Not even a week has elapsed since what was a rumor picked up as a scoop by News.com became confirmed truth by the power of Steve Jobs' renowned Reality Distortion Field. Already, though the Internet is distorting this move even further with unsupportable theories, unreasonable expectations, and unfounded predictions, most based either on simply failing to actually read what Jobs and Apple VP Phil Schiller said at this year's developer conference, or deep-seated misunderstandings about what kind of company Apple is, has always been, and, despite speculation to the contrary, is likely to be even after the change in processor architectures takes place.

I've been collecting some of these myths and predictions and put them into a list, to pop them one by one like balloons at a child's birthday party, because I'm the kind of guy who likes to pop balloons at children's birthday parties.

Myth #1: MacIntel Is Good/Bad For Mac Gaming

There's a line that says that because Mac OS X will now run on the same processor series as Windows, that porting applications with significant low-level optimizations, which is to say, games, will now be faster and easier and a flood of Windows games ports will arrive on the platform.

This ignores the fact that most games for Windows use DirectX, not OpenGL, and this is the real barrier. Unless someone makes DirectX run on Mac OS X, regardless of CPU, the amount of work saved by eliminating byte swapping issues will probably be more than offset by the short-term need to maintain so-called "fat binaries" that support Mac machines running either PPC or Intel hardware, something that mass market developers and porting houses will likely have to do for several years to come.

Developers themselves are hardly of one mind as to whether or not the move to Intel is good or bad for Mac ports of Windows games, or gaming on MacOS, or gamers who own Macs. The reactions range from "this is great news" through to "this is the death of the platform"; a good place to gauge the reactions is Tuncer Deniz's piece at IMG Magazine, Game Developers React To Intel Switch.

Meanwhile, at MacWorld, Peter Cohen is trying to be a calming influence, noting that while this means actually more work for developers and porting houses in the short-term, he doesn't know of any significant developers are going to throw in the towel over this. (It's interesting to note then that one of the developers in the IMG article who reacts most negatively is Brian Greenstone at Pangea Software, who as some may remember fought a bitter but losing battle with the forces at Apple who deemphasized what he felt was the superior QuickDraw 3D technology in favor of OpenGL. However, I'm not exactly sure how much the Mac gaming market suffers if the Bugdom and Nanosaur franchises head off into the sunset.)

Myth #2: Apple Is A Software Company

This misconception leads to a lot of unrealistic expectations and other predictions which we'll deal with separately. However, there is ample evidence that Apple is, in fact, a hardware company. This evidence is not hard to collect.

Apple sells iPods, which are hardware MP3 players. Their software MP3 player, which runs on Mac OS X and Windows and helps users manage their iPods if they have them, is free.

Apple sells personal computers which generally cost more than the cheapest possible Wintel competitors with similar basic specifications, often making up the difference in price with superior software, superior design, and extra functionality. The operating system that Apple makes for these personal computers, Mac OS X, is also sold, but unlike with Windows there is no copy protection on the discs, no serial numbers to be entered.

The reason for this is simple: Mac OS X only runs on Apple computers. They no longer license their designs to other manufacturers, as they once did for a short time and throughout their history have been urged to do, in order to reduce consumer prices and supposedly expand the user base for their OS. If you have any use for Mac OS X it is because you own an Apple computer and thereby, by definition, are an Apple customer. Apple already has some of your money. Sure, strictly speaking Apple would like every Mac owner to pay for every significant OS release rather than copying or downloading it, but they have obviously decided that it is not worth the trouble and effort to do so. If Mac OS X ran on hardware made by another company, it is likely they would do so, regardless of whether or not that manufacturer paid Apple a license fee for it. Apple's profit margin on its computers is much more than what any "clone" company would reasonably pay, and much more than the margin, if any, on sales of Mac OS X as a retail sales item.

Apple makes software that it either sells at a loss or gives away for free in order to entice users to buy Apple hardware, either music players or personal computers and accessories. This applies equally to software that runs on Apple computers or software that runs on Windows (QuickTime and iTunes). It subsidizes its software development with high margins on hardware. It must do this because the share of the market it has is smaller than the competing OS, which is Windows.

Myth #3: Apple Is Becoming A Software Company

Even those who understand that Apple has historically made money on hardware subsidized by free or cheap software think the company is reversing this, getting out of the hardware business, and making Mac OS run on commodity hardware.

In reality this is the worst combination of faulty logic and wishful thinking. People who say this like Mac OS X, and either want to see it gain market share against Windows because they like it better, or want to see it run on cheaper hardware because they can't justify the expense of an Apple computer.

