Sigh... There is a lot of hardship in trying to do things right in a legally upholdable fashion. History has shown that what appears to all involved -- coders and legal experts alike -- to be legally sound, may turn out to be evil itself after all.
Way back when, Public Domain was the legally correct term to denote free software. Along comes a lawyer who some people think should be put against the wall, and poof: asserting copyright became a necessity.
Along comes Richard Stallman, and the GPL, and all of a sudden the term "free" takes on an entirely new meaning. Whichever side of the divide you're on, the consequences baffled coders and legal experts alike.
Sun's real problem with Java has been with openness: applying patches from the field, providing easy access to the good bits, and allowing less than perfect ports out. As it stands, I'm using kaffe to much greater effect than Java, because of Sun's stoopid handling of the issues.
All of the "circumvention" instructions simply said unzip it.
If I leave my door open, stick a poster on it saying "Lot's a money inside! Don't open this door unless you sign a form promising not to take it!", I'm not only rather stupid, but I think I'd be open to a countersuit by a captured intruder or possibly even the government for solliciting a crime.
This line of thinking is always on my mind when I see letters such as this one.
If I were to ROT13 my trade secrets, publish them in the New York Times in that format, and then started suing, isn't there a legal way to nail me?
Actually, Sun started out way back when with the BSD code, calling it SunOS 3 and later, 4. SunOS 4 was pretty close to ancient BSD.
Then, Sun had the brilliant idea to ditch the BSD code and instead use the yucky System V code as its basis for Solaris 2. It took them years to get the thing stable (2.5.1 is the first Solaris that I'd consider using for anything serious).
Solaris is a serious OS these days, but it bears as much resemblance to BSD as Linux does. Right now, all three have evolved so far as to make any OS of the early ninetees look ancient.
By the way, where do you get the "mortal enemy" bit from? Sun is a heavy invester in Intel based software. It's just Microsoft they're not big fans of.
Similarly, if you wanted to determine who the most prolific scientific researcher is in a field, would you gather data by simply grepping for names in the texts of papers?
Hmmm, this reminds me of the infamous Quotation Index used in the scientific world. Back when I studied sociology, a professor of mine would spend five minutes each college blasting the practice. As it turned out, a number of his colleagues were quoting each other, thereby bumping each others ratings. "On the effects of offering free ballpoints to interviewees", being referenced by an article on "A critical review of free ballpoints", referenced by the rebuttal, ad nauseam.
Doesn't it strike a familiar note in a forum driven by mechanically established karma?
Yeah, the statistics are poor across the board. According to this site, I contributed a total of 5502 bytes of code to Open Source projects. Well, make that Ethereal. I couldn't find Mozilla at all! This is weird, because Mozilla, though not being the most important and by a large margin not the first open source project, is the project that made the term Open Source a household (and boardroom) name.
I pride myself in contributing lots of humble changes or fixes to lots of projects. Still, I'm not in the business of getting my name in the AUTHORS file of every project under the sun (even though it a nice side effect of a hobby that exploded:-) My motivation is to make my life easier and more fun, while contributing to the public good.
The most flattering thing that was ever said about my contributions was hidden in the URL of an interview by Feed Magazine. When I showed this URL to my family, the reaction was "wait a sec! Bottomfeeders? Isn't that a bit derogative?". It took quite some explaining to make it clear that it was the culmination of what I've done over the years: I've joined the hordes of folks who, by submitting small patches, fixes, bits of functionality, have made the difference between making Open Source a hobby of a select few, and making it a (possibly) useful tool.
Oh well. I hope the folks at Orbiten will improve the situation (I'm sure their mailboxes will suffer the slashdot effect), and make the relative merit of their measuring methodology more clear. It is gratifying to see that someone picked up the odious task of trying to quantify what Open Source has to offer.
As a side note, I lost my previous (very well written, thanks for noticing!) reply to this message because of accidentally clicking on a banner ad on slashdot. Oh, for the irony!
