MS's relationship with the OEMs is a different case, because MS has an actual monopoly, and would in fact charge higher prices or simply not sell OEM licenses at all to vendors. They did this in order to protect their monopoly, and this is illegal under Sherman. Apple doesn't have a monopoly of cellphone applications, and is in fact in tight competition with competitor's app stores. A customer can always elect to write their app for the Blackberry instead. With Windows the OEMs didn't have a viable alternative.
AND even though some of the specific practices MS engaged in with the OEMs were found to violate Sherman, MS was able to remedy the problem by simply imposing a "points" system where OEMs pay a lower price for OEM installs based on the number of points they earn by prominently displaying a "Works with Windows" logo, not offering other OSs for default install, volume etc. As long as MS didn't use bundling of other OSs as the positive single reason for not offering an OEM a low price, they were allowed to continue the practice.
The upshot of this is if you did finally come to the court and manage to show that Apple had a monopoly and that removing the program was an illegal monopoly action, Apple's most logical remedy would be to create a point system and simply dock points from people who write tethering apps, and tie the points system to the percentage sales cut.
In any case, AT&T finally controls tethering on their network, this is a recognized authority they have, and if Apple found itself unable legally to prevent tethering apps, ATT would just shape the traffic of people that used it, or charge them extra on their cell bill.
From what I understand, sound studio apps really, really don't need specialized hardware -- CPUs are more than fast enough.
Well, as a practical matter, if you want to have 512 independent busses with sample-accurate sync and zero latency, and dozens of independent EQs and reverbs all running in RT, you'll generally need something, like Pro Tools cards with a dozen SHARCs on them, but most people don't do that, aside from the machine this message is being posted from. All that hardware to waste... Must get back to work.
Of course, Pro Tools has a vendor-specific plugin arch where you write code that can be run on different classes of DSP hardware. The platform just loads your code onto the cards and then the CPU can forget about that audio filter... There's definitely room to grow this tech in the open platform space, and this OpenCL business looks like a good way of doing that.
You mean an estoppel was created when they momentarily accepted it to the App store? For that, you'd have to prove that Nullriver had received an expectation or assurance from Apple that it was going to receive revenue. I don't think being offered on the App store for a day creates that expectation, and Apple disclaims any warranty or promise of revenue. I'd cite the developer agreement but I'm not sure if it's not under an NDA...
I thought they meant the content of the phone. The address book, calendars, media, user data... since they mentioned it in the same breath as the remote wipe, I assume that both features address the same domains.
Apple agreed to the app and it was in the appstore for a day, then pulled it without word.
Under the terms of the App Store program, they can offer or retract an App at will. They're under no obligation to provide an app, or even give a logical reason for withdrawing it. No question it's bad business to not give a reason, but there's no legal issue there.
They killed the nullriver app because AT&T told them to behind the scenes. Collusion and abuse of monopoly power.
The App Store isn't a "monopoly" in the US legal sense. If the App store was the only way to buy an app on 90% of the phones in the US, or if Apple had colluded with Google to prevent an App to be shared on either store, then there might be a conversation to be had. In the bad analogy department, you're arguing that Gillette should be sued for a monopoly because it refuses to sell Bic blades.
The App Store's licensing and content control model is basically identical to how console manufacturers control what games are permitted to be run on their consoles.
Which competition is even close to this kind of market?
Microsoft maybe, though their problems with getting a good vertical product are endemic, excepting perhaps the Xbox, but is that yet profitable?
Amazon, but they seem to be happy leaving personal computing, email, contacts, mobile media on the table. But Amazon seems to be much better equipped to roll out a vertical than MS, even if it hasn't demonstrated the complete wherewithal in the hardware department to pull off a media/comm device. Love my Kindle though.
As Gruber pointed out the other day, the main NY Apple Store pulls as much revenue, in that one store annually as the entire Palm corporation earns in 4 months. Palm might have a chance if the hardware is as good as everyone says, but I have terrible memories of support on my Treo. Apple will always have the winning store chain of all of these to provide an on-the-ground support infrastructure.
