when it has already been done.Glider87 (talk) 23:12, 11 April 2011 (UTC)
I agree: point out the issue in the text, give a brief description of why, and link to other locations so the interested reader can find out more. The introduction of non-standard terms (not mentioned anywhere else in the article) in a techo-table can only serve to confuse.  GFHandel.   23:20, 11 April 2011 (UTC)
You are severely underestimating the importance of the issue. The capacity discrepancy between advertised and reported hard drive capacity is something which has puzzled quite a few computer enthusiasts, has been the subject of lawsuits, and is discussed many reliable sources on computer hardware (e.g. Mueller's Upgrading and Repairing PCs and Messmer's PC Hardwarebuch, the former introducing the notions of mebibyte and gibibyte in the chapters on storage devices to be able to discuss the issue). It should therefore be discussed right here, in this article, not somewhere else. —Ruud 23:47, 11 April 2011 (UTC)

Or Wikipedia can communicate to our readership precisely as the computer industry has been doing for years, e.g. the following:

Terminology like gigabyte (GB) is ambiguous as it means slightly different values depending on the application. For many uses, such as computer memory (RAM), 1 GB equals 10243 bytes, which is 1,073,741,824 bytes. For other uses, notably storage such as hard disk drives and even solid-state drives (SSDs), 1 GB equals 10003 bytes, which is 1,000,000,000 bytes.

That this sort of approach is the one recommended by MOSNUM ‘twas no accident; it avoids confusing our readership with terminology they will only see here and won’t remember after they leave our pages.

Someone on this page wrote as follows:

One of the exceptions for binary prefixes is:

" in articles specifically about or explicitly discussing the IEC prefixes. "

This table compares SI and binary prefixes. That counts as "explicitly discussing" it.

Well, the editor may think that is what the community had in mind when it adopted that MOSNUM guideline, but the community obviously had in mind articles such as Binary prefix and Megabyte. The MOSNUM guidance the community adopted is exceedingly explicit on how to describe the magnitude of binary quantities with advise and examples as follows (note how the given examples include hard drives):

  • Disambiguation should be shown in bytes or bits, with clear indication of whether in binary or decimal base. There is no preference in the way to indicate the number of bytes and bits, but the notation style should be consistent within an article. Acceptable examples include:
    A 64 MB (64 × 10242 bytes) video card and a 100 GB (100 × 10003 bytes) hard drive
    A 64 MB (64 × 220 bytes) video card and a 100 GB (100×109 bytes) hard drive
    A 64 MB (67,108,864 bytes) video card and a 100 GB (100,000,000,000 bytes) hard drive

Just three editors are the shepherding authors behind our article “Timeline of binary prefixes”. One of them wrote here as follows: The software industry has been misappropriating SI prefixes to mean something they don't for decades. The ideal solution of course would be for software makers to actually use SI prefixes correctly, just like hard drive manufactures are now. Such editors believe the world would be far better off if the computer industry better embraced the use of the IEC prefixes so that terms like GB don’t have two different meanings. But the computing world hasn’t embraced them. Wikipedia follows the majority of the RSs. If you go into a computer store, you won’t find memory upgrades marked as “1 GiB SIM card”.

As for the IEC prefixes being a recommendation of a standards body, that doesn’t matter at all. The BIPM—the mother of all standards bodies and the one behind our current incarnation of the metric system, the SI—has steadfastly insisted that a space is always inserted between the numeric value of a quantity and the unit symbol and that this applies to the percent symbol (%) just as it does to the meter symbol (m). Thus, according to the BIPM, one properly writes 75 % and not 75%. Regardless that this is consistent and logical and recommended by the mother of all standards bodies, Wikipedia follows the practices observed by the way the real world works. Even though the use of 75 % would likely not confuse anyone and using this style would be futuristic, logically consistent, and compliant with a respected standards body, we don’t use it. And that is over a mere issue of *style* over which readers would easily recognize the measure and understand its meaning; it isn’t something much more substantive like a unit of measure that exceedingly few readers have seen before and will likely never see again after they leave our pages. Seriously… that’s really against the fundamentals of Technical Writing 101. That “kibibyte” is a swell idea isn’t sufficient to offset this colossal shortcoming.

It’s just that simple. We follow the advise of MOSNUM and we follow the majority of most-reliable RSs. With precious little if any exception, the hard drive industry and the operating system industry (Ubuntu at 1% being the exception) and the computer industry and the aftermarket RAM industry doesn’t use the IEC prefixes. So we don’t either. That is a bedrock principal of Wikipedia. Greg L (talk) 23:42, 11 April 2011 (UTC)

OK, how about we recast the section to be a summary section for the Binary prefix article per WP:SUMMARY. Write one paragraph without IEC units and without a table, but that abstracts the dichotomy by saying just essentially that HDDs are advertised and packaged as one size (e.g. 1 TB), and reported as various sizes, and linking the readers to the main article, SUMMARY style. Something like what Greg L suggested above that starts with: "Terminology like gigabyte (GB) is ambiguous...". In the Binary prefix article there is a whole section with five paragraphs about the decimal vs binary capacity issue in HDDs. In other words, here we get: no tables, no IEC units, no unnecessary verbiage and explanations, no 1000 = 931, and no squabbling. — Becksguy (talk) 00:12, 12 April 2011 (UTC)
Dude. That’s fine. So long as we follow the practices of the vast majority of the RSs and don’t use unfamiliar terminology the computer industry is soundly ignoring, we can do anything we want to encyclopedically communicate to the reader all about hard drives and their storage capacity. Greg L (talk) 00:15, 12 April 2011 (UTC)
  • Some RS's found with 5 seconds of googling: [1] [2] [3] [4]

    I'm not crazy about the table with the binary prefixes. I do think "gibibyte" should be mentioned in the section or else we've got an NPOV problem since multiple RS's do use the term. Most of all I believe that the idea here is to describe the topic as clearly as possible, using or not using binary prefixes as the article editors see fit. The recommendations of MOSNUM should be taken on board but MOSNUM does not have absolute authority. As it says in the infobox at the top of the MOSNUM page: "Use common sense in applying it; it will have occasional exceptions." It's up to the editors of this talkpage whether the affected section of this article should be one of those occasional exceptions.

    Note that whether SIM cards are advertised as "1 GiB" is immaterial; the RS's to follow most closely are the ones narrowly specific to the topic at hand, not the whole wide universe of data capacity specifications. The topic at hand is explaining the disparity between "1TB hard drive" (written on the outside of the box) and "937 GB" (reported as the same drive's capacity by OS software). There are tons of RS about that topic and a lot of them do explain it in terms of gibibytes. My own attitude towards binary prefixes is I've seen them used on occasion, they seem a bit precisionistic in most contexts since one can usually figure out what type of "GB" is actually intended. But this is one of the places where precision is required. Beckguy's suggestion of kicking "GiB" off to the binary prefix article is well-meaning but sub-optimal, since it is notable reference info about hard drives rather than merely being about binary units. So it belongs in the hard drive article. Ruud has explained it pretty well. 75.57.242.120 (talk) 02:26, 12 April 2011 (UTC)

