If this suite’s a success, why is it so buggy?

This is my response to If
this suite’s a success, why is it so buggy?
by Andrew Brown of
The Guardian.

Andrew Brown claims that OpenOffice “illustrates the limitations of
open source” and establishes that my aphorism “Many eyes make bugs shallow” is
false, but his reasoning is shaky.

Mr Brown appears to be arguing that because open-source development isn’t
perfect, it isn’t any better than closed source. But there is no
silver bullet for the problem of software complexity — all
programs, open or closed, will have bugs. The figures he is waving
around (6K bugs, 5K feature requests) are meaningless in isolation.

In fact, controlled comparisons between closed- and open-source
versions of functionally equivalent programs have been done. Barton
Miller’s well-known “Fuzz Papers” suggested that open source programs
to have a 39% edge in reliability over closed-source equivalents.

So where are the comparative statistics for the bug load of Microsoft
Office? Do we know that it has fewer than 11,000 bugs and feature
requests outstanding? If Mr. Brown don’t know that, or at least have
those figures for a closed-source program of comparable size to
OpenOffice, he has no basis for asserting that the open-source method
is failing.

His article does inadvertently illustrate an important point, however.
If you make legal paperwork a requirement before volunteers can
contribute to a project, very few will do so. If OpenOffice is
failing its promise, it’s not because “many eyeballs” doesn’t work —
it’s because bureaucratic obstacles are driving the eyeballs away.

54 thoughts on “If this suite’s a success, why is it so buggy?

  1. Anything “open source” from Sun Microsystems is just their idea of a marketing gimmick.

  2. Oh, sorry for making two posts, but I just read the article and noticed that Brown describes OOo as “one of the flagship projects of open source”. Riiiiight. Later he goes on to blithely contradict everything he has said above: “OpenOffice actually started as a commercial product – StarOffice [...] Almost all the work on it is now done by about 100 full-time Sun programmers. [...] As far as I know, in the five years it has been available as open source, not one contribution to the program has come from amateurs.” So what are you saying, that open source programmers don’t want to get involved with a bunch of stinky old spaghetti code developed by some obscure German company and bought by Sun “we failed at Unix” Microsystems? Crap, you’re right, open source is… no wait.

    The guy also makes numerous references to the fact that he thinks word processors are for writing books. I agree; I also think a food processor is great for making pizza.

  3. Perhaps it’s a sign that the Open Source community has a well-developed sense of the meaning of the phrase “Open Source”, but that meaning hasn’t filtered very well into the rest of society.

    Or at least, the segment that includes almost all technology-writers for Mass Media.

    Out of curiousity, I went to the article and noticed that the author has an “Author’s Blog” link at the bottom of the article. At that blog, he has a subject-listing including “software”.

    It’s not too interesting reading for this software geek. Somewhere along the way, he talks about the difference between coders and developers, and managed to call Linux (as directed by Torvalds) an “amorphous concept”.

    At which point I let out a hearty laugh and close the page, because he doesn’t know enough about open-source to know the difference between the LInux kernel (which is a specific project with limited focus), and various Linux-based distributions, running the gamut from LFS to Fedora. The kernel is Torvald’s domain, the rest he doesn’t pay much attention to.

  4. Well, we have always known that calling something open source (or even releasing something under an open source license) doesn’t do nearly as much good as having lots of programmers reading and fixing the source.

    Personally, I’ve been interested in the UNO architecture in the past (because I like C++), but I just can’t agree to the license provisions. That, and I don’t trust Sun to make good decisions when its bank account balance runs low. Sorry, Sun, I’d rather help AbiWord pass Open Office than dance with you.

  5. Besides the legal paperwork, OpenOffice is a repulsive, frightening project that I wouldn’t get into unless I absolutely had to. You can tell just by looking at the program’s interface that its code must be hidious. I’d rather spend my hacking efforts on something that at least looks fun, like GNOME, or even Mozilla.

    Also, the more unusual and elaborate the project’s code checkout and build system, the less likely I am to start hacking on it. It took me about five hours to even get GNOME CVS installed; by that time, the enthusiasm I had was spent.

    So there are many ways besides legal paperwork that you can scare away your help. Take note, ye project maintainers.

  6. The obvious reason why Brown does’nt quote bug statistics for MS-Office is: no one knows or is supposed to know how many bugs MS programs hide. In my experience, the differences between MS Office and OpenOffice (in terms of crappiness) are subtle to the point that they don’t matter.