Eric Raymond has a pretty neat description of a process for resolving this. He calls it homesteading. The basic idea is: someone abandons a project, someone else thinks its cool, announces his intent to pick it up, and hey presto: there's a new project owner.
Losing key staff is no longer the exclusive realm of corporations. I sort of surprises me to see this argument brought up in the context of open software!:-)
Yup, and that is why the thing was taken out of the default user interface (that's what "broke too many sites" translates to).
On a related note, just when Microsoft (partially) mended their ways by making their distributed download sites more predictable and thereby less likely to cause "cache aliasing", enter Akamai, whose business model seems to be based on URL aliasing.
I'm talking about the effects that content distribution has on caches like Squid. In a typical week after a new update of Internet Explorer is released, I find my Web caches filled with identical copies of files downloaded off different sites like msvaus.www.conxion.com and mskyus.www.conxion.com -- with an "intelligent" process at Microsoft deciding which site to point this particular download to. This hurts with 1000 users sitting behind the slow connections to our offices. Renegociating the bandwidth pricing is not in the cards either:-)
So, does anyone have any info on how Akamai can work together with Web caches?
Well, I like AvantGo a lot. I'm not as happy with their "open" software. I've spent a lot of time reading the code and figuring out what the heck it does, and didn't get an inch closer to setting up my own AvantGo portal (which I only need because the stuff I want to sync on my Palm is not out there on the Internet; setting up a custom channel to use their server is trivial).
Sigh.
Incidentally, the VNC thingy is cool, but it doesn't run on my Palm V. I tried recompiling the thing, which worked, but now user input is broken. This is a problem with a lot of cool toys for the Palm: when source is available, it rarely compiles with a recent GCC and Palm SDK. More of this stuff should be on sourceforge, so it can be brought forward as the Palm platform develops.
Actually, I had forgotten about the stupid naming!
The scarcity of prerecorded adult material is what I always hear to be responsible for the rise of VHS (it was the time when the concept of the video sandwich surfaced: one family movie (Charles Bronson), one kiddie movie (Bambi) and one general interest movie sandwiched in between. Some video stores even had a special deal on sandwiches.)
Phil{,l}ips seems to have exclusive rights to really big marketing screwups!
2) These are on tapes, they must be rewound, no chapter skipping or interactive features.
Well, maybe I'm not a true geek, but I pop a tape in, hit Play, Fast Forward to skip the FBI warning, maybe Stop on the way to get a fresh beer, and finally Rewind when the flick ends (by which time I am definitely ready to go take a leak while the tape rewinds).
In contrast, DVD requires you to take a leak while the FBI warning is playing, and the shadow of the region coding still looms over it.
Speaking of which, has anyone yet filed a complaint at the WTO over this illegal market compartmentization? The excuse of protecting the European first openings against early imports from the US sort of falls flat on its face when you see titles from 1934 with region coding.
Sigh... Eons ago, there were three video standards: Phillips's Video 2000, Sony's Beta, and VHS. Debates still rage about which of the three was, but near general consensus holds that VHS was the lesser option. As evidenced here, Video 2000 is no longer even remembered.
Enter gynae^H^H^H^H^Hcontent availability. It is rather sad, but VHS got carried by adult content availability, because Phillips refused to release non-family titles.
DVHS's saving grace may well be that this thing will play existing VHS tapes. The so-called "specs" at the Web site weren't clear on whether or not the US edition would be able to play European PAL videos. If it would, the player might very well get a lift from the huge numbers of Americans taking souvenir videos back home from Europe.
Brilliant! The horrible thing that crossed my mind is: did John Taschek intend to write something that would apply to any topic by just editing a few names?