Sony is weird; they definitely could do the hardware and support but they seem to have serious problems institutionally with putting together a vertical that combines the consumer electronics division, music and creatives, Vaio, and SonyEricsson into a single Great Product.
I think the problem is too many String theorists write too many pop physics books, giving people on the outside the impression that the arguments and theories are often wrong, and further that they are fickle and driven by popularity and name recognition (Brian Greene invented the 11 dimensions, right?)
There's really no reason anybody needs to have practical knowledge of 11 dimensions. At this time. It's nice that theoretical physicists are happy to bring us up to the chalkboard and talk about what they're working on, but I don't think they work hard enough to establish that a lot of String theory is, for the time being, a lot of guesswork and that you shouldn't really hold it up as the vanguard of physical knowledge. The pop physics books try to make this or that String theorist out as the next Einstein, when in fact Einstein was just guessing until someone took a picture of an eclipse. It was a very educated, formal, beautiful guess, in agreement with everything we know about the world, but it was just as wrong as ether until it was proven.
The USB Consortium's peripheral compliance checklist is pretty clear on the matter. If the vendor IS you're using doesn't match the vendor on your application, you're not compliant. I don't know how this affects your ability to get a logo or a license, however.
This is impossible for Apple to block. [...] Nor is it even unsafe, because the code to support older iPods is pretty stable and will not change over time - the older iPods will always be supported.
As Jon points out in TFA, the Pre still identifies itself as a Pre on it's root device node, even when it's in Media Sync mode, so it's trivial to block, it only requires Apple do so.
More broadly, Apple can make any scheme like this very difficult for a lot of people for a very long time, enough to make the feature impractical for casual use, which is the whole principle of DRM anyways. Apple can push firmware updates to the old iPods and make the old owners upgrade before moving on to iTunes 9, or iTunes 8.1.1.2, or whatever, but it's true they can't make people upgrade their iTunes, as long as they just use it as a jukebox and don't need the services, like the Store. But if you're throwing that overboard, why don't you use a different jukebox that actually supports the Pre legit? Like, as Jon says, MediaTwist.
I;m just making the point that there are Scientologists out there that don't belong to the Church of Scientology. The CoS is everything you say it is, but the faith system is a separable entity. It's an admittedly strange one by my standards, but there is in fact a community of people that read Dyanetics and distribute the schematics of E-Meters and audit each other for free, because they believe in it.
Scientology is an elective class of people, on the whole.
The church membership is elective. The faith community is a protected class. It's perfectly legitimate to sanction the institution of a Church -- various diocese of the Catholic Church have been sued into bankruptcy because of their malfeasance -- but not the religious believers themselves. You don't choose your faith.
That must be why there were dozens of threads updating and reading from these massive global arrays without using any kind of synchronization mechanism.
Shows what you know, they were just using classic Mongolian Actorfuck Model:)
You appear to have a very, very, very strange idea of what is involved in "designing" an OS. In fact, what you're arguing is the equivalent of saying installing an unsigned driver in Windows (or a binary driver into Linux) changes it to some fundamentally different OS.
Yes, I guess that is what I think. The code of someone else's OS isn't yours, doesn't belong to you, and cannot be modified without changing its fundamental identity and character. If you do want an OS where this doesn't happen, I don't think you want Mac OS X. I would consider support and an unbroken trust with the developers as intrinsic to an operating system.
Specific hardware drivers are not - or certainly shouldn't be - part of an OS's design (for a general purpose OS like IS X). Interfaces for the drivers to use ? Sure. The drivers themselves ? Typically not.
Well, that's an interesting argument, because some OSs (like Windows) provide a stable ABI for device driver writers, while some (like Linux) don't. In both cases it's intentional and driven by matters external to the technology. Linus could put in a stable driver ABI, but he doesn't want to because then OEMs could ship binary-only drivers.