However, it is difficult to see how this practice represents a sustainable business model for Apple in the short to medium term. It is not possible to interpret Apple's announcement that it will begin shipping PCs based on Intel processors in 2006 and have a completely Intel-based lineup by the end of 2007 as a shift away from hardware and towards a software-only business model without ignoring the simple fact that the announcement was for Intel-based PCs, not an operating system that runs on Intel processors.

That OS X runs, and has run for several years, on Intel processors is ample evidence of Apple's excellent management and skepticism towards IBM and Freescale's ability to fulfill its requirements. However, it should not come as a surprise to anyone, given OS X's NextStep roots; nor should it indicate anything more than what it is: a processor architecture switch.

If Apple were announcing that it was becoming a software, rather than a hardware company, they would not have bothered announcing this as a "switch". They'd have announced that OS X now runs on PPC as well as Intel x86, made their PPC designs available for licensing, and added DRM to their OS distributions (otherwise OS X for Intel processors basically end up becoming de facto free-as-in-beer software, leaving the company with no revenue stream at all).

To continue to spend money on designing new computers to work on Intel processors would be a foolish and unnecessary investment. That the company is announcing a "switch" in architectures at all is a clear signal that they intend to continue being a hardware company, making their own proprietary designs, which they intend to be the sole platform on which Mac OS X will run. The difference is that the CPU will be made by Intel rather than by IBM; neither more nor less.

Myth #4: Apple Should Become A Software Company

While I think it's easily demonstrable that Apple has no intention of doing so, it is worth explaining the reasons why they've probably decided not to. This expectation again comes from those who say they would like to run Mac OS X, but can't afford Macs, as well as those who consider Mac OS X superior and would like to see it more prevalent than it is, and see the high cost of Apple hardware as an unfair and unnecessary barrier to the operating system's expansion.

These arguments inevitably draw comparisons to Linux and other free Unix variants, like FreeBSD and OpenBSD, that run on various hardware platforms and are generally available either for free download or at a minimal cost. They allow the user to run on hardware that is (more or less) of their choice, depending on the specific distribution and drivers available.

However, there is a great deal more risk in a business model based on this idea than in Apple's current business model. Here's why:

Making Apple hardware costs Apple money. They have to design them, and they have to manufacture them. A certain portion of the cost is fixed, meaning it is the same regardless of how many Macs they either build or don't build, sell or don't sell: the design, etc. At a certain point, they have to pay for parts and assembly; they have to predict how many they will sell, how many parts they need, and spend as little as possible to meet the demand. Once they stop selling a given computer, it's possible to calculate how much that model actually cost them over its life, and how much revenue it generated. As long as the latter is more than the former, everything is OK; and since the fixed costs, the costs that have to be split over all the units ever sold, is only a portion of the unit costs, it should be possible to control their profit margin if they can predict sales volumes with a certain amount of precision.

Software development costs aren't like that at all. Aside from packaging, distribution and post-sales support, the initial fixed costs of development are, essentially, the whole enchilada. Mac OS X 10.4 "Tiger" costs the same to develop whether one client buys it or ten million; the percentage of the costs related to a sale of the software that is "fixed" and therefore cannot be reduced when sales volumes decrease is much higher than for hardware.

So let's look and see what happens if we apply this dynamic to the business model for Apple proposed by those who want to see commodity PC hardware run Mac OS X. Currently, Apple sells Mac OS X for around the same amount as other professional software: in other words, hundreds of dollars, rather than thousands. Even if one argues that expanding to support other hardware platforms increases costs minimally, it is not arguable that Apple's average margin per unit sold would decrease drastically if they eliminate their hardware products.

Should Apple want to make Mac OS X an architecture-blind operating system that runs on PCs from many different manufacturers and give up its hardware margins brings to mind the joke of the businesses that lose money on every sale but make up the difference in volume.

Microsoft can do this because they have such a large user base; their margins on each copy of Windows, either from retail sale or OEM licensing, may be much slimmer than Apple's, but their unit sales are much higher.

Even with Apple's hardware costs all eliminated, it would likely require an extremely dramatic increase in the user base for Mac OS X to make up the lost revenue. This is likely a risk that Apple is unwilling to take. Supporting a wider variety of hardware inevitably means higher support costs as well as a degraded user experience for those with hardware Mac OS X doesn't initially work with. Since each individual Apple customer now represents much less gross profit for Apple, the quality of customer support would likely fall, in favor of spending money on marketing to expand the reach of the platform. Combating piracy would become a real problem for Apple, as users who never owned an Apple computer but run Mac OS X on their own PC hardware now represent real lost revenue.