The CNNfn poll has its weaknesses in the way the questions are put, but that is as nothing compared to an interview I saw Moneyline's Stuart Barney do with Bill Gates. Barney: Mr Gates, do you think the governments actions will slow down innovation by Microsoft if it succeeds? (not a verbatim quote, but pretty damn close). I mean, that reeks of Microsoft interviewing Microsoft! Not a question about the alleged illegal coercive tactics, not a word about stifling competition by using hidden API's, no nothing that even relates to what Joel Klein mentioned just minutes before to explain why Big Brother DOJ was going after innovator Microsoft.
If I were to improve the poll without sacrificing CNN's trademark soundbite style, I'd word it:
Do you think antitrust protection is relevant when applied to the software industry?
Yes
No
Unsure
Do you think a company breakup would help or hinder the software industry as a whole?
Help
Hinder
Unsure
Do you believe the proposed behavioral remedies alone could suffice in keeping the market competetative?
Yes
Maybe
I didn't listen/understand/forgot what Joel Klein had to say
Oh well, you get the point. I'm very frustrated with CNN, their reliance on company executives to tell the eternal truths is so big, they usually forget to ask the real questions.
like they only NOW figured out that people hate it?
Intel doesn't give a damn about whether consumers like it ot not. The serial number was killed before the PIII was even on the market: most motherboard vendors disable the processor id through the BIOS by default.
The reason it got killed was most likely the fact that it is utterly useless for its stated purpose: to provide a secure identity on the Internet. Even if a web browser reports the serial# to a Web site, how is the Web site to know that the user didn't just stick his own header line in? See http://www.counterpane.com/crypto-gram-9902.html for Bruce Schneiers scorching review of its value as a security tool.
I expect that the terms of break-up will REQUIRE that communication between these two SEPARATE companies be conducted on open channels, via published APIs and public company press releases.
I just listened to Joel Klein from the government side on CNN, and one provision in the proposal requires the OS division to disclose the API to all software developers in as much detail as it would to the Office division. This is good, this finally makes it illegal to cripple competing developers by building in secret API's (and there are lots of them in Windows).
Second, the proposal explicitly makes it illegal to threaten hardware vendors with less favorable conditions if they wish to ship hardware with competing software.
As far as I'm concerned, Microsoft need not be split up if these (and a few other) stipulations can be enforced. Find a hidden API in Windows used by Office? Sue them. Internet Explorer using a better API to DNS than the documented one? Sue them.
Sigh. If the provisions of the proposal that are just sidelines to the case now were more easily enforced in the industry generally, the world would look a lot cleaner, and the entire trial circus that is going on now would be moot. I am as tired of the government case as I am of Microsofts, I do not think a breakup per se will do anyone any good, but I'm glad to see some rather common-sense ideas about fair competition in the software market gain a foothold in law.
I have some problems with the underlying assumption that the vendors mentioned do not care about Linux.
In a lot of commercial software companies, there's a strong sentiment not only in managers, but also in development staff, that supporting Linux opens up a Pandora's box. It's not so much compiling for all those platforms, because most developers and managers can be convinced that the API is not very different, so few actual code changes are required. The big problem is in testing. For a glimpse of the problem, look at Netscape's Tinderboxes, and remember that Netscape is one of the most Linux minded commercial developers out there.
For a commercial developer that sees Linux as a platform deserving to be served with tested software, the cost of testing each of their new releases on each distribution quickly becomes prohibitive, and higher up management resolves this issue by drawing a line in the sand. There is a risk that this line winds up including no Linux at all.
Oh well. There is no clear answer here. I'd like to see more distributions supported (as much as I'd like to see commercial software for Linux being tested on *BSD under Linux emulation). I'm just afraid that some companies will get second thoughts about Linux if they get roasted over not supporting less prominent distributions.
Cool idea. Not likely, but cool nonetheless. Problem is, it won't buy the UNIX users anything, because the API to the GUI and to the OS is pretty entrenched in the Explorer code.
Just look at the differences between IE on Mac and IE on Windows. I don't believe that much code is shared.