The _only_ point I have made, is that OS X is not _designed_ to work solely with Apple's hardware lineup,
And my point is that when they put platform validation into the OS, it becomes designed to work solely with Apple's hardware lineup. But even if we accept your more expansive definition of "designed," what effective difference does it make if you have to hack the OS in order to get it to run on off-label hardware? It's not really Mac OS X anymore, is it? I suppose that it would do all of the things Mac OS X does, but the trust that normally would extend to the developers will be broken and only go back to the hacker. The code would then effectively be the hacker's design, or his "interpretation" of Mac OS X, and the support would devolve to him.
It's like Apple's USB plugs on their keyboards. They're USB, but they have a notch that only lets you plug them in to the their extension cord. Plainly the plug is designed to only plug into the extension cord. It's irrelevant what's actually on the wire, as long as you plug everything together like Apple says and "A" appears on the screen when you press "A" on the keyboard. That's all Apple says is supposed to happen, and thus all you should expect. There might be all kinds of interesting historical designs and technologies at work, but it's a closed system, you run a Mac because you specifically want to ignore how things work. For a closed system, design is support, because the design is private and subject to change without notice or consent.
We really understand how to optimize imperative languages well, we don't have the same level of knowledge / experience regarding functional.
On the other hand, I was surprised at how many of the languages in the "optimal" corner were compiled Schemes, even if, in practice, Scheme is so pure that it's hard to use for real projects...
Well yeah, that's the point. It's sarcasm smitty, you've uh missed my point. By a mile. Or something:\ You've picked a definition of "design" which leads inescapably to the conclusion the Mac OS X is designed to run on any Intel platform. So I picked a definition of "design" that led inescapably to the fact that Linux should be distributed as a binary only. This is a bad argument because there's a lot more to the Linux kernel than its build system, and there's a lot more to Mac OS X than merely what CPU it compiles to.
You can say that Mac OS X is designed for generic Intel system $X by dint of it successfully compiling to x86 assembly, but I could argue it isn't by dint of its lack of available kexts for the motherboard accoutrements system $X, or the OS's contingent operation on the "Dont_Steal_Mac_OS_X.kext" device. Clearly the inclusion of the required "Don't Pirate" device indicates the intention of the writers to forbid operaton on off-label hardware. Maybe you could say this "forbidding" is essentially different from "design," and maybe that's your point. But it's a dumb point, because by that standard Windows NT was "designed" to run on a Mac 9600, because NT can be compiled to Intel and PowerPC, and thus MS must have been placing some "non-designed" encumberance on those PPC 9600 users that wanted to run NT. I think this use of the word "design" makes it almost meaningless, since it's pretty clear NT wasn't designed (by any reasonable understanding of the word) to run on a 9600.
It's perfectly legit to argue over why Apple restricts the operation of Mac OS X. I don't deny the Mac OS X kernel, and a lot of the userland services, are designed to run on any operating system, and they're open source and you can run xnu and Apple's build of Apache tomorrow for free, on any system you please. But these aren't Mac OS X.
I think you're working too hard to build your argument around the foregone conclusion...
Really, I think I should distribute a closed-source Linux distro.
The Linux kernel is written in C, which is a language designed to be obfuscated in a binary format. As evidenced by the Linux kernel's ability to be built into a binary, it was clearly designed to be distributed in a locked-down format. The license on the tarball I downloaded has little bearing to the actual build and distribution system. In this day and age you'd need to be an awesomely incompetent developer to give away all your changes to the kernel for free.
Not exactly the same thing, but I think you have to respect the vendor as being the person who says what something is designed to do, and their rights to control their works within the bounds of the law. Don't like it? Change da law.
A small point, but musicians don't use sound cards anymore... everyone uses interfaces that talk over USB or firewire. You still need PCIe for Pro Tools HD but that's $10k worth of hardware, minimum, and clearly then you're not going to be buying a midrange tower to put all that stuff in... You might be able to make a case for MADI bridges but nobody uses that stuff...
Note, if you really want a Mac Pro with slots and last years tech specs and no support you can spend between $800-$1000 today.