It would be a disaster. Somewhere in Apple I'm sure somebody has calculated what the value of X is, being the percentage of the operating system market share that Mac OS X would have to achieve at market price point Y to deliver Z, being the amount of money that Apple makes in any given quarter in the past few years using their current business strategy. Whomever has calculated X has either concluded that it is unachievable or that rising costs would make it undesirable to achieve.

Myth #5: Mac OS X On Intel Will Help/Hinder Linux

There's a line of reasoning that says that what keeps open source developers away from tinkering with Mac OS X's open source foundation, Darwin, is that it only runs on PPC processors.

This is only part of it. Darwin, in fact, will run on Intel but only on a particular chipset. Given that Apple most likely intends to continue being a hardware company, it is reasonable to expect that Darwin and Mac OS X will only run on Intel CPUs inside Apple computers, with chipsets built to Apple's proprietary specifications. The only advantage to developers with regard to Darwin will be not having to worry about byte order (little Endian vs. big Endian) issues when coding. And since both OS X and Darwin already benefit from a huge library of *nix software that run on it with a simple recompile, like Apache, Sendmail, Postfix, MySQL, and countless others, and there are several distributions of Linux that also run on PPC hardware, it seems that this was never really a serious obstacle anyway.

Owners of Apple hardware that, for whatever reason, don't want to run Mac OS X on their computers will now be running some flavor of Linux or Unix targeted at x86 CPUs, instead of the flavors of PPC Linux, which will most likely fall into disuse except on older Apple hardware. Or perhaps they'll try and run Windows, something which Apple VP Phil Schiller said he expects users to do and that Apple would not prevent them from doing. (That's an important point, it comes up again later. Many prognosticators, it seems, have ignored this point entirely.)

PC Magazine pundit John C. Dvorak thinks that Mac OS X running on Intel is bad for Linux. The primary assumption of this, of course, is that OS X will run on commodity hardware. If that were so, he might be right. Even though Linux is free and runs on just about anything and makes a great server, it still lags behind Apple's far more expensive combination of proprietary hardware and software when it comes to desktop use, primarily because of Apple's massive investment in the interface over the years. If suddenly you could, for free or just the price of a software distribution, get the Apple experience instead of KDE or Gnome on your desktop, then I'd agree that Linux's days on the desktop would be numbered.

But as I've already outlined, this isn't going to happen; because the only thing that OS X running on cheap hardware would kill faster than Linux would be Apple itself. Linux is great as a server, OS X is great on the desktop. OS X running on Intel processors changes neither of these things, so I don't think it's going to have any appreciable effect at all on the fate of Linux.

Myth #6: Mac OS X Will Run On Parts-Bin PCs

This myth underpins several others, and because I've already dealt with it in detail I won't belabor the point too much here, but I'm including it for completeness.

Without fat hardware margins, Apple can't survive to invest what it does in developing OS X, and it most likely cannot reach a high enough market penetration fast enough to make up the difference on low-margin software. Designing and manufacturing custom hardware is cheaper than supporting cheap commodity hardware. Apple knows this. What happens to the first hacker who gets OS X to run on non-Apple hardware will make what DVD Jon went through look like tea with the Queen Mum. The very notion of this threatens Apple's very existence.

If that isn't enough to convince you, take a look at Schiller's other quote, which is "We will not allow running Mac OS X on anything other than an Apple Mac."

Myth #7: Apple Will Stop You From Running Windows On MacIntel Computers

The other edge of the "this means OS X will run on cheap PCs" blade is the "Apple will try and stop you from running Windows on Macs" angle, which says that hackers will figure out how to make Windows run on Apple's Intel-based PCs, and that this will hurt Apple and reduce Mac OS X's market share to the company's detriment.

Hogwash.

For one, no less than Apple VP Phil Schiller said during the developer conference that Apple would not stop users from attempting to run Windows on Macs using Intel processors. That's certainly no pledge of support, nor a promise that it would work unmodified.

Apple uses its OS to sell hardware. Once they've sold it, I doubt they really care what you run on it, at least in financial terms. They polish OS X to bring more users to the hardware platform, not to keep you using it after you've bought the thing. That may sound harsh, but that's the economic reality.

In fact, if iTunes for Windows and the iPod was the leading edge of the so-called "halo effect" designed to bring PC users to the Macintosh platform, then a MacIntel PC that could dual boot Windows and Mac OS X is the fat part of that wedge. Given that Apple is a hardware company, the real enemies are HP and Dell, not Microsoft. Microsoft is and has been for some time an ally; without Microsoft Office, Apple cannot survive.