  • The test isn’t that some hits come up on Google that mention the IEC prefixes; of course they are going to be there. The test is what the “majority of reliable RSs” do. This advise is repeated right in MOSNUM’s guidance on expressing quantities of bytes and bits, where it states that we may use such terminology when the article is on a topic where the majority of cited sources use the IEC prefixes. Clearly this is not the case; not by a long shot because the imbalance is wildly disproportionate against the IEC prefixes. So to avoid running afoul with WP:WEIGHT, we simply follow mainstream practices. The use of the IEC prefixes should be most prevalent amongst manufacturers and retailers of RAM, yet "1GiB" RAM produces 128,000 hits and "1GB" RAM produces 14.3 million hits, including all manufacturers and distributors of RAM. If we didn’t have WP:WEIGHT as a bedrock principle, our Apollo 11 article would talk about how “Most experts believe Apollo 11 was when men walked on the Moon” in order to allow for the conspiracy theorists, which is something like 20% of Americans, which is far in excess of the 0.9% of Google hits in IEC-prefix example. Greg L (talk) 02:48, 12 April 2011 (UTC)
  • We're not talking about RAM, we're talking about hard drives, and specifically we're talking about a narrow topic within the subject of hard drives. What do the majority of RS's about that narrow topic say? Have you checked? Comparing NIST-recognized units to moon landing conspiracies isn't persuasive. There's no facts in dispute here; it's just a question of the clearest way to write the exposition. GiB isn't fringe. It may be minority but in that case it should still be represented per NPOV (i.e. with some citations like the ones I gave above). Fwiw, the article's description of why hard drive capacity uses decimal figures is itself uncited and looks more OR-ish than the gibibyte thing, from what I can tell, and a lot of the "binary prefix timeline" looks OR-ish too. I know there are a bunch of computer books at my local library with names like "hard drive bible", so next time I'm there, I'll try to remember to look at a few of them about this. 75.57.242.120 (talk) 03:25, 12 April 2011 (UTC)
  • I'm beginning to think VMS's approach of reporting everything in "blocks" was the right idea all along. If you're going to use a different unit, make it an obviously different one. Jeh (talk) 04:01, 12 April 2011 (UTC)
  • Agree w/Becksguy. Hi. I've been chatting with Raptor on his tp about other issues, and his comments directed my attention here. In the discussion there, admittedly, we found ourselves seeing some things differently, stemming from a completely different substantive--but similar behavioral--4-months-old issue at the TSA article. In any event, as to the issue at hand, I think Becksguy makes good points, and his suggestion seems to best reflect consensus, to whose numbers I add my own.--Epeefleche (talk) 03:57, 12 April 2011 (UTC)
I'm fine with turning the table into a paragraph, but using long and obtuse language like "in the binary sense" to describe GiB without actually using the word GiB is just censorship.--RaptorHunter (talk) 04:09, 12 April 2011 (UTC)
It isn't censorship, calling it censorship is pejorative. It is good editing and following the policies WP:NPOV WP:UNDUE.Glider87 (talk) 04:35, 12 April 2011 (UTC)
If there's any article that deserves a line or two to talk about binary prefixes, it's this one. If you think binary prefixes aren't notable, then take the article to AFD. In the meantime, the hard drive article will continue to mention them.--RaptorHunter (talk) 05:13, 12 April 2011 (UTC)
You have not given any good reason to include IEC prefixes. WP:ILIKEIT is not a good reason. The hard drive article will not continue to mention them because that would be against the RfC consensus.Glider87 (talk) 05:29, 12 April 2011 (UTC)
  • I guess I'll wait to see Becksguy's edit before weighing in on it. I do get the impression that Greg and Glider87 trying to suppress terminology that they don't like, even at the cost of good exposition. Greg, do you have a summary of RS you have examined on this topic? 75.57.242.120 (talk) 04:43, 12 April 2011 (UTC) Add: it looks like the "binary prefix" article is also loaded with OR, and that a lot of its hard drive info should be moved to this article (and similarly for some other sections re other articles). I also start to comprehend that this discussion is yet another tiny eruption in an enormous lame edit war spanning many articles and talkpages, so descending into it is probably counterproductive unless towards a pretty drastic outcome that's not worth pursuing in the scheme of things. 75.57.242.120 (talk) 04:58, 12 April 2011 (UTC)
  • I gave 4 citations showing use of the IEC prefixes in RS references about the units discrepancy. You actually haven't cited a single source of that type that doesn't mention the IEC units. So far, you're the one using ILIKEIT. 75.57.242.120 (talk) 05:36, 12 April 2011 (UTC)
  • Given that you and I know there are many thousands of references showing real world use of non-IEC prefixes then four citations is not enough to show notability or to counter WP:UNDUE. "Generally, the views of tiny minorities should not be included at all." To push for a minority system to be used is WP:ILIKEIT, so I am not using WP:ILIKEIT you are. The use of IEC prefixes is in the tiny minority so it should not be included on this page at all. To make the case for IEC prefixes to be used on this page you need to show that they are widely used by the majority of the computing industry for the conversion task used in this article. Glider87 (talk) 05:43, 12 April 2011 (UTC)
No I'm not aware of those thousands of references, I have asked you for them multiple times and you have not cited a single one, so your credibility is poor at this point. Yes I know there's thousands of spec sheets, drive packaging, manuals, etc. that use decimal units. I'm asking for sources of the type I cited: 1) printed book about hard drives or computer systems, not specific to some particular device family; 2) use (or lack of it) of IEC prefixes specifically in the topic under discussion: explaining the decimal-binary discrepancy for drive capacity. You're mistaken (per NPOV) that "majority" sourcing has to be shown for the prefixes to be in the article--significant minority is enough, and I'm claiming that (at minimum) is established (by the 4 books I listed). Can you look at 10 random books on the subject and see what they say? 75.57.242.120 (talk) 06:34, 12 April 2011 (UTC)
Question to "75.57.242.120". Are you aware that IEC prefixes use is in the tiny minority?Fnagaton 09:21, 12 April 2011 (UTC)
Looking for Google books involving "drive 1024 1000 specifically without memory" and a SI prefix then specifically with and without the IEC prefix. I filtered out results with "memory" to try to remove as many books talking about binary types of memory.
Search term Hits
terabyte without tebibyte 57
terabyte with tebibyte 0
gigabyte without gibibyte 625
gigabyte with gibibyte 5
Like I said IEC prefixes are not notable given the majority of references out there not using them. Glider87 (talk) 10:14, 12 April 2011 (UTC)
Google is not an appropriate tools to measure usage in reliable sources. Upgrading and Repairing PCs (the best-selling book on computer hardware [5]) discusses the IEC binary prefixes in relation to storage capacity. Do you know of any recent computer hardware books which do not? —Ruud 10:47, 12 April 2011 (UTC)
This is not Google it is Google books. Different search. Anyway the links from 75.57.242.120 that were claimed to be "notable" were from Google books. The links I posted contain many more recent computer books that do not. Glider87 (talk) 10:51, 12 April 2011 (UTC)
For example from the link I posted you have Beginning Ubuntu Linux, Fifth Edition published 2010 and even though it is a book about Ubuntu it doesn't seem to mention IEC prefixes at all. Glider87 (talk) 11:12, 12 April 2011 (UTC)
Seagate don't use them either. If the manufacturer does not then why should we?Glider87 (talk) 11:21, 12 April 2011 (UTC)
Also Ruud why are you not questioning the Google books links from 75.57.242.120? Those three links (a fourth is not even a book by the way) are all several years older than the more recent books I produced. To try to question my Google books references and not question the links provided by 75.57.242.120 does not look neutral. Glider87 (talk) 11:46, 12 April 2011 (UTC)
Here's a Seagate product manual from March 2011[6] that uses IEC prefixes to describe a drive's transfer speed (though not capacity) on p. 18 of the pdf. I don't see anything in "Beginning Ubuntu Linux" about the prefix discrepancy--can you give a page number? If it doesn't discuss the prefix discrepancy it's not an on-subject source. 69.111.194.167 (talk) 08:08, 13 April 2011 (UTC) (new address) Added More generally, re "if the manufacturer doesn't use them, why shouldn't we?", the manufacturer (in some viewpoints) has a self-serving goal of wanting to show bigger numbers for marketing purposes. So to find the NPOV we ideally want RS's that are independent of the manufacturers. 69.111.194.167 (talk) 08:16, 13 April 2011 (UTC)

Replace table with this paragraph:

A one terabyte (1 TB) disk drive would be expected to hold at least 1000 GB with a GB defined to equal 1,000,000,000 bytes; and indeed most 1 TB hard drives will contain slightly more than this number. However some operating system utilities would report this as around 931 GB or 953,674 MB with a GB defined to equal 1024^3 or 1,073,741,824 bytes. To alleviate confusion, a new standard has been put forward by the IEC that would refer unambiguously to powers of 1024 with special binary prefixes. In this system 1024^3 or 1,073,741,824 bytes equals 1 GiB and 1000^3 or 1,000,000,000 bytes equals 1GB. However, as of 2011 adoption has been slow and usage has been limited in the marketplace and in the press.

This way the article doen't use the term GiB, instead the article describes the term GiB. This should keep the WP:MOSNUM people happy.--RaptorHunter (talk) 05:51, 12 April 2011 (UTC)

Not OK. It still mentions 1 GiB which ignores the RfC consensus and violates WP:UNDUE and is also not what this article is meant to discuss. Glider87 (talk) 06:00, 12 April 2011 (UTC)
This version would be OK...
A one terabyte (1 TB) disk drive would be expected to hold at least 1000 GB with a GB defined to equal 1,000,000,000 bytes; and indeed most 1 TB hard drives will contain slightly more than this number. However some operating system utilities would report this as around 931 GB or 953,674 MB with a GB defined to equal 1024^3 or 1,073,741,824 bytes. To alleviate confusion, a new standard has been put forward by the IEC that would refer unambiguously to powers of 1024 with special binary prefixes. However, as of 2011 adoption has been slow and usage has been limited in the marketplace and in the press.
This is alse more concise while linking to the article that describes it in more detail. Glider87 (talk) 06:00, 12 April 2011 (UTC)
It is ridiculous to say that one mention of binary prefixes in the ENTIRE hard drive article is UNDUE WEIGHT. It is extremely common to have a paragraph in an article with a quick description and a link to a more complete article. — Preceding unsigned comment added by RaptorHunter (talk • contribs) 06:06, 12 April 2011 (UTC)
WP:UNDUE applies because this article is about hard drives and not about IEC prefixes. To want to use IEC prefixes on this page is not neutral. Glider87 (talk) 06:13, 12 April 2011 (UTC)
Actually I prefer this version, even shorter and more concise:
A one terabyte (1 TB) disk drive would be expected to hold at least 1000 GB with a GB defined to equal 1,000,000,000 bytes; and indeed most 1 TB hard drives will contain slightly more than this number. However most operating systems would report this as around 931 GB due to differences with binary prefix use.
Glider87 (talk) 06:38, 12 April 2011 (UTC)
Again, this only states there is a difference, it does not explain why there is a difference or how one could compute the difference. —Ruud 09:48, 12 April 2011 (UTC)
Since that is explained in another article there is little point explaining it here in this article.Glider87 (talk) 11:16, 12 April 2011 (UTC)