An OS specifically designed for even a particular version of an AMD or Intel CPU, would quite feasibly not run on any of the others (eg: take advantage of features specific to Core 2). Similarly in terms of specific chipset, NIC, or other hardware implementations.
Strictly speaking, Mac OS X is "designed" (in the intel case) to run on an Intel CPU with EFI and a specified closed set of hardware... Windows isn't "designed" to support a set of hardware as much as it's "designed" to support OEMs writing their own drivers, which has the effect of Microsoft claiming that it "supports" a wide range of hardware, when really all they do is give vendors a platform and the "support" is finally the vendor's problem. If your Mac won't boot it's Apple's problem, period. If your Dell won't boot it might be Dell's, but you'll have to wait and see what their diagnostic tool says, and if they decide it's Windows, you can take your problem to Microsoft. I suppose you can say that Mac OS X is "designed" for vendors writing their own hardware drivers, because it does allow this, but the OS itself relies on no outside vendors to run the complete package in a clean configuration. From the perspective of an end user that wants a running computer, the distinction is without effect. It's not designed to run on anything but what Apple sells, because the people who wrote the thing were only writing for a closed an predictable set of platforms.
It's not like builds of Mac OS X run on broad swathes of commodity hardware, and then at some critical phase Apple gelds it and makes it check the TPM (they don't). Or maybe I'm conflating support with design, because I consider how you resolve problems to be part of the design of the OS. Huh. Whatever... In any case, I think it's safe to say that design relates intention, and Mac OS X is only intended to run on Apple-branded hardware. So here we are.
You've missed my point. By a mile.
I thought your point was that Apple didn't have the technical wherewithal to accomplish the feat of destroying their hardware business, and I'm just saying it's within their power if they wanted to but it really wouldn't benefit them.
Slashdot Mirror
MS's relationship with the OEMs is a different case, because MS has an actual monopoly, and would in fact charge higher prices or simply not sell OEM licenses at all to vendors. They did this in order to protect their monopoly, and this is illegal under Sherman. Apple doesn't have a monopoly of cellphone applications, and is in fact in tight competition with competitor's app stores. A customer can always elect to write their app for the Blackberry instead. With Windows the OEMs didn't have a viable alternative.
AND even though some of the specific practices MS engaged in with the OEMs were found to violate Sherman, MS was able to remedy the problem by simply imposing a "points" system where OEMs pay a lower price for OEM installs based on the number of points they earn by prominently displaying a "Works with Windows" logo, not offering other OSs for default install, volume etc. As long as MS didn't use bundling of other OSs as the positive single reason for not offering an OEM a low price, they were allowed to continue the practice.
The upshot of this is if you did finally come to the court and manage to show that Apple had a monopoly and that removing the program was an illegal monopoly action, Apple's most logical remedy would be to create a point system and simply dock points from people who write tethering apps, and tie the points system to the percentage sales cut.
In any case, AT&T finally controls tethering on their network, this is a recognized authority they have, and if Apple found itself unable legally to prevent tethering apps, ATT would just shape the traffic of people that used it, or charge them extra on their cell bill.
From what I understand, sound studio apps really, really don't need specialized hardware -- CPUs are more than fast enough.
Well, as a practical matter, if you want to have 512 independent busses with sample-accurate sync and zero latency, and dozens of independent EQs and reverbs all running in RT, you'll generally need something, like Pro Tools cards with a dozen SHARCs on them, but most people don't do that, aside from the machine this message is being posted from. All that hardware to waste... Must get back to work.
Of course, Pro Tools has a vendor-specific plugin arch where you write code that can be run on different classes of DSP hardware. The platform just loads your code onto the cards and then the CPU can forget about that audio filter... There's definitely room to grow this tech in the open platform space, and this OpenCL business looks like a good way of doing that.
You mean an estoppel was created when they momentarily accepted it to the App store? For that, you'd have to prove that Nullriver had received an expectation or assurance from Apple that it was going to receive revenue. I don't think being offered on the App store for a day creates that expectation, and Apple disclaims any warranty or promise of revenue. I'd cite the developer agreement but I'm not sure if it's not under an NDA...