Re:This is a BIG Win for Microsoft
on
Microsoft Loses
·
· Score: 1
Personally, I'm rather disappointed. I couldn't give a damn about most aspects of the trial, short of one: the alleged Microsoft business practice of ramming stuff down other vendors throats. For all I care they'd just be fined US$1 over their threats to hardware vendors that wanted to ship Netscape, as long as they are chastised for this alleged conduct.
Well, I guess the judge just said that there has been no bullying of vendors, just of the consumers by the pricetag of Windows'98. The judge will have to relent on this one, development cost of IE4 is nothing on Microsofts budget, pennies per copies of Windows'98.
In my view, Microsoft didn't quite win. The marketplace lost.
It is pretty standard for extended versions of GCC to break portability. It's hard enough to put, say, Altivec code generation in a compiler and get it to work, let alone doing so in a portable fashion.
Same thing happened with the Intel Pentium optimized GCC. It just takes time and a lot of hard work to clean up such a code base. The success of the Apple patches will be judged on the question of how much work it will be to clean it up, and how much time Apple or volunteers will be able to devote to cleaning up the code.
I for one would be happy to see ObjectiveC go forward with Apples backing.
I used the Ericsson I888 cellphone, and while it works great in California between LA and SF, it doesn't work along the bits of Highway one not in this part of CA. My next stop was Wisconsin and Illinois, and no coverage at all there. A small sample of the US, I know, but...
Besides, roaming charges will kill you, and I find the I888 user interface sucks (try reading an 80 character SMS message on it).
Renting a mobile phone locally is the better option for now.
Well, surely it is in their best interest to work together with the "pre-fork" maintainers. Merging in changes from a different tree (whether it's from Corel to Mainstream WINE or vice versa) is a major undertaking.
Netscape is doing a tremendous effort at integrating "outside" patches even if they come with no immediate added value to themselves, and if you track netscape.public.mozilla.* and look for the Netscape internal developers comments on that situation, you'll realize just how formidable the task is.
I am extremely relieved to see companies like Netscape and Corel sticking it out, trying to meet the conflicting goals of keeping the shareholders happy (release on time, on target) and the goals of the open source community (release when it's ready, and fix *this* bug now even though it doesn't hamper customer acceptance of the product).
I just got a Dell CPx (with Win95 preloaded -- let's all say a heartfelt "Duh" for corporate purchasing).
Anyway, the X server doesn't support APM -- so, the X server has to be out of the way when travelling, and forgetting that will result in a hard crash. I'm not sure whether this is because the ATI Mobility chip is too complex to get X right, or if there is a bug in the chip. Whatever the cause, it's annoying, and Windows works just fine (well, at least in this respect:-)
The cool thing about getting Dell to ship real complex hardware like laptops with Linux, is that if Dell gets too many complaints, they'll chat up ATI to do something about the APM problem, and help the Xfree people to bring hardware support up to par with Windows.
When Digital Equipment sold out on DECWindows to turn it over to OSF, my trust in closed source died. The DECWindows code contained bugs, sure enough, but as part of VMS at least some customers got source free of charge, allowing me to work around those bugs. The OSF bits were not included in the microfiches, so: no source code, no bug fixes.
Letting the openness of the source out of the picture, there's the user experience. While some nice features from (then new) window systems such as Windows and CUI were brought in, the eye candy brought in as part of the Motif effort effectively killed the clarity of the user interface.
My favorite example is the radio button. DECwindows had a clear concept: an unchecked radio box is depicted by and empty circle, a checked radio button is a black circle with a big red dot in it. With Motif's "improved" 3D look, one really had to look hard to see which was selected and which wasn't. "Hmmm, the lower pixels are darker than the top pixels, so this is probablyu the selected radio button".
Hey, come off it: NT *is* POSIX compliant. Needless to say, POSIX doesn't mandate the POSIX layer to be useful and therefore NT's implementation isn't, but it is *compliant*.