The paragraph is on the right track and I don't think it has an undue weight problem towards too much IEC. I'd wikilink GiB and gloss it as "gibibyte". I think given the amount of user frustration that has arisen around the issue, it's ok to give it more space than Glider87's version. 75.57.242.120 (talk) 06:41, 12 April 2011 (UTC)

If we try to include IEC units in the summary paragraph, then we will still have content that is contentious, just a shorter version of it. The whole point of the suggestion for a summary section is to avoid the contentious content. One editor suggested that it's "sub-optimal" and I can't disagree. But it's better than the disruption we been having, and if it doesn't work, toss it out and try something else. We should not mention IEC or it's units at all, since all that is in the Binary prefix article. The shear amount of disruption over this issue totally overwhelms any possible benefit in using IEC units in this article, since we can describe the capacity issue without using them. I think it speaks volumes that two established editors came in supporting the inclusion of IEC units, but have switched. I also think it's amazing that we are actually collaborating here. Lets please keep it up. This isn't about us as editors, it's about serving our readers. — Becksguy (talk) 11:01, 12 April 2011 (UTC)

"I think it speaks volumes that two established editors came in supporting the inclusion of IEC units, but have switched" I agree, it does indeed speak volumes. It is good to see. Glider87 (talk) 11:33, 12 April 2011 (UTC)

Thank you. User:Jeh was the first editor to rethink his position. See here on his talk page. — Becksguy (talk) 11:56, 12 April 2011 (UTC)

Proposed new section and table

PC hard disk drive capacity (in GB) over time. The vertical axis is logarithmic, so the fit line corresponds to exponential growth.

Capacity measurements

Hard disk manufacturers quote disk capacity in multiples of SI-standard powers of 1000, where a terabyte is 1000 gigabytes and a gigabyte is 1000 megabytes.[1] Many operating systems and utilities – including Microsoft Windows, Apple's MacOS X prior to version 10.6, and Ubuntu – report hard disk capacity in powers of 1024, where a terabyte is 1024 gigabytes and a gigabyte is 1024 megabytes. Therefore available space appears less than advertised capacity. The relative difference between the two capacity measurements increases logarithmically with the capacity of the hard disk. The actual number for a formatted capacity will be somewhat smaller still, depending on the file system.

The discrepancy between the two methods of reporting sizes had financial consequences for at least two hard drive manufacturers when a class action suit argued the different methods effectively misled consumers (see Orin Safier v. Western Digital Corporation and Cho v. Seagate Technology (US) Holdings, Inc.).[2][3]

Starting in about 1998, a number of standards and trade organizations approved standards and recommendations for a new set of binary prefixes, proposed earlier by the International Electrotechnical Commission (IEC), that would refer unambiguously to powers of 1024. According to these, the SI prefixes would only be used in the decimal sense, even when referring to data storage capacities. As of 2011 they have seen little adoption in the computer industry. Microsoft Windows reports disk and file sizes using customary binary prefixes, Apple MacOS X switched to decimal SI prefixes in version 10.6. and Ubuntu uses the IEC binary prefixes.

Advertised, expected, reported and formatted capacities of a 100 MB, 100 GB and 100 TB hard disk drive.
Source Advertised capacity
by manufacturer 		Expected capacity
by consumers in class action suit 		Reported capacity Formatted capacity
Apple Mac OS X 10.6[4] Microsoft Windows Ubuntu[5] and Fedora[6] HFS+ NTFS ext3
Units SI prefixes Bytes Customary binary prefixes Bytes Difference SI prefixes Customary binary prefixes IEC binary prefixes SI prefixes
100 MB 100000000 100 MB 104857600 4.36% 100 MB 95.37 MB 95.37 MiB (todo)
100 GB 100000000000 100 GB 107374182400 6.87% 100 GB 93.13 GB
95367.40 MB 		93.13 GiB
100 TB 100000000000000 100 TB 109951162777600 9.05% 100 TB 90.95 TB
93132.30 GB
95367431.64 MB 		90.95 TiB

Comments

Not good. It includes IEC prefixes which do not have consensus. Not to mention it includes more confusing table that duplicates the work found in binary prefixes section "Deviation between powers of 1024 and powers of 1000". Basically just remove the whole section, add a small paragraph and link to binary prefixes is much better. Glider87 (talk) 10:55, 12 April 2011 (UTC)

Oppose – Why are we considering the Ubuntu market share? Is it 5 million units? (How about a reliable source.) English Wikipedia has hundreds of millions of readers. The tablet and smart phone platforms dwarf the desktop Linux market.
The market research group Gartner has just released this study.
"Media tablet sales are growing exponentially, according to Gartner’s figures, with approximately 70 million units to be sold in 2011, about 300% more than in 2010. By 2015, Gartner expects tablet sales to almost hit 300 million."
Article in Forbes. Gartner press release. -- SWTPC6800 (talk) 14:06, 12 April 2011 (UTC)

Look at the article Operating_systems: Examples of popular modern operating systems for personal computers are: Microsoft Windows, Mac OS X, GNU/Linux, and Unix. [4]
If we talk about ubuntu there, there's no reason not to talk about it here.--RaptorHunter (talk) 16:25, 12 April 2011 (UTC)
If you look at reference [4] you will see that Linux has a 0.98% share. The iPhone has a 1.02% share and the iPad has a 0.70% share. Smart phones and tablets are growing while desktop Linux is stagnant. (I use Red Hat Enterprise Linux at work.) -- SWTPC6800 (talk) 00:41, 13 April 2011 (UTC)

Looks Good The table reports 3 major operating systems and how they display hard drive space. Seems informative and unbiased without give undue weight to binary prefixes.--RaptorHunter (talk) 16:22, 12 April 2011 (UTC)

However, you should drop the outer two columns. They are empty.--RaptorHunter (talk) 16:29, 12 April 2011 (UTC)

Oppose — this is convoluted, unreadable, and seems to have been designed solely for the purpose of getting the rejected binary prefix garbage into the article. *** Crotalus *** 16:42, 12 April 2011 (UTC)

Oppose. IEC units have been introduced (against consensus). Progress was being made in rewriting the paragraphs, so using a table is an unwanted and backward step. The table is bizarre as it makes it look like the world has three major OSs: Apple, Windows, and Ubuntu. The lack of references for any of the data in the table is poor. So the Apple column only refers to 10.6, but the Windows column is universal? I was feeling a little down today, but this amateurish and transparent attempt to push a cause (using Ubuntu instead of Linux is priceless) has given me a laugh and perked me right back up again. Thanks.  GFHandel.   21:22, 12 April 2011 (UTC)

Oppose. Per wp:consensus. After all the above discussion, reflecting an opposite consensus, this seems to be a strikingly odd proposal.--Epeefleche (talk) 21:15, 12 April 2011 (UTC)

Support Needs some wordsmithing but basically the correct approach and entirely consistent with MOSSUM since the section is all about the various ways of reporting capacity, including but not limited to IEC Binary Prefixes. Tom94022 (talk) 22:37, 12 April 2011 (UTC)

Oppose Clearly a contrivance that shoehorns reality to the side too aggressively in order to “get those units in there somehow.” If one were to objectively approach that last one (Ubuntu), it would just say “Linux” and the units would read “varies”. I see our own Usage share of operating systems mentions Linux and it bifurcates that into “mainstream” and “Android”. This outside article, A Guide To Linux Flavours lists quite a number of different flavors, as follows: Redhat, Fedora, Mandrake, Suse (by Novel), Debian, Knoppix, Mepis,, Ubuntu (there it is), Slackware, Collage Linux, and (*sigh*) Gentoo. Given that all of Linux totals around 2%, slipping the ol’ Ubuntu®™© seems quite the contrivance. The sensible thing to do here is dump it because it comprises cherry-picked flavors of the highly fragmented Linux cloud. Everyone else’s computers are just reporting KB, MB, and GB. Even Apples iOS for iPads and iPhones has a greater share than all of Linux combined. So…

The addition of two flavors of Linux in this table, when Linux in total is around 2% of the OS market, is unwise, cherry-picked, and it’s obvious why.