There is one very gutsy move I think Apple probably won't make, although they probably could: bundle Windows with every MacIntel computer, either as a dual-boot or a virtual machine solution. What reason would a premium computer purchaser possibly have to rule out a Mac under those circumstances? He'd get chance to try out a virus-free, malware-free OS with a great interface and some killer software, with no risk-- if he doesn't like it, he can just boot into Windows. Apple gets its money either way, and the price of an OEM license probably wouldn't make a ripple in the price of a Mac. Microsoft wouldn't care now-- they can legitimately claim to be holding on to their OS hegemony, as all Macs would come packaged with and be able run Windows. They'd actually be increasing their market share. Mac owners would be paying them what amounts to tribute; I'm sure Bill Gates would love that. Apple might bristle but wouldn't really care, because Windows compatibility would be yet another weapon in the fight against cheap PC hardware, eliminating once and for all the "my software won't run" complaint that keeps the hesitant from switching over. It would be a killer inclusion for Mac users who are also hardcore gamers, who have to get their fix by having a second machine or a game console. Just make sure there's a driver for the newest video card from Nvidia and ATI, as well as support for a good surround sound card (something sorely lacking in Macs now for years) and the MacIntel could be a killer gaming rig. Want to play Half-Life? Don't curse the dead Mac port, just load Windows. When you're ready to really get some work done, go back to OS X.

It's a win-win situation for Apple and Microsoft. The losers are the companies that make PCs with uninspired designs, who can only compete on price and CPU speed, and lag behind in adding new technologies like WiFi, USB, Firewire, DVD-ROM drives, CD-ROM drives, and 3.5" floppies-- all things that Apple added to significant portions of its product lines before they became industry standards.

The danger to Apple here, of course, is what happened to OS/2, that had a Windows compatibility mode that worked so well that eventually developers decided that no native OS/2 development was worth it, since everbody could run Windows apps anyway. However, there is a potential upside here, too. If the safety net of Windows compatibility increases the Mac's market share, and those new users prefer to run MacOS rather than Windows, it means a potential larger audience for native Mac applications. Obviously this works out better for Apple if the machines are simply dual-boot capable, rather than a Virtual Machine solution, which would be more threatening.

Myth #8: Moving To Intel Will Kill Apple

This is, again, based mostly on the presupposition that if people can run Mac OS X on whatever they like, they won't buy expensive Apple PCs and the company will die. This is actually correct, which is why Apple won't allow that.

Of course, even those who recognize that Apple does not intend this simply believe that the hacker community will make OS X run on commodity hardware, because hackers can do anything they put their minds to.

This seems to ignore the fact that, to date, there's no evidence of anybody making Darwin run on more than the single PC chipset Apple designed it for; there's no Mac emulator that runs acceptably on Windows, even though several Windows emulators run on the Mac, and several versions of Linux run on Mac hardware even though Mac OS X has not apparently ever run on an Intel-based PC of any design outside of Apple itself. The barriers to this are far more than just the processor; in fact, the fact that individual software titles run on more than one architecture indicate that the CPU alone is not a significant obstacle. Apple will build in enough obstacles that even if running Mac OS X on non-Apple hardware is impossible, it will be difficult enough and have enough drawbacks that it will not be preferable to a real Mac and therefore threaten the product in the marketplace.

The PowerPC chip itself is not some esoteric piece of hardware that can't be bought at any price. The reason why there aren't clones anymore is becuase besides the processor there are things you need to know to build a Mac that only Apple knows. For years this was included in the custom ROMs that even today are needed to successfully emulate 680x0 machines like old Macs and Amigas. These days, there are other restrictions. Since this is the only thing that makes sure that Apple gets the benefit from the money they spend on developing OS X by making sure that only Mac owners can use OS X, there will be little the company won't be willing to do to protect it. Letting Mac OS X run on anybody else's hardware would be giving away the store.

Myth #9: MacIntel Is Bad For Macintosh Development And Windows Ports

If you accept that Myth #6 is not true, because Apple VP Phil Schiller said so, then you might be liable to think that MacIntel spells death to Macintosh software development; that if software companies know that Mac owners can run Windows if they want to, they'll just keep making Windows apps and tell their users who own Macs to boot into Windows in order to run them.

Certainly I think some companies will do this; however, while some will certainly view this as basically the elimination of Mac OS X as an operating system, others will, I think, rightly acknowledge that just because Mac owners can boot into Windows doesn't mean they'll want their software to require it. And if the ability to run Windows and Windows applications-- at full speed, without emulation-- means more Mac owners, then the potential market for Mac applications and thus the incentive to port actually goes up.