Seeing how easy uncertified platforms like Linux and BSD compile each others code with less system specific patches than certified platforms like NT, Solaris or HP-UX, one has to wonder what the role for certification is.
Slashdot Mirror
Way back when, Public Domain was the legally correct term to denote free software. Along comes a lawyer who some people think should be put against the wall, and poof: asserting copyright became a necessity.
Along comes Richard Stallman, and the GPL, and all of a sudden the term "free" takes on an entirely new meaning. Whichever side of the divide you're on, the consequences baffled coders and legal experts alike.
Sun's real problem with Java has been with openness: applying patches from the field, providing easy access to the good bits, and allowing less than perfect ports out. As it stands, I'm using kaffe to much greater effect than Java, because of Sun's stoopid handling of the issues.
There are no easy answers.
If I leave my door open, stick a poster on it saying "Lot's a money inside! Don't open this door unless you sign a form promising not to take it!", I'm not only rather stupid, but I think I'd be open to a countersuit by a captured intruder or possibly even the government for solliciting a crime.
This line of thinking is always on my mind when I see letters such as this one.
If I were to ROT13 my trade secrets, publish them in the New York Times in that format, and then started suing, isn't there a legal way to nail me?
I'd certainly deserve it :-)
Then, Sun had the brilliant idea to ditch the BSD code and instead use the yucky System V code as its basis for Solaris 2. It took them years to get the thing stable (2.5.1 is the first Solaris that I'd consider using for anything serious).
Solaris is a serious OS these days, but it bears as much resemblance to BSD as Linux does. Right now, all three have evolved so far as to make any OS of the early ninetees look ancient.
By the way, where do you get the "mortal enemy" bit from? Sun is a heavy invester in Intel based software. It's just Microsoft they're not big fans of.
Hmmm, this reminds me of the infamous Quotation Index used in the scientific world. Back when I studied sociology, a professor of mine would spend five minutes each college blasting the practice. As it turned out, a number of his colleagues were quoting each other, thereby bumping each others ratings. "On the effects of offering free ballpoints to interviewees", being referenced by an article on "A critical review of free ballpoints", referenced by the rebuttal, ad nauseam.
Doesn't it strike a familiar note in a forum driven by mechanically established karma?
I pride myself in contributing lots of humble changes or fixes to lots of projects. Still, I'm not in the business of getting my name in the AUTHORS file of every project under the sun (even though it a nice side effect of a hobby that exploded :-) My motivation is to make my life easier and more fun, while contributing to the public good.
The most flattering thing that was ever said about my contributions was hidden in the URL of an interview by Feed Magazine. When I showed this URL to my family, the reaction was "wait a sec! Bottomfeeders? Isn't that a bit derogative?". It took quite some explaining to make it clear that it was the culmination of what I've done over the years: I've joined the hordes of folks who, by submitting small patches, fixes, bits of functionality, have made the difference between making Open Source a hobby of a select few, and making it a (possibly) useful tool.
Oh well. I hope the folks at Orbiten will improve the situation (I'm sure their mailboxes will suffer the slashdot effect), and make the relative merit of their measuring methodology more clear. It is gratifying to see that someone picked up the odious task of trying to quantify what Open Source has to offer.
As a side note, I lost my previous (very well written, thanks for noticing!) reply to this message because of accidentally clicking on a banner ad on slashdot. Oh, for the irony!
Losing key staff is no longer the exclusive realm of corporations. I sort of surprises me to see this argument brought up in the context of open software! :-)
On a related note, just when Microsoft (partially) mended their ways by making their distributed download sites more predictable and thereby less likely to cause "cache aliasing", enter Akamai, whose business model seems to be based on URL aliasing.