It’s simple: the majority of the citations in the hard drive article don’t use the IEC prefixes. The vast majority of the hard drive manufacturers themselves don’t. To wedge these virtually unheard-of units into such an article by riding the coat tails of Ubuntu is transparent and not in keeping with the bedrock principal of “follow the majority of most-reliable modern RSs for the discipline.” (I revised this after Ruud’s comment, below) Greg L (talk) 00:21, 13 April 2011 (UTC)

I think this includes most recent (desktop/workstation) Unix variants. I don't have shell access to a SuSE or Solaris machine at the moment, so I can't verify this right now. —Ruud 00:12, 13 April 2011 (UTC)
Well, given the context of the table we should probably restrict ourselves to desktop operating systems (ignoring mobile and server platforms, the first generally not having a hard disk drive, the second certainly not weakening the argument for the inclusion of Linux). Of desktop Linux distribution Ubuntu and Fedora account, by a fair margin, for the most installations. I have not yet checked, but suspect the other distributions have by now also switched to IEC units. —Ruud 00:39, 13 April 2011 (UTC)

Oppose. Show me that the top ten linux distributions (Ubuntu, Fedora, openSUSE, Debian, Mandriva, Linux Mint, PCLinuxOS, Slackware, Gentoo, CentOS - see distrowatch.com) use IEC Binary Prefixes and I will support the usage in the table under "Linux." Otherwise, use "varies" as Greg L suggested. Guy Macon (talk) 00:19, 13 April 2011 (UTC)

Sounds reasonable. I'll give it a try sometime. —Ruud 00:23, 13 April 2011 (UTC)
Actually they all use binary prefixes. They all run the linux kernel and the linux kernel uses binary prefixes.--RaptorHunter (talk) 00:28, 13 April 2011 (UTC)
If you can prove that, RH, I might drop my objection. Linux is only 2%. But if there is one consistent unit of measure being displayed by all the GUIs these Linux flavors present, then I suppose there is nothing wrong with adding Linux to the table… it’s out there. But your assertion needs to be proven true, and we also need to ascertain for certain that the GUI isn’t doing its own thing. Fortunately, I have a good friend who is an atomic-level programmer and owns a triple-boot Mac (Mac, XP, and Linux). I’m not sure what flavor. One call–that’s all… Greg L (talk) 00:34, 13 April 2011 (UTC)
Proof: Binary_prefixes#Operating_Systems--RaptorHunter (talk) 01:43, 13 April 2011 (UTC)
P.S. No answer. I left a message on his answering machine. But this Web site: Red Hat Enterprise Linux AS 2.1: The Official Red Hat Enterprise Linux AS Installation Guide doesn’t bode well for any argument that the various flavors of Linux consistently uses any particular unit of measure for hard drive capacity. Megabytes (MB). It might be sloppiness. Probably not. Greg L (talk) 00:48, 13 April 2011 (UTC)
The Linux kernel doesn't use any prefixes at all. Perhaps you are confusing the Linux kernel with a Linux Shell or a Linux Window Manager? Guy Macon (talk) 01:44, 13 April 2011 (UTC)
My Ubuntu and Suse installations do not display IEC prefixes. Fnagaton 00:49, 13 April 2011 (UTC)
Ubuntu only started this with the 10.10 release. However, if you run gparted in any release you will see it.--RaptorHunter (talk) 01:41, 13 April 2011 (UTC)
This chart is so dead… Greg L (talk) 00:57, 13 April 2011 (UTC)
I'm fairly sure they currently do. The long term plan of Ubuntu is to move to decimal SI prefixes for reporting all file and disk sizes and use IEC prefixes only for reporting memory sizes. This makes the Linux column in the table I proposed misleading. —Ruud 01:09, 13 April 2011 (UTC)
It’s not the Superman’s cape one clutches at 13,000 feet as the basis for trying to slip an “Oh… didn’t-cha know??”-use of the IEC prefixes into an article about hard drives. Indeed. If we gotta include Linux, just put “Varies all over hell & creation and isn’t stable either.” Greg L (talk) 01:16, 13 April 2011 (UTC)
https://wiki.ubuntu.com/UnitsPolicy does not appear to say what you claim it says. Guy Macon (talk) 01:45, 13 April 2011 (UTC)
Which is basically what I just said. —Ruud 02:03, 13 April 2011 (UTC)

Oppose IEC units have been proposed against consensus. The next proposal should not include IEC prefixes. Fnagaton 00:49, 13 April 2011 (UTC)

The RFC above shows a consensus that they are acceptable if the article is describing them, not using them.--RaptorHunter (talk) 01:50, 13 April 2011 (UTC)
The RfC doesn't say that and it certainly doesn't open the way for proposals containing IEC prefixes to be used in this article.Fnagaton 02:04, 13 April 2011 (UTC)
President Obama said the American people owe a debt of gratitude to me for contributions to the fuel cell industry. (God that was fun.) Greg L (talk) 02:22, 13 April 2011 (UTC)

Here’s what my programmer friend says about Ubuntu (or Linux for that matter) using the IEC prefixes. He is sooo into the depths of this, it will be hard to explain. First, I explained that I was working a debate on Wikipedia over the units of measure used in the computer industry. He said “Oh, you mean that ‘mebi-whatever’ crap?” Yes, I responded. I told him how this article was an old holdout that slipped beneath the radar years ago when the issue was settled on MOSNUM and how the principals had run off to work on articles about the IEC prefixes, such as Timeline of binary prefixes. He responded “Where they can write how ‘This is what it was’.”

Unfortunately, he doesn’t have any of his Linux machines at home this evening. His triple-boot Mac uses Ubuntu and a computing brick he is programming to use as an industrial control device uses Debian. First observation from him: Linux traces its roots to well before the IEC prefixes were even proposed so—at least for the most part—the kernel-level stuff uses the conventional prefixes. Also, Linux in all its carnations runs on a huge variety of hard drive file systems (HFS, ext2, etc.). He has been partially sensitized to the mebibyte (*sigh*… the built-in spell checker in Safari flagged one of those prefixes with a dashed-red underline) because he uses Wikipedia and used to encounter the Porky Pig stutters. And he knows me and knows about the MOSNUM jihads and suicide bombers and all. So he is sensitized to the issue and notices the IEC prefixes when he encounters them in Linux—which isn’t often at all. To his recollection, in virtually call cases, he sees the standard decimal prefixes being used for reporting free memory as well as storage capacity. He added that it depends upon what tool one uses to work on a hard drive. For instance, if one uses a command-line partition editor, it reports using “KB” and “MB”, etc. But he does recall seeing an instance of the IEC prefixes being used and it was a GUI-based partition editor.

When I told him that some sort of Ubuntu place was saying that official policy was to use the IEC prefixes in the future, he asked who (thinking it some discussion group). So I came here to this page to look up Ubuntu.com’s Units Policy and told him who it was. He said “Oh. The guy who started Ubuntu is that South African internet mogul who is so rich, he bought a ride on Soyuz to the Space Station. Yeah, he’s out to change the world.”

But, in a nutshell, it is soooo very misleading to say that “Ubuntu” uses the IEC prefixes. That’s like saying “People in Mississippi use bad grammar.” It’s probably worse than that; for the majority of user-interface things to which users of current installations of Ubuntu are being exposed, they aren’t seeing the IEC prefixes.

It’s time to slam-dunk this one. They IEC prefixes have absolutely no excuse for being in this article. This “Ubuntu” thing was apparently all predicated on a Web site showing where the South African guy wants Ubuntu to go in the future; that’s all. Greg L (talk) 03:17, 13 April 2011 (UTC)

Everytime you partition a hard drive in ubuntu you see this:

— Preceding unsigned comment added by RaptorHunter (talk • contribs) 03:26, 13 April 2011 (UTC)

  • No, all you’ve shown is a particular GUI-based partition editor on Ubuntu that does that. That’s what you see when you partition your Ubuntu hard drive with that tool. That is the exemption to the rule I wrote of above. If you read what I wrote above about my programmer friend, you’d know that. I suspect you did read what I wrote above and exploited that and then came here with this misinformation. That is so low. If you wanted to include the IEC prefixes, you’d have to write that “Some flavors of Linux, specifically Ubuntu, have some GUI-based partitioning tools (v.s. command-line ones) that expose the computer user to RaptorHunter’s IEC prefixes so he thinks there is some sort of basis upon which to make an outrageously exaggerated and lame excuse to include them in Wikipedia’s table on the Hard disk drive article.” Upon that you’re trying to make a case of adding the IEC prefixes to this table?? Give it up; it’s dreadful seeing you clutch at straws like this. Greg L (talk) 03:38, 13 April 2011 (UTC)
You seem to have endless energy to type out these replies.--RaptorHunter (talk) 04:05, 13 April 2011 (UTC)
Facts are facts. Try it some time. Like how you have 46 posts on this page to my 43. And that excludes any of yours here when you were too sloppy to actually sign in and posted as an I.P. Ample electronic white space is provided below this line for you to make it #47. Happy editing. Greg L (talk) 04:13, 13 April 2011 (UTC)
  1. 1 I've never edited this talk page as an IP address. I don't have to "bother to sign in". The computer stays signed in all the time. I've seen your accusations of sockpuppetry before. This is one of the most mild ones yet, but even when you have outright declared me to be some nemesis of yours from years past your accusations don't go anywhere. User:Gwen_Gale shot you down. WP:ANI seems to be ignoring you. (I have the sneaking suspicion it's because no one actually wants to read the pages and pages you type in these type of debates) Maybe you should try a tl;dr version next time you call me a sock on ANI.
tl;dr Go ahead and submit my name to WP:SPI. Lump me in with whatever editors you want: thunderbird, 75.57.242.120 and anyone on this page that voted "compliant" Should be fun.--RaptorHunter (talk) 04:42, 13 April 2011 (UTC)
This would be the Gwen Gale who wrote [7] "there is still not yet a hint there would be any consensus for IEC prefixes in the MoS, given their ongoing lack of use in the wide world". The last editor who kept on ignoring the consensus against IEC prefixes and who kept on pushing so hard for them to be used was warned to stop by Gwen Cale. It is interesting you mention thunderbird as having voted "compliant" when thunderbird has not voted. Glider87 (talk) 04:52, 13 April 2011 (UTC)
You misunderstand me. I mention thunderbird because greg keeps accusing me of being thunderbird. Sorry to disturb whatever crazy theory you were passively aggressively suggesting.--RaptorHunter (talk) 04:56, 13 April 2011 (UTC)
It is interesting because one of the thunderbird accounts was warned by Gwen Gale for ignoring the consensus against IEC prefixes. I note you didn't comment on the "there is still not yet a hint " quote so I take that to mean you now accept there is consensus against using IEC prefixes.Glider87 (talk) 05:05, 13 April 2011 (UTC)
lol. That's right. If I don't dispute what you say then I must now completely agree with it. :) Are you just trolling? C+ for effort but D- for execution. --RaptorHunter (talk) 05:10, 13 April 2011 (UTC)
If you cannot reply with a reasonable counter argument then for the purposes of determining WP:CONSENSUS you are not disputing what I've posted. Glider87 (talk) 06:56, 13 April 2011 (UTC)