I thought they meant the content of the phone. The address book, calendars, media, user data... since they mentioned it in the same breath as the remote wipe, I assume that both features address the same domains.
Apple agreed to the app and it was in the appstore for a day, then pulled it without word.
Under the terms of the App Store program, they can offer or retract an App at will. They're under no obligation to provide an app, or even give a logical reason for withdrawing it. No question it's bad business to not give a reason, but there's no legal issue there.
They killed the nullriver app because AT&T told them to behind the scenes. Collusion and abuse of monopoly power.
The App Store isn't a "monopoly" in the US legal sense. If the App store was the only way to buy an app on 90% of the phones in the US, or if Apple had colluded with Google to prevent an App to be shared on either store, then there might be a conversation to be had. In the bad analogy department, you're arguing that Gillette should be sued for a monopoly because it refuses to sell Bic blades.
The App Store's licensing and content control model is basically identical to how console manufacturers control what games are permitted to be run on their consoles.
What would be their allegations in such a suit? What agreement has Apple broken? What law has apple violated?
I'm gonna buy one and put it next to my Apple II GS. :)
If YDK what IDK means, than IDK what to do with you and you're probably SOL.
Which competition is even close to this kind of market?
Microsoft maybe, though their problems with getting a good vertical product are endemic, excepting perhaps the Xbox, but is that yet profitable?
Amazon, but they seem to be happy leaving personal computing, email, contacts, mobile media on the table. But Amazon seems to be much better equipped to roll out a vertical than MS, even if it hasn't demonstrated the complete wherewithal in the hardware department to pull off a media/comm device. Love my Kindle though.
As Gruber pointed out the other day, the main NY Apple Store pulls as much revenue, in that one store annually as the entire Palm corporation earns in 4 months. Palm might have a chance if the hardware is as good as everyone says, but I have terrible memories of support on my Treo. Apple will always have the winning store chain of all of these to provide an on-the-ground support infrastructure.
Sony is weird; they definitely could do the hardware and support but they seem to have serious problems institutionally with putting together a vertical that combines the consumer electronics division, music and creatives, Vaio, and SonyEricsson into a single Great Product.
Anyone else come to mind?
The precedent would have been sufficient to cripple any number of open source developers...
I think the problem is too many String theorists write too many pop physics books, giving people on the outside the impression that the arguments and theories are often wrong, and further that they are fickle and driven by popularity and name recognition (Brian Greene invented the 11 dimensions, right?)
There's really no reason anybody needs to have practical knowledge of 11 dimensions. At this time. It's nice that theoretical physicists are happy to bring us up to the chalkboard and talk about what they're working on, but I don't think they work hard enough to establish that a lot of String theory is, for the time being, a lot of guesswork and that you shouldn't really hold it up as the vanguard of physical knowledge. The pop physics books try to make this or that String theorist out as the next Einstein, when in fact Einstein was just guessing until someone took a picture of an eclipse. It was a very educated, formal, beautiful guess, in agreement with everything we know about the world, but it was just as wrong as ether until it was proven.
But since this isn't a USB device
Explain? From Palm's website:
The USB Consortium's peripheral compliance checklist is pretty clear on the matter. If the vendor IS you're using doesn't match the vendor on your application, you're not compliant. I don't know how this affects your ability to get a logo or a license, however.
This is impossible for Apple to block. [...] Nor is it even unsafe, because the code to support older iPods is pretty stable and will not change over time - the older iPods will always be supported.
As Jon points out in TFA, the Pre still identifies itself as a Pre on it's root device node, even when it's in Media Sync mode, so it's trivial to block, it only requires Apple do so.