I'm talking about the effects that content distribution has on caches like Squid. In a typical week after a new update of Internet Explorer is released, I find my Web caches filled with identical copies of files downloaded off different sites like msvaus.www.conxion.com and mskyus.www.conxion.com -- with an "intelligent" process at Microsoft deciding which site to point this particular download to. This hurts with 1000 users sitting behind the slow connections to our offices. Renegociating the bandwidth pricing is not in the cards either :-)
So, does anyone have any info on how Akamai can work together with Web caches?
Sigh.
Incidentally, the VNC thingy is cool, but it doesn't run on my Palm V. I tried recompiling the thing, which worked, but now user input is broken. This is a problem with a lot of cool toys for the Palm: when source is available, it rarely compiles with a recent GCC and Palm SDK. More of this stuff should be on sourceforge, so it can be brought forward as the Palm platform develops.
The scarcity of prerecorded adult material is what I always hear to be responsible for the rise of VHS (it was the time when the concept of the video sandwich surfaced: one family movie (Charles Bronson), one kiddie movie (Bambi) and one general interest movie sandwiched in between. Some video stores even had a special deal on sandwiches.)
Phil{,l}ips seems to have exclusive rights to really big marketing screwups!
Well, maybe I'm not a true geek, but I pop a tape in, hit Play, Fast Forward to skip the FBI warning, maybe Stop on the way to get a fresh beer, and finally Rewind when the flick ends (by which time I am definitely ready to go take a leak while the tape rewinds).
In contrast, DVD requires you to take a leak while the FBI warning is playing, and the shadow of the region coding still looms over it.
Speaking of which, has anyone yet filed a complaint at the WTO over this illegal market compartmentization? The excuse of protecting the European first openings against early imports from the US sort of falls flat on its face when you see titles from 1934 with region coding.
Enter gynae^H^H^H^H^Hcontent availability. It is rather sad, but VHS got carried by adult content availability, because Phillips refused to release non-family titles.
DVHS's saving grace may well be that this thing will play existing VHS tapes. The so-called "specs" at the Web site weren't clear on whether or not the US edition would be able to play European PAL videos. If it would, the player might very well get a lift from the huge numbers of Americans taking souvenir videos back home from Europe.
Brilliant! The horrible thing that crossed my mind is: did John Taschek intend to write something that would apply to any topic by just editing a few names?
If I were to improve the poll without sacrificing CNN's trademark soundbite style, I'd word it:
Oh well, you get the point. I'm very frustrated with CNN, their reliance on company executives to tell the eternal truths is so big, they usually forget to ask the real questions.
Intel doesn't give a damn about whether consumers like it ot not. The serial number was killed before the PIII was even on the market: most motherboard vendors disable the processor id through the BIOS by default.
The reason it got killed was most likely the fact that it is utterly useless for its stated purpose: to provide a secure identity on the Internet. Even if a web browser reports the serial# to a Web site, how is the Web site to know that the user didn't just stick his own header line in? See http://www.counterpane.com/crypto-gram-9902.html for Bruce Schneiers scorching review of its value as a security tool.
I just listened to Joel Klein from the government side on CNN, and one provision in the proposal requires the OS division to disclose the API to all software developers in as much detail as it would to the Office division. This is good, this finally makes it illegal to cripple competing developers by building in secret API's (and there are lots of them in Windows).
Second, the proposal explicitly makes it illegal to threaten hardware vendors with less favorable conditions if they wish to ship hardware with competing software.
As far as I'm concerned, Microsoft need not be split up if these (and a few other) stipulations can be enforced. Find a hidden API in Windows used by Office? Sue them. Internet Explorer using a better API to DNS than the documented one? Sue them.
Sigh. If the provisions of the proposal that are just sidelines to the case now were more easily enforced in the industry generally, the world would look a lot cleaner, and the entire trial circus that is going on now would be moot. I am as tired of the government case as I am of Microsofts, I do not think a breakup per se will do anyone any good, but I'm glad to see some rather common-sense ideas about fair competition in the software market gain a foothold in law.