Proposed new section and table (2)

PC hard disk drive capacity (in GB) over time. The vertical axis is logarithmic, so the fit line corresponds to exponential growth.

Capacity measurements

Hard disk manufacturers quote disk capacity in multiples of 1000, where a terabyte is 1000 gigabytes and a gigabyte is 1000 megabytes.[7] Many operating systems and utilities – including Microsoft Windows and Apple's MacOS X prior to version 10.6 – report hard disk capacity in multiples of 1024, where a terabyte is 1024 gigabytes and a gigabyte is 1024 megabytes. Therefore available space appears less than advertised capacity. The relative difference between the two capacity measurements increases logarithmically with the capacity of the hard disk. The actual number for a formatted capacity will be somewhat smaller still, depending on the file system.

The discrepancy between the two methods of reporting sizes had financial consequences for at least two hard drive manufacturers when a class action suit argued the different methods effectively misled consumers (see Orin Safier v. Western Digital Corporation and Cho v. Seagate Technology (US) Holdings, Inc.).[8][9]

Starting in about 1998, a number of standards and trade organizations approved standards and recommendations for a new set of prefixes for binary multiples, proposed earlier by the International Electrotechnical Commission (IEC), that would refer unambiguously to multiples of 1024. According to these, the SI prefixes would only be used for decimal multiples, even when referring to data storage capacities. As of 2011 they have seen little adoption in the computer industry. Microsoft Windows continues to reports disk and file sizes using SI prefixes for binary multiples, while Apple's MacOS X switched to decimal multiples in version 10.6,[10] with Ubuntu planning to do so as well.[11]

Advertised, expected, reported and formatted capacities of a 100 MB, 100 GB and 100 TB hard disk drive.
Source Advertised capacity
by manufacturer 		Expected capacity
by consumers in class action suit 		Reported capacity Formatted capacity
Windows Mac OS X 10.6 NTFS HFS+ ext3
Units Decimal multiples Bytes Binary multiples Bytes Difference Binary multiples Decimal multiples Decimal multiples
100 MB 100,000,000 100 MB 104,857,600 4.36% 95.37 MB 100 MB (todo)
100 GB 100,000,000,000 100 GB 107,374,182,400 6.87% 93.13 GB
95,367.40 MB 		100 GB
100 TB 100,000,000,000,000 100 TB 109,951,162,777,600 9.05% 90.95 TB
93,132.30 GB
9,536,7431.64 MB 		100 TB

Comments

If there has to be a table then this version is OK. As per Jeh I would prefer a version without a table. But this version is better than the version in the article. Fnagaton 02:15, 13 April 2011 (UTC)

I see you left in the ext3 filesystem, but removed linux. More censorship from the WP:IDONTLIKE IEC crowd.--RaptorHunter (talk) 02:36, 13 April 2011 (UTC)
I proposed both this version and the one above. While GParted and the Gnome System Monitor use the IEC prefixes, Nautilus (the primary user interface to the filesystem) and the Disk Utility do not: they use decimal SI prefixes. The Ubuntu Unit Policy actually states the latter is preferred/correct. As the current usage is inconsistent I did not think it would be wise to include a column at this time. —Ruud 03:57, 13 April 2011 (UTC)
  • Meh. Whatever. There are no IEC prefixes, which addresses the issue at hand. The rest if window dressing. I changed the delimiting to the conventional format used on Wikipedia. Yes, thinspaces are *allowed* on Wikipedia but they are most suited for “scientific articles”, which was a concession for those who love thinspaces and want to use them on every article that seems “sciencey” to Sarah Palin. For this article, we follow the majority of the RSs for guidance. Even the space-cadets at Ubuntu (Units Policy) use the conventional way of delimiting numbers (commas). Yes, it could be argued that hard drives are designed by scientists, but this isn’t an article directed to readers of science journals; it’s directed to a general-interest readership. Greg L (talk) 02:51, 13 April 2011 (UTC)
I can't decide if your trying to insult ubuntu users or not.--RaptorHunter (talk) 03:21, 13 April 2011 (UTC)
In the case of my programmer friend, who has designed industrial-control equipment his entire life, he uses Ubuntu at home and Debian for process control because Linux isn’t a colossal hunk of hammered dog crap from Microsoft. And he’s built industrial-control technology upon licensed Microshaft software only to have Microshaft pull the rug out from under it all after a few years because it was just another half-baked wet dream of Bill Gates (to mix metaphors) that failed to crush the grassroots competition that actually understood process control. Never again. He uses derivatives of Linux and uses open-source software for everything else, including the database libraries when you want your machine to keep track of how many widgets were made that week. Then you don’t attach your machine control to the Internet. If anything works in this world, thank an engineer. Greg L (talk) 03:29, 13 April 2011 (UTC)
Greg tries to insult anyone, if you hadn't noticed that yet. That also explains his non-empty block log. This is quite disruptive during what is supposed to be a calm and rational debate and will be the reason he is going to be more trouble if this dispute ever reaches a more formal stage. For now you're best of ignoring this behaviour as much as you can. —Ruud 03:57, 13 April 2011 (UTC)
Ah! but since Greg likes to type so much, you can cherry pick his quotes to make yourself look good. For instance, quoting Greg: "RaptorHunter is rather fearless" [8] and "ride in atop a tall steed of Truth, Justice, and the Wikipedia Way™©®—its nostrils flaring in the morning mist." [9] — Preceding unsigned comment added by RaptorHunter (talk • contribs) 04:49, 13 April 2011 (UTC)

To all editors on this page: see my 03:17, 13 April 2011 (UTC), above. In a nutshell, it is not a factual statement to say “Ubuntu uses the IEC prefixes.” Only some tools do.

When you consider that the kernel underling Linux, which is at the bottom of Ubuntu, was written when everyone used only the decimal prefixes (before the IEC proposal), and when you consider that all the flavors of Linux combined enjoy ~2% of the O.S. share, which is less representation than iOS, and when you consider that there are dozens of flavors of Linux out there, only one of which is Ubuntu, and when you consider that command-line partitioning tools show the decimal prefixes for storage capacity and only some GUI-tools on Ubuntu (notably or a particular disk partitioning tool) use the IEC prefixes, the use of the IEC prefixes to describe the storage space available on hard drives is such a small subset of the universe of hard drive capacity-reporting, the chart here in this article would have to be three feet wide denoting practices down to the hundredth of a percentile point to capture nuances this fine. That? To get IEC prefixes into the chart?

It is only because of galactic tendentiousness that this issue has persisted for so long. It’s time to stop horsing around; every guideline and bedrock principal of Wikipedia as well as half a dozen principles from Technical Writing 101 have to be flouted to pull off this feat of magic. The IEC prefixes have no business in this article in any shape or form. Greg L (talk) 04:12, 13 April 2011 (UTC)

Well I guess now that Greg and Glider are in agreement, the debate must be over. Nevermind the countless numbers of reasonable editors that don't have the energy for these debates that never seem to go anywhere, but would say that the prefixes are just fine.--RaptorHunter (talk) 05:14, 13 April 2011 (UTC)
I left a blank open below so Greg can bang out 2 pages of text no one will actually read:
  • Well, OK… I’ll bite. I just love that log/log graph of hard disk drive capacity over time. I can see right where the 20 MB “Hyperdrive” came out, just as the “Fat Mac” was being introduced by Apple (circa 1985). I couldn’t afford all that stuff and was computing full-ghetto back then. That was a long time ago, darn near the beginning of that graph. Who here on this page goes back further than that and recalls coveting a hard drive that had less capacity than the Hyperdrive?