    I think there is something deeply wrong with the whole concept of an `office suite’ that tries to be everything: a dumbed-down toy for casual users, an industry-standard power tool, a brainwashing machine penetrating styles of thinking and communication in the whole business world, a writing crutch for the clueless. You cannot fix this by changing the development model. Nor by reimplementing `the standard’ bug for bug. Time for fresh ideas.

  7. OpenOffice is obviously not a typical open source project. The codebase is absolutely colossal, and it was released under a free license only at a very late stage, with no developer community outside StarDivision/Sun (Netscape did the same with Navigator, and the developers deemed the codebase unsalvageable and rewrote it from scratch). Just compiling OO is a major effort, and until the recent reorganization of the source, was impracticable on a normal home PC. I was not aware of the legal paperwork required from contributors until now, but that certainly just adds to the already high barrier for potential contributors.

    That said, the project is in fact getting a lot of very high quality bug reports from people outside Sun (see their issuezilla), and companies like Red Hat, Novell and Canonical are contributing. I cannot imagine a way Sun could have got these companies to participace if it was not for the free license. OO has been translated to more languages than MS Office, and that should tell you something about the interest in it. I think the project is making progress in making the code more accessible and easier to build, and it is still early days of the 2.0 series. I haven’t found OO buggy, but it is slow in many respects. I’m quite sure it can and will be optimized. Just see how far KDE and GNOME have come in the performance department.

  8. The important factor with the “many eyeballs” model, is simply the amount of developer/code ratio. Blame scary complex code or bureaucratic obstacles, it’s not really important to this discussion, the core is that OpenOffice has a low developer/code ratio. And actually making it a rather good example of the eyeballs concept, since most other open source programs have both a higher developer/code ratio and higher reliability.

  9. One point that I realized after reading the referred paper is the fact that when a piece of code is so monolithic, as OpenOffice seems to be, it is more difficult to attract the casual programmer into looking at or playing with the source. If you have to spend 5 hours trying to find where a bug resides, they you are not going to be very likely to even have the desire of fixing it, no matter how bad the bug is. Small, modular seems to still be a possitive thing in the software world :)

  10. Byteshack:

    But… but… only monolithic applications are really professional!!!

    :)

  11. Gee, I actually like Open Office – I use it every day, hours per day, for documents, spreadsheets, and presentations. Sure, I encounter bugs, but not nearly as bad as those I encounter on those relatively rare occasions when I get stuck using Microsoft products. I find Microsoft Office to be cumbersome and Microsoft Project to be just shy of unusable.

  12. Jim, did I give you permission to talk!? Well, now that you’ve gone and run your damn fool mouth, I guess I might as well address it. Just like fighting women and children, arguing with idiots is a great way to boost my fragile ego.

    Your post is titled “ESR doesn’t understand Copyright law.” You then go on to not prove that at all. You make a case — which could potentially be compelling if you weren’t such an incontinent anal avenger — that requiring copyright assignment could be a Good Thing, contrasted against ESR’s opinion to the contrary (bureaucratic obstacles are driving the eyeballs away.).

    From this, I deduce that you both probably understand Copyright law fairly well. But whereas ESR is just an asshole, you’re a bitchy little cunt filled with sand that’s got nothing better to do than wax menopausal about completely asinine shit simply because… shit, WTH is your malfunction anyway?

    I’m betting you’re the guy behind this little gem also: (http://esr.1accesshost.com/)

    And lest you think I’m part of ESR’s fan club, let me list some of the ways in which I dislike him:

    1) Lose the mustache, or grow a mullet. Allah wills it.
    2) Paganism is queer, albeit somewhat less queer than the popular alternatives.
    3) .45 is for people who are too insecure about the size of their penis to shoot 9mm.
    4) Lisp may not be dead, but it’s a zombie, and deservedly so.
    5) Anyway who uses emacs with anything but a begrudging attitude needs a swift kick in the nuts.
    6) Your analysis of the weaknesses of OSS useability is spot now. Now go and fix the fetchmailconf gui for chrissakes. My little brother is DEAD because of that shit. I showed it too him, and his head just imploded. Stop the violence — gtk is your friend.
    7) Tae-kwon-do is what reject ballerinas pass off as a martial art.

    The list goes on, but in the interest of self preservation, I’m gonna cut it short. (One of the problems with ESR is that if you can see ESR, he can see you. And if you can’t see ESR, you might only be seconds away from death.) I’m also going to avoid writing a list about things I don’t like about you because attempting to push it through my pipe would be the end of teh intarweb as we know it. Plus, I don’t want CTS.