In a lot of commercial software companies, there's a strong sentiment not only in managers, but also in development staff, that supporting Linux opens up a Pandora's box. It's not so much compiling for all those platforms, because most developers and managers can be convinced that the API is not very different, so few actual code changes are required. The big problem is in testing. For a glimpse of the problem, look at Netscape's Tinderboxes, and remember that Netscape is one of the most Linux minded commercial developers out there.
For a commercial developer that sees Linux as a platform deserving to be served with tested software, the cost of testing each of their new releases on each distribution quickly becomes prohibitive, and higher up management resolves this issue by drawing a line in the sand. There is a risk that this line winds up including no Linux at all.
Oh well. There is no clear answer here. I'd like to see more distributions supported (as much as I'd like to see commercial software for Linux being tested on *BSD under Linux emulation). I'm just afraid that some companies will get second thoughts about Linux if they get roasted over not supporting less prominent distributions.
All my PC's run BSD/OS. The other computers run NetBSD (or PalmOS).
Just look at the differences between IE on Mac and IE on Windows. I don't believe that much code is shared.
Well, I guess the judge just said that there has been no bullying of vendors, just of the consumers by the pricetag of Windows'98. The judge will have to relent on this one, development cost of IE4 is nothing on Microsofts budget, pennies per copies of Windows'98.
In my view, Microsoft didn't quite win. The marketplace lost.
Same thing happened with the Intel Pentium optimized GCC. It just takes time and a lot of hard work to clean up such a code base. The success of the Apple patches will be judged on the question of how much work it will be to clean it up, and how much time Apple or volunteers will be able to devote to cleaning up the code.
I for one would be happy to see ObjectiveC go forward with Apples backing.
Besides, roaming charges will kill you, and I find the I888 user interface sucks (try reading an 80 character SMS message on it).
Renting a mobile phone locally is the better option for now.
Netscape is doing a tremendous effort at integrating "outside" patches even if they come with no immediate added value to themselves, and if you track netscape.public.mozilla.* and look for the Netscape internal developers comments on that situation, you'll realize just how formidable the task is.
I am extremely relieved to see companies like Netscape and Corel sticking it out, trying to meet the conflicting goals of keeping the shareholders happy (release on time, on target) and the goals of the open source community (release when it's ready, and fix *this* bug now even though it doesn't hamper customer acceptance of the product).
Hail Netscape! Hail Corel!
Anyway, the X server doesn't support APM -- so, the X server has to be out of the way when travelling, and forgetting that will result in a hard crash. I'm not sure whether this is because the ATI Mobility chip is too complex to get X right, or if there is a bug in the chip. Whatever the cause, it's annoying, and Windows works just fine (well, at least in this respect :-)
The cool thing about getting Dell to ship real complex hardware like laptops with Linux, is that if Dell gets too many complaints, they'll chat up ATI to do something about the APM problem, and help the Xfree people to bring hardware support up to par with Windows.
When Digital Equipment sold out on DECWindows to turn it over to OSF, my trust in closed source died. The DECWindows code contained bugs, sure enough, but as part of VMS at least some customers got source free of charge, allowing me to work around those bugs. The OSF bits were not included in the microfiches, so: no source code, no bug fixes.
Letting the openness of the source out of the picture, there's the user experience. While some nice features from (then new) window systems such as Windows and CUI were brought in, the eye candy brought in as part of the Motif effort effectively killed the clarity of the user interface.
My favorite example is the radio button. DECwindows had a clear concept: an unchecked radio box is depicted by and empty circle, a checked radio button is a black circle with a big red dot in it. With Motif's "improved" 3D look, one really had to look hard to see which was selected and which wasn't. "Hmmm, the lower pixels are darker than the top pixels, so this is probablyu the selected radio button".
Seeing how easy uncertified platforms like Linux and BSD compile each others code with less system specific patches than certified platforms like NT, Solaris or HP-UX, one has to wonder what the role for certification is.