    Oh, uhmm… since the last count, when the post-count on this page was 47-43 / advantage RaptorHunter, the score is now 50-44 with RaptorHunter pulling away! Greg L (talk) 05:51, 13 April 2011 (UTC)

Re: "Who here on this page goes back further than that", My experience as an engineer goes back to the Perkin Elmer/Wangco plug comparable clone of the IBM Model 3340 30MB/30MB Winchester drive - around 1975. (And Fnagaton is entirely correct when he states that any attempt to force IEC prefixes to be used on this page is against the WP:MOSNUM consensus, is in violation of WP:NPOV policy and can be seen as disruptive editing behaviour.) Guy Macon (talk) 06:41, 13 April 2011 (UTC)
  • Comment The table can be made narrower by writing 1011 instead of 100,000,000,000 and so forth. I still favor IEC prefixes based on the sources I looked at the other night. I'd be interested in seeing some examples of sources on the same topic that don't use them (google hit counts based on pages about different topics are irrelevant). The term "SI prefix" should IMO also not be used without citation to an RS saying it's an SI prefix. Prefixes like K for kilo have been around since long before the Système international was created in 1960, and describing informal uses (like on HD packaging) as SI units without explicit sourcing seems to me like original synthesis. For example, WDC's advanced format technology information sheet[10] describes the capacity of an advanced format sector as 4K, meaning 4096 bytes, not 4000. In other words they use prefixes like "K" both ways, as most of us knew all along. Similarly, the "20MB Hyperdrive" Greg mentions appears to be an approx 21.6 million byte drive based on some googling. 69.111.194.167 (talk) 08:25, 13 April 2011 (UTC)
  • Referring to "SI units" as original synthesis: In a way it is, since SI does not define any units for quantities of information. However SI does state that the prefixes defined in SI can be used with non-SI units for such cases. Hence "SI prefixes." Now HD makers are pretty damn universal in stating "1 GB = one billion bytes," or perhaps "1,000,000,000 bytes" - it's not just on their packaging. If referring to that usage as an "SI prefix" is "original synthesis" then so is just about everything on WP that isn't a direct quotation from source. Jeh (talk) 16:07, 13 April 2011 (UTC)
  • I changed “decimal prefixes” and “binary prefixes” in the above table to “decimal math” and “binary math” since the distinction isn’t the prefixes—which stay the same for 99.9 percent of users—but is a distinction of the math underlying the meaning of the conventional prefixes. Greg L (talk) 20:55, 13 April 2011 (UTC)

Conclusion

RaptorHunter's recent editing pattern has been nothing but tendentious and bizarre. Normally a certain amount of eccentricity can be tolerated, but RaptorHunter is now taking up the valuable time of editors here who love to add content, and are therefore being distracted from doing that. It's time to bring this debate to a conclusion, and I've sought help to achieve that. BTW RaptorHunter, the vandalism you recently performed (twice) on Jimbo Wales' page was especially disturbing—as was your justification for doing it.

A table is a lazy way of explaining this issue, so let's continue with the good progress made with the descriptive paragraphs, and get this finalised.
 GFHandel.   08:27, 13 April 2011 (UTC)

Since RaptorHunter is obviously ignoring the very strong consensus on this talk page then I propose we move ahead with using the editprotected template or ask for the article to be unprotected so we can improve it. Fnagaton 12:45, 13 April 2011 (UTC)
What consensus? I do not see any. The requested comments above are evenly divided whether the table is contrary to the MOS guidelines. So it should be given the benefit of the doubt and not censored. −Woodstone (talk) 14:19, 13 April 2011 (UTC)
The consensus is obvious and against using IEC. The comments are not evenly divided. —Preceding unsigned comment added by 220.255.2.34 (talk) 15:41, 13 April 2011 (UTC)
The RFC had 6 violations and 9 Compliants. It seems that the consensus is that the table compliant with the manual of style.--RaptorHunter (talk) 17:51, 13 April 2011 (UTC)
I guess it has to be said yet again: Consensus is not established by counting the votes of whoever shows up. It's established by comparing the validity of arguments. There have been no valid arguments for "compliant". All of the arguments for "compliant" are either specious reasoning (e.g. "IEC prefixes are being used, therefore they are a topic of the section", or else contrary to previously established consensus in MOSNUM (e.g. "IEC prefixes will help explain the issue," when MOSNUM consensus is that they won't). Nor have there been any compelling counters posted to the various arguments for "not compliant." Jeh (talk) 18:17, 13 April 2011 (UTC)
Jeh is entirely correct. If we simply counted heads, all sorts of stupid things would occur on Wikipedia. There are many an occasion when editors essentially say I DON’T LIKE IT for slightly different reasons where none withstand logical scrutiny while those who better understand encyclopedic technical writing will speak with one voice that makes sense. It happens with regularity in a collaborative writing environment where *anyone* can edit. Greg L (talk) 21:11, 13 April 2011 (UTC)
I completely agree that consensus is not established by counting votes and it is established by comparing the validity of arguments. Consensus is not defined by specious reasoning or people turning up with WP:IJDLI. Woodstone you are a long term editor so you must know this is how consensus works by now? To continue to ignore the consensus against using IEC prefixes, as defined by the logical compelling counter arguments posted, is not helpful to making the article better. Woodstone can you provide any reasonable argument to counter the those compelling counter arguments?Glider87 (talk) 00:44, 14 April 2011 (UTC)
  • All arguments offered up by the proponents of the use of the IEC prefixes here to describe hard drive capacity haven’t withstood even a cursory inspection. Their use on the table in the Hard disk drive article persisted as long they did merely because of tendentiousness. Their use in that table in that manner was clearly intended to highlight and help promote wider adoption of the IEC prefixes in the computing world and the proponents own words here show that they have a great disdain for the current ambiguous nature of terms like “GB” and greatly believe in the virtue of the IEC prefixes. However, Wikipedia follows the way the real world works and is not to be used as a means of advocacy of all-things-cool.

    Wikipedia’s three-year-long attempt at using instances of “The Dell Inspiron 1501 originally came with 256 MiB or RAM” only confused readers and clearly accomplished nothing insofar as getting the world to follow our wise lead. That sort of terminology came about because a single editor—nearly overnight—managed to revise hundreds of articles and a small cabal lead by a rogue admin prevented all those edits from being reverted. At least two holdouts from the MOSNUM battle that ended that unwise practice settled on this article, which they have used as their final stand for two years. Greg L (talk) 17:03, 13 April 2011 (UTC)

You're WP:ANI was swiftly closed and archived as inappropriate. Maybe you should focus on the article instead.--RaptorHunter (talk) 17:46, 13 April 2011 (UTC)

  • My ANI?? WTF?!? Methinks the man is exceedingly confused this morning. Try Starbucks’ House Blend; works for me. Someone else started that ANI. I asked that it be swiftly closed because WQAs are better suited for your special brand of Wikipedia conduct. (*Mumbling under my breath*: As if facts matter at all here now…). I believe this is where you write By “you,” I meant GFHandel all along. Greg L (talk) 19:29, 13 April 2011 (UTC)
Greg_L take a deep breath and look at the indention. I was replying to GFHandel.--RaptorHunter (talk) 19:32, 13 April 2011 (UTC)

(*sound of deep breath)* Oh, well… Right under my post. How clear was that? GFHandel’s only post is some four inches up. BTW, I’m not responding to you, RaptorHunter; look at my indenting! ;-) Greg L (talk) 19:45, 13 April 2011 (UTC)

Well the last time I attempted to reply to a post by putting it directly under that post and indenting, GFHandel filed an incident report against me. [11] Maybe he should file another one...--RaptorHunter (talk) 19:51, 13 April 2011 (UTC)
Raptor, try responding like I just responded with this post: by addressing an individual with their name rather than the personal pronoun “you”; particularly when the individual to whom you are addressing is damn near off-screen above on a thread. That’s like walking into a room full of corporate attorneys and yelling “You! Blood-sucking parasite!”: they’re all going to simultaneously respond if you address them that way. Greg L (talk) 20:00, 13 April 2011 (UTC)
Hey you, Go ahead and file and incident report about my pronoun usage.--RaptorHunter (talk) 20:04, 13 April 2011 (UTC)
The ANI was closed because it was the wrong venue, however everyone there agreed that your editing is bizarre and disruptive. Do you really think that going around vandalising pages (that takes the valuable time of other editors to fix) is "fun"? Remind us all why any one would want to write "Jimbo never sleeps" as the caption of his picture on his own user page? You need to take a long look at how you work in a collaborative environment.
BTW, there is a clear consensus here for not using the IEC prefixes. The reasons were established on the MOSNUM page a while back, and the arguments given here reinforce those reasons. The only reason for using IEC prefixes to explain things on this page is: because they exist; however as pointed out (so many times) they only serve to introduce another confusing concept into the text—a confusion that doesn't need to exist for our average readership.
BTW II: you are still not indenting and responding in an accepted style on talk pages. I've been very polite and have tried to explain things to you, but to no avail. Do you not want to follow the conventions other editors master very quickly, or are you not capable of following them? There has been over a dozen examples lately where you have either not indented correctly, or where you have posted out of order. Would you like more assistance in learning the conventions that we all use—conventions that assist all editors to follow a debate?
 GFHandel.   20:10, 13 April 2011 (UTC)
Nice patronizing tone there. Also, I really wish you would quit wasting everyone's time with all of these incident reports. All you've managed to do so far is annoy the admins.--RaptorHunter (talk) 20:14, 13 April 2011 (UTC)
(Fixed further unconventional indenting.) So you see my offer to help you with your indenting problems as nothing more than "patronizing"? Is there any chance you would address the other issues I raised (such as why you consider adding "Jimbo never sleeps" to his page to be "fun")? You might also like to comment on the IEC issues raised.  GFHandel.   20:26, 13 April 2011 (UTC)