More broadly, Apple can make any scheme like this very difficult for a lot of people for a very long time, enough to make the feature impractical for casual use, which is the whole principle of DRM anyways. Apple can push firmware updates to the old iPods and make the old owners upgrade before moving on to iTunes 9, or iTunes 8.1.1.2, or whatever, but it's true they can't make people upgrade their iTunes, as long as they just use it as a jukebox and don't need the services, like the Store. But if you're throwing that overboard, why don't you use a different jukebox that actually supports the Pre legit? Like, as Jon says, MediaTwist.
I;m just making the point that there are Scientologists out there that don't belong to the Church of Scientology. The CoS is everything you say it is, but the faith system is a separable entity. It's an admittedly strange one by my standards, but there is in fact a community of people that read Dyanetics and distribute the schematics of E-Meters and audit each other for free, because they believe in it.
Scientology is an elective class of people, on the whole.
The church membership is elective. The faith community is a protected class. It's perfectly legitimate to sanction the institution of a Church -- various diocese of the Catholic Church have been sued into bankruptcy because of their malfeasance -- but not the religious believers themselves. You don't choose your faith.
That must be why there were dozens of threads updating and reading from these massive global arrays without using any kind of synchronization mechanism.
Shows what you know, they were just using classic Mongolian Actorfuck Model :)
You appear to have a very, very, very strange idea of what is involved in "designing" an OS. In fact, what you're arguing is the equivalent of saying installing an unsigned driver in Windows (or a binary driver into Linux) changes it to some fundamentally different OS.
Yes, I guess that is what I think. The code of someone else's OS isn't yours, doesn't belong to you, and cannot be modified without changing its fundamental identity and character. If you do want an OS where this doesn't happen, I don't think you want Mac OS X. I would consider support and an unbroken trust with the developers as intrinsic to an operating system.
Specific hardware drivers are not - or certainly shouldn't be - part of an OS's design (for a general purpose OS like IS X). Interfaces for the drivers to use ? Sure. The drivers themselves ? Typically not.
Well, that's an interesting argument, because some OSs (like Windows) provide a stable ABI for device driver writers, while some (like Linux) don't. In both cases it's intentional and driven by matters external to the technology. Linus could put in a stable driver ABI, but he doesn't want to because then OEMs could ship binary-only drivers.
The _only_ point I have made, is that OS X is not _designed_ to work solely with Apple's hardware lineup,
And my point is that when they put platform validation into the OS, it becomes designed to work solely with Apple's hardware lineup. But even if we accept your more expansive definition of "designed," what effective difference does it make if you have to hack the OS in order to get it to run on off-label hardware? It's not really Mac OS X anymore, is it? I suppose that it would do all of the things Mac OS X does, but the trust that normally would extend to the developers will be broken and only go back to the hacker. The code would then effectively be the hacker's design, or his "interpretation" of Mac OS X, and the support would devolve to him.
It's like Apple's USB plugs on their keyboards. They're USB, but they have a notch that only lets you plug them in to the their extension cord. Plainly the plug is designed to only plug into the extension cord. It's irrelevant what's actually on the wire, as long as you plug everything together like Apple says and "A" appears on the screen when you press "A" on the keyboard. That's all Apple says is supposed to happen, and thus all you should expect. There might be all kinds of interesting historical designs and technologies at work, but it's a closed system, you run a Mac because you specifically want to ignore how things work. For a closed system, design is support, because the design is private and subject to change without notice or consent.
We really understand how to optimize imperative languages well, we don't have the same level of knowledge / experience regarding functional.
On the other hand, I was surprised at how many of the languages in the "optimal" corner were compiled Schemes, even if, in practice, Scheme is so pure that it's hard to use for real projects...
Well yeah, that's the point. It's sarcasm smitty, you've uh missed my point. By a mile. Or something :\ You've picked a definition of "design" which leads inescapably to the conclusion the Mac OS X is designed to run on any Intel platform. So I picked a definition of "design" that led inescapably to the fact that Linux should be distributed as a binary only. This is a bad argument because there's a lot more to the Linux kernel than its build system, and there's a lot more to Mac OS X than merely what CPU it compiles to.