    I can point out the weaknesses in your fair and balanced argument, though. First, the projects you cite which require copyright aren’t exactly none for swiftness of development. (*cough* hurd *cough*) Second, I know of no rants accusing the FSF of being anti-property on ESR’s part, and being pro-property does not mean being pro-attaching-rivalrous-property-rights-to-every-goddamn-thing-in-site. Forth, if ESR is a poop-flinging monkey, then you are the guy getting shat upon in german scheisse films. (Actually, number four is probably just true in general.) Five, you are aware that ESR’s seminal (har har, but I digress) fetchmail project is released under the GPL? I mean, you knew that, right? ‘Cuz you’ve totally checked out his software and looked at the licensing he uses, ja?

    And here’s what makes it so fucking sweet: I’m member number 1301 of the FSF. I support software freedom (I just don’t support forcing people into it). I think this gives me the right to tell you to eat your own penis. And you know what? No mint jelly for you.

    Eric, sorry for this massive turd I just took on your comments section, but goddamn this dude is harshin’ my mellow.

  13. I’d think some too-enthusiastic arguments for OS (like Eric’s) are a bit overdone.

    A) The fact the OS is a better development model than CS only means that if two projects start on on the same day, the OS will be better.

    B) MS Office has over a decade of headstart over OOo.

    C) It means that one day OOo (or a forked project) will be better than MSO.

    D) Why the hell do we have to argue that this day is now? Why can’t we say – what Andrew Brown seems to admit too – that OOo is a well started project, it will make up for its leeway in five years, but it’s still yet no better? And this is for other projects too. Why can’t we just say the generally, Linux environment will beat MS environment in 2010 – which seems to be true, as Vista is a complete bullshit, totally devoid of ideas – , why do we have to argue that it is better now? For me, it took ’till Ubuntu Breezy to stop parallely using Win/Linux and use just Linux. What I can honestly say that somewhere in 2003-2005, Linux became usable. Remember, we did not have a well working GAIM or Kopete or Totem or KPlayer in 2002, we could not chat with our MSN friedns or watch a wmv video. It is a good prediction, that by 2010 it will be unanimously better. Why do we have to argue that it is better now, while it is not true. Remember, I was often told Linux is better several years ago, somewhere aroung SuSE 6.x (maybe 6.4?). It was a complete crap, ALSA was experimental phase, OSS did not work for a simple SoundBlaster Pro card etc. It is an amazing development Linux and other OS applications had in the recent years. However, there is still many ways to improve – especially integration of different apps, f.e. having a central place to regiser databases like ODBC Data Sources in Windows, and not register them in OOo.

    I believe stating that everything is cool now lowers our credibility to the outer world, and when in 3 or 5 years it will be really cool, they would just not believe us.

  14. RMS started the open source movement, eh? I’m sure he’ll be _very_ glad to hear that. This dude is without clue 1.

    Please, don’t feed the troll.

    Richard

  15. Pete, your turds are up there for entertainment value with a Steven Wright monlogue; you can dump them in my comment section any time.

    Yes, Jim Thompson is an annoying little shite, but he trashes his own credibility a lot more effectively than he attacks mine every time he opens his mouth. So I figure on just ignoring him until he chokes on his own bile.

    But you are so wrong about 1911s. Your medium calibers like .40 and .45 are the signature of a gentleman who doesn’t care to shoot namby-pamby little popguns, but has too much taste and discretion to carry the true declaration of penis-size insecurity — the Desert Eagle. Harrumph.

  16. Specialized software needs specialized coders
    When you talk about software that involves specialized knowledge, like word processors, than you’re talking about a different entry level. Some softwares, like compilers, get a lot of attention from specialists because they’re such a fundamental piece of the puzzle. Other projects (say, a compiler in a non-Algol-derived language) attracts less attention.
    Many hackers will get people to join in projects that involve infrastruture (mail, etc) or systems programming, because many people are knowledgeable in that domain.
    To hack on a word suite, it’s not enought to know the language (C++, etc). You must know *what* to code.
    The same applies to very specialized software, like image processing or multimedia. For instance, the BBC wants to develop an open-source codec for streaming media. The prerequisite is that you’re able to understand wavelets. To do that, you must be knowledgeable in mathematics at least at the level of an engineering career to even begin to understand the documentation on wavelets.
    In OO.org’s case, there’s a huge code base to be learned. This is the part where documentation really is crucial. If you don’t have by the beginning of the project, then you better take remedial steps and contract a technical writer (GNOME did this recently). If you project grows to a big size, you can’t expect help from people if they don’t even have a roadmap.
    The part that ESR gets wrong is where he creates this mistique that coding is easy. The free software community is ridden with sloppy code, made by people with very little discipline in programming, as well as the opposite end, where you have the very best experts in something (OpenBSD hackers are the guys who most understand C programming in the open-source arena).