While Jeh, Greg L and others talk about "consensus" at MOSNUM it is only fair to point out that there are more editors in favor of use of IEC Binary Digits in this article than achieved the so-called consensus at MOSNUM and that the so-called consensus was achieved by a small group of editors in the face of an immediately prior consensus by a larger group of editors, 11 to 1 as I recall, that IEC Binary prefixes not be deprecated. I think we all agree that “The Dell Inspiron 1501 originally came with 256 MiB or RAM” was silly, even stupid, but there have been good arguments made by many editors here as to why they are appropriate in this article and so far I have seen nothing of substance to rebut the arguments, only the usual dismissive and insulting remarks by the IEC Binary Thought Police. Tom94022 (talk) 20:07, 13 April 2011 (UTC)

Oh dear! And we tried to be logical. So, the argument against their use that goes as follows:
When you consider that the kernel underling Linux, which is at the bottom of Ubuntu, was written when everyone used only the decimal prefixes (before the IEC proposal), and when you consider that all the flavors of Linux combined enjoy ~2% of the O.S. share, which is less representation than iOS, and when you consider that there are dozens of flavors of Linux out there, only one of which is Ubuntu, and when you consider that command-line partitioning tools show the decimal prefixes for storage capacity and only some GUI-tools on Ubuntu (notably or a particular disk partitioning tool) use the IEC prefixes, the use of the IEC prefixes to describe the storage space available on hard drives is such a small subset of the universe of hard drive capacity-reporting, the chart here in this article would have to be three feet wide denoting practices down to the hundredth of a percentile point to capture nuances this fine. So it amounts to grasping at diaphanous straws in a disingenuous attempt to justify using and helping to promote the IEC prefixes—in an article on hard drives of all things.
…you found that to be pretty silly. Right? And you must really object to User:Guy Macon’s views stated in his 06:41, 13 April post, above. Greg L (talk) 20:23, 13 April 2011 (UTC)
  • The above-proposed graph needs lots of work and there is no point trying to resolve atomic-level details to everyone’s satisfaction right here, right now. Those sort of details take time and discussion to improve things incrementally. The proper resolution is that the table in the article merely have the “Binary Prefixes” column deleted. There will be no mention of “binary prefixes” since well over 99% of users see no such thing.

    FYI, I changed “decimal prefixes” and “binary prefixes” in the above table to “decimal math” and “binary math” since the distinction isn’t the prefixes—which stay the same for 99.9 percent of users—but is a distinction of the math underlying the meaning of the conventional prefixes. Greg L (talk) 20:51, 13 April 2011 (UTC)

"Decimal math" and "binary math" sounds a bit silly and is somewhat of an original definition. According to the NITS the correct terminology would be "multiples" as in "prefixes for binary multiples". —Ruud 22:42, 13 April 2011 (UTC)
You missed my point: it’s not about the prefixes. Except for a subset of drive partitioning software utilities in a subset of Linux, the industry uses one set of prefixes two different ways. The table as now drafted shows the two meanings of “GB.” One could call it “Decimal multiples” and “Binary multiples” too. If you’re certain you’re technically correct, be my guest and revise the table; no objections here, Ruud. Greg L (talk) 23:32, 13 April 2011 (UTC)

Regarding the conclusion to this, here's a modification of a suggested paragraph:

A one terabyte hard disk drive stores 1,000 gigabytes, however operating systems report the size of such a hard disk drive to be smaller, e.g. 931.3 gigabytes. The discrepancy occurs because a gigabyte has a binary definition of 1,024 x 1,024 x 1,024 bytes (instead of the 1,000 x 1,000 x 1,000 decimal definition used by hard disk manufacturers). As the term "gigabyte" therefore has two practical uses, a system of binary prefixes has been proposed to account for the discrepancy, however the prefixes have not been adopted by the manufacturers of hard disk drives, nor are they used by the majority of operating system manufacturers.

 GFHandel.   21:31, 13 April 2011 (UTC)

I think this contains sloppily worded:
  • "... a gigabyte has a binary definition of ...", implies there is one definition of gigabyte while two problem is that there are two definitions in use.
  • "... a system of binary prefixes has been proposed to account for the discrepancy ..." The prefixes do not account for the discrepancy, the two definitions of gigabyte do, the new prefixes where intended to end the discrepancy by making it unambiguous to which of the two definitions is referred.
—Ruud 22:51, 13 April 2011 (UTC)
I don't understand your first point as my text explicitly states that there are two definitions of "gigabyte" in use. I've addressed your second point with this amendment:

A one terabyte hard disk drive stores 1,000 gigabytes, however operating systems report the size of such a hard disk drive to be smaller, e.g. 931.3 gigabytes. The discrepancy occurs because a gigabyte has a binary definition of 1,024 x 1,024 x 1,024 bytes (instead of the 1,000 x 1,000 x 1,000 decimal definition used by hard disk manufacturers). As the term "gigabyte" therefore has two practical uses, a system of binary prefixes was proposed to end the discrepancy, however the prefixes have not been adopted by the manufacturers of hard disk drives, nor are they used by the majority of operating system manufacturers.

 GFHandel.   23:05, 13 April 2011 (UTC)
  • That’s fine too. Mentioning them like that is perfectly fine. Fate and reality trampled the IEC prefixes like that Roman chariot driver in Ben Hur. People talk about how he wheezed for the last fifteen minutes of his pitiful life and move on. Enough wikidrama has transpired on this. The IEC prefixes have gone the way of microuno, which was also a keeno idea for addressing the ambiguity of terms like “ppm”. Didn’t catch on either. Greg L (talk) 23:15, 13 April 2011 (UTC)
Fine by me.Glider87 (talk) 00:46, 14 April 2011 (UTC)
I would like to see a Wikipedia article with a title something like List of proposed measurement units that never caught on Guy Macon (talk) 01:48, 14 April 2011 (UTC)

Momentary unlock or simply "unlock"

There seems to be a general consensus on what needs to be done. A brief mentioning of the binary prefixes (spelled out by name of the subject) and its failure to gain traction is fine. And further, there would be no use of the IEC prefixes themselves in a table (if a table is truly needed) or in text to describe the magnitude of memory, disk capacity, or file size. With that in mind, we have two options moving forward:

  1. Temporary unlock Agree on the details of verbiage here on this talk page and ask HJ Mitchell to allow a shepherding editor of that revised section to post it and then lock the thing down, or…
  2. Indefinite unlock Ask HJ Mitchell to indefinitely unlock the article. Then a team of editors acting in good faith can go rework the disputed section.


  • #2, Indefinite unlock - I move that we get back to normalcy ASAP. Good-faith editing in line with the general consensus is welcome. Those who might disrupt peaceable collaborative writing in a collegial manner would be shown no mercy and be given the ANI-equivalent of a Chinese execution with a bullet to the back of the head. Greg L (talk) 01:04, 14 April 2011 (UTC)
  • #2, Indefinite unlock - However I have a question. Will RaptorHunter, Tom and Woodstone respect the consensus and agree not to revert such changes in the article?Glider87 (talk) 01:17, 14 April 2011 (UTC)
Anyone may collaboratively join. Like I wrote, good-faith editing in line with the general consensus is welcome. Think of the task at hand like the local townsmen piling sandbags along the riverbank in the driving rain to keep Main St. from flooding: having the town grocer bring a pickup truck-full of sand is much appreciated. Seeing someone slashing sandbags with a knife when your back is turned is met with despair, to say the least. Greg L (talk) 02:11, 14 April 2011 (UTC)
  • Sorry to barge in here. There is no consensus to add the material in question; the onus is on those wishing to add it; pending such consensus to add it being demonstrated nobody should be adding it. Anyone disagree with that? If there is no disagreement the article c an probably be unprotected. --John (talk) 04:03, 14 April 2011 (UTC)
    What do you mean "no consensus"? There is consensus to not IEC prefixes because as Jeh says above "Consensus is not established by counting the votes of whoever shows up. It's established by comparing the validity of arguments.". The people who want to use them have not answered the compelling arguments that involve Wikipedia policy and guidelines. Glider87 (talk) 04:22, 14 April 2011 (UTC)

Unlocked

Reading over the discussion everyone seems reasonably okay with "Proposed new section and table (2)". I'm going to unlock the article now and see if this going to work. —Ruud 10:13, 14 April 2011 (UTC)

I'm now lost because I didn't think the consensus was for using a table at all. We were working on a paragraph under the Conclusion section above, so why the rush to update the article?  GFHandel.   11:09, 14 April 2011 (UTC)
The original dispute was about whether the table should contain any IEC prefixes (current table doesn't and everyone seems okay with that.) Hopefully any remaining tweaks to the wording of the section can be worked out without edit warring. —Ruud 11:36, 14 April 2011 (UTC)
  • Ruud’s effort is an excellent start and, if not ideal, is certainly in keeping with the general consensus on this talk page. We can now go back to the slow, evolutionary editing process—as is normally done on Wikipedia—as we refine how to describe the different ways the computer world uses terms like “GB” and “MB” to measure hard disk capacity, file sizes, and memory.