You can say that Mac OS X is designed for generic Intel system $X by dint of it successfully compiling to x86 assembly, but I could argue it isn't by dint of its lack of available kexts for the motherboard accoutrements system $X, or the OS's contingent operation on the "Dont_Steal_Mac_OS_X.kext" device. Clearly the inclusion of the required "Don't Pirate" device indicates the intention of the writers to forbid operaton on off-label hardware. Maybe you could say this "forbidding" is essentially different from "design," and maybe that's your point. But it's a dumb point, because by that standard Windows NT was "designed" to run on a Mac 9600, because NT can be compiled to Intel and PowerPC, and thus MS must have been placing some "non-designed" encumberance on those PPC 9600 users that wanted to run NT. I think this use of the word "design" makes it almost meaningless, since it's pretty clear NT wasn't designed (by any reasonable understanding of the word) to run on a 9600.
It's perfectly legit to argue over why Apple restricts the operation of Mac OS X. I don't deny the Mac OS X kernel, and a lot of the userland services, are designed to run on any operating system, and they're open source and you can run xnu and Apple's build of Apache tomorrow for free, on any system you please. But these aren't Mac OS X.
I think you're working too hard to build your argument around the foregone conclusion...
Really, I think I should distribute a closed-source Linux distro. The Linux kernel is written in C, which is a language designed to be obfuscated in a binary format. As evidenced by the Linux kernel's ability to be built into a binary, it was clearly designed to be distributed in a locked-down format. The license on the tarball I downloaded has little bearing to the actual build and distribution system. In this day and age you'd need to be an awesomely incompetent developer to give away all your changes to the kernel for free.
Not exactly the same thing, but I think you have to respect the vendor as being the person who says what something is designed to do, and their rights to control their works within the bounds of the law. Don't like it? Change da law.
Sound cards for musicians,
A small point, but musicians don't use sound cards anymore... everyone uses interfaces that talk over USB or firewire. You still need PCIe for Pro Tools HD but that's $10k worth of hardware, minimum, and clearly then you're not going to be buying a midrange tower to put all that stuff in... You might be able to make a case for MADI bridges but nobody uses that stuff...
Note, if you really want a Mac Pro with slots and last years tech specs and no support you can spend between $800-$1000 today.
An OS specifically designed for even a particular version of an AMD or Intel CPU, would quite feasibly not run on any of the others (eg: take advantage of features specific to Core 2). Similarly in terms of specific chipset, NIC, or other hardware implementations.
Strictly speaking, Mac OS X is "designed" (in the intel case) to run on an Intel CPU with EFI and a specified closed set of hardware... Windows isn't "designed" to support a set of hardware as much as it's "designed" to support OEMs writing their own drivers, which has the effect of Microsoft claiming that it "supports" a wide range of hardware, when really all they do is give vendors a platform and the "support" is finally the vendor's problem. If your Mac won't boot it's Apple's problem, period. If your Dell won't boot it might be Dell's, but you'll have to wait and see what their diagnostic tool says, and if they decide it's Windows, you can take your problem to Microsoft. I suppose you can say that Mac OS X is "designed" for vendors writing their own hardware drivers, because it does allow this, but the OS itself relies on no outside vendors to run the complete package in a clean configuration. From the perspective of an end user that wants a running computer, the distinction is without effect. It's not designed to run on anything but what Apple sells, because the people who wrote the thing were only writing for a closed an predictable set of platforms.
It's not like builds of Mac OS X run on broad swathes of commodity hardware, and then at some critical phase Apple gelds it and makes it check the TPM (they don't). Or maybe I'm conflating support with design, because I consider how you resolve problems to be part of the design of the OS. Huh. Whatever... In any case, I think it's safe to say that design relates intention, and Mac OS X is only intended to run on Apple-branded hardware. So here we are.
You've missed my point. By a mile.
I thought your point was that Apple didn't have the technical wherewithal to accomplish the feat of destroying their hardware business, and I'm just saying it's within their power if they wanted to but it really wouldn't benefit them.
This is beginning to get out of hand.