  17. “incontinent anal avenger”

    Fsck me…that’s going down in history HA HA HA

    ;-)

  18. ESR, I suspect that the only man alive that can carry off the DE .50 look is Vinnie Jones (in the movie Snatch).

    My brother in law has the 8″ extended barrel DE…for hunting only. I think we can all agree that to normally carry *that* around would reek of homo-erotica.

    Me…I’m a .40S&W guy…I like my lumps of lead to soundly leap the 1000fps hurdle ;-) …but not by too much…1100-1200 is just fine

    1911s *do* kick ass though…no denying it…but my P226 whups ‘em

  19. I don’t see how Open Office’s history as a proprietary program somehow makes it less open source than Postgres (oops), AbiWord. Opensource.org encourages companies to make a fortune on the secrets in their code as long as they’re secrets, and then open source the project when they’re not so secret, and cash in on the movement. The Magic Cauldron makes a similar argument.

    However, I do understand the point that Sun simply can’t let go of Open Office. FSF gets copyright in order to release things under the GPL. Sun gets copyright under terms premitting them to grab your code and put it into proprietary projects while denying you that same privelege of their code.

    Now, I know the BSD licenses do the same, and that Raymond has said the GPL isn’t necessary, because closing off the source by definition closes off the benefits of open source. OTOH, I’m not arguing that Sun shouldn’t be *allowed* to do what it’s doing; I’m arguing that Sun’s actions are closing off the source, and they’re getting predictable results.

  20. Back on topic…

    The author of the linked article strikes me as being a ‘power user’ (for want of a better term), in that he his comfortable within the linux world and is confident enough to take advantage of its offerings at many different levels of maturation. Bravo for him. What he *isn’t*, though, is a ‘member’ of the OS community, he is only a benificiary.

    His decision to hold up OOo as an example of how the OS development model (as championed by ESR et al) is critically flawed because, being an ‘outsider’, he failed to grasp the ways in which Sun *themselves* are to blame for failing to correctly engage the OS community.

    I am no expert on the fine details of the licensing Sun uses for OOo, but simply declaring that one can have full access to the source, while *technically* establishing it as an OS project, is rendered somewhat impotent when there is a significant ‘cost of entry’ to the project, as I understand there is in Sun’s case, with all the bureaucracy.

    What this means is that Sun have effectively isolated themselves from the very OS ‘gift culture’ that would otherwise get involved.

    The author’s fallacious reasoning has caused him to mischaracterise Sun’s failure to correctly participate in the OS community, as a failure of the OS community itself.

  21. For the record:

    – I agree with Russ, that was a fine flame. Best since Mark Wisner(?) flamed me way back.

    – I have nothing to do with http://esr.1accesshost.com/, but thanks for the pointer.

    – I shoot (and used to carry before I moved to a state with no carry law) a .45. Why does a 9mm handgun hold so many rounds? Because it has to.

    – My penis is fine.

    – I think Lisp is a fine language. It needs improvement in several areas, most notably networking and the ability to launch code on a parallel architecture.

    – I’m also a FSF “associate member”.

    – I (also) use emacs

    – I (also) wear a mustache, no mullet.

    – I don’t need your permission.

    – Linux wasn’t particularily ‘swift’ to develop, either. (And don’t take this as a linux-flame, please. Linux is fine.) Good software takes time, especially as the number of people dropping code into it increases.

    My “malfunction” (to use your term) is that ESR moved to usurp software freedom and sell it to business, moving himself into the spotlight. He makes claims that the evidence shows to be false, and won’t issue corrections for things he says that are plainly wrong. “Open Source” as a development model and idiology is fine, but ESR’s continuing fight to pit “Open Source” against “Software Freedom” is wrong.

    Yes, I cut the post short. Duties in the “big room” called. If Eric understood why (current) Copyright law requires the “paperwork”, then he wouldn’t attempt to point to it as a weakness.

    If you enjoy flaming, flame away. I enjoy it, even when its directed at me.

    As for aging hippy… well, my hair is sure to be shorter than yours, and Eric is older than I am. Most of the parents of my kid’s friends think I’m somebody’s boss at MCBH. Must be the haircut.