    The table is fairly large, but it appears to be an effective tool for the speed-readers amongst our readership.

    At least as I write this post, when the article was at this stage, it seems that the body text in the article and the way it discusses the IEC’s proposal (it’s *out there*) is an encyclopedic and appropriate treatment, notwithstanding the colossal extent to which the computing world has eschewed the IEC’s proposal.

    What has now been corrected as a result of the protracted discussions on this talk page is the IEC’s proposed units of measure for binary values and their symbols (the “IEC prefixes”) are no longer being used in body text or tables as a measure of hard drive capacity. Their previous use in this article placed undo emphasis on the units and implied ubiquity in the discipline wildly in excess of reality; very few of our readers have ever seen them and will ever see them in the real world. Greg L (talk) 14:19, 14 April 2011 (UTC)

There is no consensus and removal of the lock is improper. There is agreement amongst those advocating deprecation of IEC Binary Prefixes but a majoritiy of the editors so far in this discussion have favored using them. Please revert to the locked page prior to your unlocking Tom94022 (talk) 22:09, 14 April 2011 (UTC)

Agreed. What we seem to have here is a agreement among a very small group of editors dedicated to the elimination of the prefixes. The much broader RFC showed no consensus to remove the table. In fact it showed a strong majority who thought that it complied with wikipedia guidelines.--RaptorHunter (talk) 22:12, 14 April 2011 (UTC)
John tries to help RaptorHunter understand how consensus works but RaptorHunter just removes that offer of help. RaptorHunter and Tom94022 the fact is consensus is against use of the IEC prefixes because compelling arguments using Wikipedia policies were made and both of you did not tackle those arguments. If you both don't understand how consensus works you are in for untold misery until you do. 220.255.2.20 (talk) 23:14, 14 April 2011 (UTC)
Unlike the old section, the new version now explicitly discusses the IEC prefixes, and the new table, in my opinion, manages to convey the same amount, if not more, information than the old table without using the IEC prefixes. Could you given some concrete comments on how you think the current version of the section and table could be improved? (Note that in the RfC I found the old version to be "compliant", however I think the current version is even better.) —Ruud 23:37, 14 April 2011 (UTC)
I agree the new section is better than the old section in many ways; all u have to do is use the IEC Binary Prefixes to make it acceptable to me and a number other editors. Perhaps all that is necessary is a last column illustrating them. In the meantime u have a serious violation of Wiki policy by removing the lock in the absence of consensus. Please don't make me appeal to another administrator - do what is right and revert until there is consensus. Tom94022 (talk) 23:59, 14 April 2011 (UTC)
  • It’s an inevitability in RfCs when one side has ‘head count’ but is unable to mount a coherent and cohesive argument that there will be claims that there was no consensus. That was the same complaint the community saw after the MOSNUM guidelines covering this subject were adopted a couple of years ago. Yet there was indeed a consensus. Those here who are now dissatisfied with the outcome of this discussion and debate actually undermined their own objectives when a lead proponent of using the IEC prefixes directly contacted over 100 individual editors. Gaming the system may afford a short-term advantage but is seldom a winning ploy strategically. In this case, that conduct likely resulted in increased scrutiny on the quality and logic of the arguments and the consistency of the views amongst the editors who shared them.

    Notwithstanding the deeply held desires of a certain segment here, the simple reality is that a tiny, tiny fraction of 1% of computer users (only certain users using certain tools of a certain flavor of Linux) will ever see a binary prefix reported to them by their operating system. So to use them in a table in this article in the measure of binary capacity would violate WP:WEIGHT and WP:RS by placing undo emphasis on the units and would suggest ubiquity at odds with reality. Well beyond 99.5% of the OS industry does not use the IEC prefixes. All computer magazines directed to a general-interest readership do not use them. NO manufacturer of hard drives—what this article is all about—use the IEC prefixes on their packaging or promotional materials when communicating to their customer base. Even manufacturers of RAM (which isn’t even the subject of this article but would be the most logical industry to adopt the IEC prefixes) don’t use them when communicating to their general-interest customer base. Fate did not smile upon the IEC prefixes in the computing world.

    Note that WP:RS requires that our articles be balanced and based upon all majority and significant minority views (my emphasis). A small fraction of 1% does not so qualify. Moreover WP:MOSNUM, here states that the IEC prefixes could be used when the article is on a topic where the majority of cited sources use the IEC prefixes. Given that the vast majority of the RSs here in this article don’t use them, the arguments to use them utterly crumble because they have no logical foundation that is remotely compliant with the principles of Wikipedia and technical writing. That’s why there is indeed a consensus—to follow the advise of MOSNUM. Greg L (talk) 00:57, 15 April 2011 (UTC)

  • Oops. Sorry. That was an editing mistake. I accidentally inserted a sentence fragment into your post (I wondered where that went…). It was unintentional and likely aggravated by the fact that the wikicode for out two posts were back to back. I concede to sloppy editing; no doubt. Please see WP:AGF. Greg L (talk) 01:41, 15 April 2011 (UTC)
Tom94022 there are already two other admins looking at this article and they have built on the edits made by Ruud. I don't think it is a good idea to threaten an admin with baseless accusations of breaking Wiki rules. I don't think you would get very far given that there is no consensus for your point of view. I advise you read WP:CONSENSUS and understand why there is consensus here. I can help you with answering any questions you have and the policy. The policy also goes on to describe how someone trying to filibuster indefinitely, which can be said of your edits about IEC prefixes over the last three years, can be seen as tendentious editing. I think you would be much happier if you accepted the consensus exists, you don't have to agree with it, then just stop trying to beat the dead horse. Fnagaton 01:54, 15 April 2011 (UTC)
Fnag, "Consensus is a decision that takes account of all the legitimate concerns raised." I understand and accept your concerns but I do not accept your right to declare my concerns and the concerns of other editors illegitimate as you and the rest of the IEC Binary Prefix Thought Police do. In this article at last count 9 editors thought the use of IEC Binary Prefixes was compliant and 6 did not. That certainly does not indicate consensus on either side which is my point as to why unlocking was improper. "Editors who maintain a neutral, detached and civil attitude can usually reach consensus" - you exhibit none of these attributes so it is not surprising that we cannot reach consensus. Unfortunately it appears that Ruud moved from being neutral to becoming an advocate when he announced at 10:13 that "everyone seems reasonably okay" and at 10:14 unlocked the page. I hope he reverts because if he doesn't I intend to take this to the next level. Tom94022 (talk) 03:07, 15 April 2011 (UTC)
Regarding the nine editors you cite Jeh says "All of the arguments for "compliant" are either specious reasoning" and "or else contrary to previously established consensus in MOSNUM". Guy Macon says "Fnagaton is entirely correct when he states that any attempt to force IEC prefixes to be used on this page is against the WP:MOSNUM consensus, is in violation of WP:NPOV policy and can be seen as disruptive editing behaviour". As Jeh says there have been compelling arguments against using IEC prefixes, these arguments use Wikipedia policy. Your point of view is WP:IJDLI which is not a valid argument for consensus building. Your WP:IJDLI concerns are not legitimate because they are weak arguments. You need to explain how your point of is better than that consensus. In the whole if this talk page and other talk pages you have not done that. I think if you take this to the next level you have a good chance of being blocked for WP:TEND. Glider87 (talk) 03:35, 15 April 2011 (UTC)
  1. ^ i.e. see HGST, Samsung, Seagate, Toshiba and Western Digital websites
  2. ^ http://www.foxnews.com/story/0,2933,201269,00.html Western Digital Settles Hard-Drive Capacity Lawsuit, Associated Press June 28, 2006 retrieved 2010 Nov 25
  3. ^ Seagate lawsuit concludes, settlement announced, bit-tech.net
  4. ^ http://support.apple.com/kb/TS2419
  5. ^ https://wiki.ubuntu.com/UnitsPolicy
  6. ^ http://fedoraproject.org/wiki/Docs_Project_Style_Guide_-_Quick_Reference#Units_of_Measure
  7. ^ i.e. see HGST, Samsung, Seagate, Toshiba and Western Digital websites
  8. ^ http://www.foxnews.com/story/0,2933,201269,00.html Western Digital Settles Hard-Drive Capacity Lawsuit, Associated Press June 28, 2006 retrieved 2010 Nov 25
  9. ^ Seagate lawsuit concludes, settlement announced, bit-tech.net
  10. ^ http://support.apple.com/kb/TS2419
  11. ^ https://wiki.ubuntu.com/UnitsPolicy
Retrieved from "https://en.wikipedia.org/w/index.php?title=Talk:Hard_disk_drive/Archive_8&oldid=1253058622"