    – I may be annoying you, but I guarantee that I’m anything but ‘little’, and that its likely I can take you in almost any form of 1-on-1 combat. Yeah, we can even go to the run range and punch holes in paper if thats your choice of contest. iI you or Eric want to fight, bring it. Come to Hawaii in January for TPOSSCON, I’ll be there.

  22. Hey man, saying a thing is one thing — proving it is an entirely different matter. It’s all well and good to talk about “the evidence,” but without actually citing it, you’re engaged in nothing more significant than yap-flappery. Why don’t you write up a nice, well supported article on the problems of ESR and stick it on your blog, and we’ll take it from there; because thus far, I can’t see the reason for the rhyme.

    As for your challenge to a fight — well, you sound like a Big Scary Man (TM). Sorry, I’ll pass. You just come across as way too tough and secure when you engage in internet posturing. I’m going to go cry myself to sleep now, ta ta.

  23. Jim, I think your understanding of ESR’s position on open source is flawed. As far as I can tell he is not pitting open source against software freedom — if anyone is doing that, RMS and the FSF are. Eric’s position is that software freedom is a wonderful thing but it cannot be pitched to business because business doesn’t operate on such lofty principles. Business decisions are based on cost/benefit analyses and, unfortunately, fads and trends (“industry best practices” is the handwavey term for this). So “open source” is a fad and trend deliberately crafted to inject free software into the business space (or at least, legitimise it there).

    This idea, while on the surface seems great, is IMHO fundamentally misguided. Microsoft is the most powerful software company in the world not because they make the best software — clearly they don’t. But they fill a social niche that OSS probably can never fill. Microsoft listens to feedback from business customers and crafts solutions to improve the operations of business. They then use their power of “benevolent dictator” on the PC platform to say “this is how things will be done”. The result is not always the best solution but it is one which helps office workers and business developers do their jobs more efficiently.

    With open source, there are competing standards, various factions (Linux vs. BSD, Perl vs. Python, KDE vs. GNOME), no one clear standard. With Microsoft there is one standard. You use Windows on the desktop, Windows Server in the back office, Word for documents, Excel for calculations, Outlook for email. For development you have a choice — pick your poison, Visual Basic or C#. This frees business developers and managers from having to make decisions on what to use and deploy.

    It’s basically business managers avoiding the dangerous novelty of thinking but unfortunately that’s how the free market works. It is why Microsoft has enjoyed so much success and Linux/open source, less so. When it comes to the enterprise, the biggest Linux success stories have generally been socialist countries with a healthy fear of Murkan slash-and-burn cowboy capitalism standardising on a non-American technology.

  24. Businesses don’t run Linux. They run Red Hat (which is a simplification, but it’s one way for business managers to avoid the dangerous novelty of thinking).
    -russ

  25. Russ,

    What is the economic incentive for business to run RedHat (.vs Unbuntu or gentoo or debian, or CentOS or Fedora, or hey… SUSE)?

  26. I would like to announce my new website, “Jim Thompson is a Gaywad”, to be written with the noble purpose of fostering civilized debate in the F/OSS community. By revealing all the ways in which Jim Thompson is a gaywad, I hope to encourage a better understanding of difficult subjects such as copyright law and programming. My sarcasm is becoming self-defeating so I’ll stop here.

  27. Bruce Perens (who continues to watch his step) comments in the relevant Slashdot article:
    “To get the full benefit of Open Source, you need a big enough community to drive work for many different agendas rather than mostly one agenda. The problem is tha OO is still mostly Sun.”

    Hey Eric, any chance of adding a “preview” feature to comments? I’m always insecure about my html.

  28. Gosh… why are Linux lovers so angry ? I use windows and MS office. Have not found a bug that stops me from doing my work. I noticed many of the OO users say the same. If you think Unix is good, that OO is good, use them. Never used them, but I think they migth be good. Stop arguing what is better. All the time you spend arguing could go into fixing a bug in your favorite piece of software or pressing your favorit software company to fix that bug that bothers you.

  29. Pete Bessman:

    4) Lisp may not be dead, but it’s a zombie, and deservedly so.

    I’m going to forgive you for this one because it would certainly be true if everyone wrote Lisp or Scheme the way ESR did in his example. :)

  30. Open source only facilitates detection and removal of bugs – it does not remove the bugs by itself. The author seems to be confusing a product with the process behind it. As for bug removal, my experience has taught me that the people in charge of the project matter most – as in, how motivated they are. Abiword (another open source word processor) has problems compiling on Solaris. They did not even acknowledge when I reported bugs and even sent them the fixes for it. This is not the problem of the process. The process allowed me to find the bug and report it, which would have been nearly impossible in a closed source program
    Regards,
    Maidros

  31. Eric — I was modestly irritated by your citation of Barton Miller’s “Fuzz Papers” — mainly b/c they seem to have been last updated in 2000. I did some digging around, and there are references to several Coverity studies released in 2004 & 2005 that strongly support the argument that Open Source has a significantly lower bug rate than closed/proprietary software, as well as some Forrester research paid for my MSFT that is highly unfavorable to Windows. You can find the data sources (such as they are, meaning, derived from inferences and unpublished sources) referenced up at http://yadomejutsu.typepad.com/my_weblog/

    Regards,

    Jeff T.

  32. Even though nobody seems to be paying attention to me (and for good reason), lemme respond to Jacque Terrence and Jeff Read (among others):

    Jacques:
    /* When you talk about software that involves specialized knowledge, like word processors, than you’re talking about a different entry level. Some softwares, like compilers, get a lot of attention from specialists because they’re such a fundamental piece of the puzzle. Other projects (say, a compiler in a non-Algol-derived language) attracts less attention.


    To hack on a word suite, it’s not enought to know the language (C++, etc). You must know *what* to code.
    */

    I’m sorry, Jeff, but the behind-the-scenes work of a word processor simply isn’t all that complex. Emacs breaks up files to be worked on in basically the same way word processors do (compare http://www.abisource.org/mailinglists/abiword-dev/98/August/0028.html and http://www.chemie.fu-berlin.de/chemnet/use/info/emacs/emacs_19.html). The main difference between the two is that emacs doesn’t keep formatting information for the text, but that’s easy enough to add.

    And spell check? Ever hear of aspell? No, it’s not ready-to-go, but it’s a start. Just work around the text formatting. As for grammar check, ever hear of compilers? That’s probably a good place to start.

    I might agree with you in regards to complex software, but not word processors.

    Jeff:
    /* As far as I can tell [ESR] is not pitting open source against software freedom — if anyone is doing that, RMS and the FSF are. Eric’s position is that software freedom is a wonderful thing but it cannot be pitched to business because business doesn’t operate on such lofty principles. Business decisions are based on cost/benefit analyses and, unfortunately, fads and trends (”industry best practices” is the handwavey term for this). So “open source” is a fad and trend deliberately crafted to inject free software into the business space (or at least, legitimise it there).
    */

    I agree with this here. Of course, my opiion of what ESR believes doesn’t really add up to much.

    /* This idea, while on the surface seems great, is IMHO fundamentally misguided. Microsoft is the most powerful software company in the world not because they make the best software — clearly they don’t. But they fill a social niche that OSS probably can never fill. Microsoft listens to feedback from business customers and crafts solutions to improve the operations of business.
    */

    How is OSS not able to “listen to feedback from business customers to craft solutions”? Ever hear of Cygwin?

    I can’t see how competing standards make this any harder. Could you elaborate?

  33. Max Lybbert:

    Sure.

    The crucial bit of insight came from an acquaintance of mine in the video game industry. Game developers are notoriously Windows only. If you’re a Linux user and want to play recent games you have two options: dual-boot or (in some cases) Cedega. This is partially due to economics (the revenues to be gained from catering to the Linux crowd are near negligible) and partially because Microsoft’s Direct3D API is considered by many to be the superior solution for high-speed 3D graphics. The reason why it is so superior is because Microsoft solicits information on what game developers want, what gamers want, and what kinds of features hardware vendors are implementing (or are likely to implement in the future) and then they write the API. Because they’re Microsoft, they alone get to set the standard and nine times out of ten, it’s more than good enough. The end result is that Direct3D is the standard for games and is quickly becoming the standard for other 3D applications too and OpenGL faces rapid marginalisation (I hear tell in Vista it is simply a compatibility wrapper for D3D). In addition, the Xbox series of game consoles are by far the most beloved of developers because of the robustness of the Direct3D API vs. the paucity of their competitors’ APIs, and will pose strong competition if not trouncing Sony and Nintendo outright in the next round of console wars.

    In the case of OpenGL, you have those “competing standards” everyone talks about, and usually they are vendor-specific extensions. The OpenGL ARB has been heavily criticised for sluggishness when it comes to implementing features like pixel and vertex shaders and as Microsoft sets the next round of Direct3D standards it finds itself falling behind when it comes to features developers want again, and therefore losing more hearts and minds to Microsoft.

    Understanding the Direct3D vs. OpenGL debacle helped clarify for me the history of Microsoft “standards”: Word as a standard document format, etc. As the 800-pound gorilla, Microsoft gets to set immediately the standards that people want immediately, that take standards committees and competing developers years to hash out and get the market/mindshare for. As long as they are not too annoying to the vast majority of their users (who really care little for technical or software-freedom issues), they will be accepted. This kind of thing is far less feasible in open source, wherein everyone is free to create their own standard and there could be many competing, somewhat incompatible ways of doing the same thing on the same machine.

    All of which puts the lie to the idea so beloved by ESR and other free software hackers — that you can love the free market and still hate Microsoft. Microsoft is exemplary of the free market in action, and because of its “monopoly power” can provide a legitimate competitive advantage over open source (albeit not a technical one; but then when has technology for its own sake ever mattered to anyone but hardcore geeks?).

  34. FYI, I’m no fan of Microsoft’s, but playing “spot the inconsistency in ESR’s philosophy” appears to be the game of the month ’round here; and I’m not exactly a fan of the free market either, not when it exists on such a chronically mentally ill substrate as the American population proves itself to be. Besides, if what those free market proponents say about monopolies is true, then surely with strange aeons even Microsoft may die…

  35. Jeff Tyeryar: The Fuzz ppapers are still the only study to directly compare the error rates of pairs of functionally-equivalent programs. Thus I think they’re a stronger argument against Brown than the Coventry study or similar results.

  36. Jeff, I can see your point now. My understanding of it isn’t that open source is incapable fo listening to the customer, but that open source (more often than not) tries to follow standards (as a poor man’s way of listening to customers), and those standards (more often than not) languish in committee and debate the placement of commas. Meanwhile, agile companies are able to set their own standard and ship the product.

    OK. While I can’t think of any recent (or complex) examples, I do know of times that open source ate somebody else’s lunch precisely by a fast-moving advantage. The two that come to mind most readily are mbox (BSD) and TCP/IP. So it is possible, and it is a strategy that open source tries to use.

    But mbox or even TCP/IP aren’t quite on the level of DirectX. OTOH, DirectX, like most Microsoft technologies, took three or four versions before being something worth looking at. That’s not a fast-moving technology, although it hasn’t been seriously challenged by open source as of yet.

  37. To close the circle: it really explains the above article. Microsoft Office is buggy, but ubiquitous enough to make it a Fact Of Life, like income taxes. OpenOffice had better have a bug count approaching zero, or else something enormously advantageous over M$, to dislodge the user from their installed base. The article is basically “I don’t see what the open source hype is about; nothing’s really different” piece. Business users who haven’t cuddled up to the buzzwordy open source mystique are likely to feel likewise.

  38. So, if Open Office sucks and isn’t a real open source project, then what is the open source community’s answer to MS office?

  39. >So, if Open Office sucks and isn’t a real open source project, then what is the open source community’s answer to MS office?

    Nobody’s denying that OO is a real open-source project. It’s one with some chronic problems, that’s all. As JWZ said once,. open source is not magic pixie dust.

  40. There’s always Uncle Duke’s attitude: hunt deer with a mortar. As the guy from the NRA said, it messes up the trophy, but it sure does soften up the target.

  41. “Nobody’s denying that OO is a real open-source project. ”

    Your very first poster:

    ‘Anything “open source” from Sun Microsystems is just their idea of a marketing gimmick. ‘

    Another:
    OpenOffice is obviously not a typical open source project.

  42. Quibble:

    “Anything open source is just . . .a marketing gimmick”, i.e., it is only a marketing gimimick and nothing more. Unless open source is no more than a marketing gimmick. . .

    Of course, that semantics web page would say there are no buckets such as these, but from a logic perspective, I say the logic is well founded.

  43. Big projects require structure and documentation. Otherwise, you get 17 people working on the Really Kewl Thing, and one guy who’ll get around to writing the boring user interface (and its testing) when he done playing NetHack.

    It’s been my experience that Open Source software falls behind Closed Source software in the areas I care about:

    1) File interoperability. I don’t care if your file format is better. If I send it to a print bureau, they want it in the file format they use. This is also why I don’t use OpenOffice for much of anything.

    2) User interface design. Open Source software has ALWAYS fallen down on UI design and usability in my direct experience. I characterize it as being written by coders for coders…and actually listening to the whining lusers about what THEY want the software to do is like, a major downer, man, when you could be writing a Wicked Kewl Hack.

    3) Attitude and disdain for the end user. There is a very strong perception that if you can’t program, at least in a shell scripting language, that you’re “too dumb” to use Open Source software.

    Software isn’t religion. It’s creating tools, optimally tools that are useful for people other than the creator, and ideally, tools that are useful to people who couldn’t create them themselves.

  44. Ken: There’s some justice to your complaints, but we’re getting better. There’s a lot less disdain for end users trhan there was five years ago, and much more effort to get UI right (though we’re still in early days for that).

  45. Neal:
    /* Anything open source is just . . .a marketing gimmick”, i.e., it is only a marketing gimimick and nothing more. Unless open source is no more than a marketing gimmick. . .
    */

    The original quote is “Anything “open source” *from* *Sun*” is a marketing gimmick. To Sun, open source *is* nothing more than a marketing gimmick. Try looking up the Jini Sun Community Source License (SCSL — http://www.jini.org/jiniorgfaq.html#using4 they called this “innovative” at the time, but they’ve since moved on to using a real open source license).

  46. Ken:
    /* It’s been my experience that Open Source software falls behind Closed Source software in the areas I care about:

    1) File interoperability. I don’t care if your file format is better. If I send it to a print bureau, they want it in the file format they use. This is also why I don’t use OpenOffice for much of anything.
    */

    You do realize that file interoperability comes from open standards. For instance, Scribus ( http://scribus.net/ ) outputs PDFs so that you can take the document to a print bureau. Many open source programs are proofs of concept for open standards. And many other open source programs handle open standards for this very reason. Microsoft Word is not an open standard, but Open Office (and AbiWord) can write it just fine.

    /* 2) User interface design. Open Source software has ALWAYS fallen down on UI design and usability in my direct experience.
    */

    I’ll agree to a point. This is, in fact, why I’m in school right now — to learn interface design.

    /* I characterize it as being written by coders for coders…
    */

    I’m trying to figure out GUI paradigms that are as elegant as command line interfaces, etc. I agree with Apple on this one — if you need wizards it’s proof your UI is broken. I especially hate programs that consider me a complete idiot, and drop useful features because they wouldn’t want to confuse little ol’ me. OTOH, I’m currently working at a help desk (largely to learn what works and what doesn’t), and I realize many users just don’t get it. In that case, we need to take another look at the UI. Remember the iPod — nobody needs to read the manual about how to select a song, it all makes sense for even the biggest klutz.

    /* and actually listening to the whining lusers about what THEY want the software to do is like, a major downer, man, when you could be writing a Wicked Kewl Hack.
    */

    Don’t let the respectable companies involved in open source hear that. IBM, HP, Red Hat, Novell, etc. all have real programmers who aren’t so concerned about “wicked kewl hacks.” In fact, I doubt you’ll ever find decent software from the strawman programmer you created.

    /* 3) Attitude and disdain for the end user. There is a very strong perception that if you can’t program, at least in a shell scripting language, that you’re “too dumb” to use Open Source software.
    */

    That may have been once upon a time, but it’s not true in the UI circles. I agree with Torvalds that Gnome errs too far on the side of the stupid user ( http://mail.gnome.org/archives/usability/2005-December/msg00021.html ), but KDE is just as worried now about useful UI. So is Mozilla. It may have something to do with a paper ESR wrote ( http://www.catb.org/~esr/writings/cups-horror.html ).

  47. Ken Burnside:

    Exactly! I mean, end users don’t want to compile their own kernels! Why should they have to?

  48. Jeff Read spoo’d forth:

    >Exactly! I mean, end users don’t want to compile their
    >own kernels! Why should they have to?

    Because every user who learns to compile his own kernel (And thereby take on fill responsibility for the state of his system) is one less user I have to do it for.

  49. Kyle, I was being facetious because Ken’s complaints about Linux seem to hail from about 1996 or thereabouts. He seems blissfully ignorant of KDE, as well as the PDF export support in OpenOffice. Unless his print bureau specifically rejects PDFs in which case I’d like to hear that convo. “Sorry, we don’t take widely supported, open industry standards. We only accept proprietary formats.” So I jokingly made reference to the “compile your own kernel” canard repeated by Slashdotters years after commercial distros made automatic hardware detection and loading of the appropriate modules a commonplace part of the Linux experience.

    (I haven’t compiled a kernel in a couple of years myself. And I like that kind of thing…)

  50. As for the whole penis size / caliber nonsense, a .32 revolver proved sufficient for the woman I married, but in all probability, the man who kicked in the door of her apartment probably wouldn’t have done so had it been the apartment of someone with a penis of whatever size.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>