There's simply too much functionality that is lost when you require a user to boot into an alternate OS; too many of the cool integrated features that Apple is known for. This, of course, might be obviated by running Windows in a Virtual Machine; but while this might be faster than emulation it is probably still not going to be ideal. New users exposed to the Mac way of doing things may just like it better and prefer Mac versions of applications when available. A new Mac owner coming from Windows may very well appreciate that buying a new computer doesn't force him into buying a new copy of Microsoft Office or Photoshop, for instance, since he can still run his old one-- but when it comes time to upgrade, which do you think he'll prefer?

If MacIntel has performance and features as good or better than Apple's previous PPC offerings, and if the ability to run Windows in some shape or fashion that is better, performance-wise, than solutions like VirtualPC today, it seems likely to me that Apple's market share will start to increase a bit, and with it, sales of Mac software as users upgrade to new versions. More Mac software sales will mean more Mac software. In the long run, this will probably be good for Macintosh development-- just not for the mostly minor technical reasons, like byte order swapping, that are being cited right now.

Myth #10: Apple Moving To Intel Hurts IBM And Motorola

The PowerPC architecture was, is, and will probably always be, designed primarily for embedded devices. IBM's other chip lines, like Power4, are for workstations. Neither are particularly suited for use in consumer PCs.

The inability or unwillingness of the other AIM partners to cater to Apple's needs (multicore processors, mobile processors) simply underscores that as a single manufacturer, Apple's volumes do not justify the discount they demand. Only the most rabid Macintosh zealots are going to consider it more than a minor public relations issue that Apple chose to go with another vendor. Intel, which sells far more CPUs designed for just what Apple wants to build-- personal computers-- can likely afford to give Apple a good price, will have to diverge from its own design trends less than IBM and Motorola had to in order to please Apple, and will benefit from the public relations boost. In short, Intel can afford to keep Apple as a prestige client without bending over backwards. IBM and Motorola couldn't any longer, and Apple had no more tricks to pull out of its sleeves.

When PPC clock speeds didn't keep pace, Apple pushed the megahertz myth and used some creative benchmarks in Altivec-optimized applications to prop up the architecture's reputation. When multicore PPCs were not forthcoming, they went multi-processor. When mobile G5s were not forthcoming, they gave a speed-bump to the lower ends of the laptop market with G4 iBooks and made a plethora of PowerBook models with varying screen sizes to cater to as many users as possible. There are no more tricks left. The move to Intel is almost surely indicative of the fact that there is, nor will there ever be, a G5 chip with suitable thermal and performance characteristics to be put in a laptop. Without it, there's nowhere for the portable line to go next year.

In the short term, Apple's sales are going to take a beating-- especially at the top end of the price range, and especially in the PowerBook line. It may take all of Apple's institutional fortitude not to go into fire-sale mode next year on PPC hardware. I know for myself I was planning on upgrading my G4 PowerBook as soon as a G5 (or perhaps even a significantly faster or more fully-featured G4 PowerBook) was available. Now I'll at least consider waiting until the Intel-based 'books come out, whenever that is. If Apple is able to pull it off, I'm betting it'll be sooner rather than later, as it is the mobile line which needs the switch more than the desktops.

category: 
platform: 
topic: 

Comments

Anonymous's picture

How exactly does Apple justify being a "hardware" company and continue to over charge its loyal customer base? Truth of the matter is that running on an Intel and possibly allowing Windows is simply manufacturing another PC. They only charge more because of the OS? That would make them a software company yes? How do they plan to increase market share if they do not venture into the corporate realm and make a serious dent in the home of Joe Lunchbox? Nothing I've seen or read leads me to believe they have a goal to increase market share. Great thoughts though. I enjoyed reading your article :)
narcogen's picture

In reply to: Market Share

[quote=]How exactly does Apple justify being a "hardware" company and continue to over charge its loyal customer base? Truth of the matter is that running on an Intel and possibly allowing Windows is simply manufacturing another PC. They only charge more because of the OS? That would make them a software company yes?

How do they plan to increase market share if they do not venture into the corporate realm and make a serious dent in the home of Joe Lunchbox? Nothing I've seen or read leads me to believe they have a goal to increase market share.

Great thoughts though. I enjoyed reading your article :)[/quote]

Well, in terms of what the hardware and software are worth, it depends what angle you approach it from. If things are worth what you pay for them, then Apple is a hardware company. This is the consumer perspective. The hardware is expensive, the OS is cheap. But if you don't have the hardware, the OS is useless.

If you approach if from a cost accounting standpoint, the OS costs a lot to develop and the hardware probably not so much, mainly because other factors come into play with hardware-- parts, manufacturing.

I do not believe that the architecture switch as stated is designed to increase market share. I think it is always Apple's goals to sell as many computers as they can profitably make, but this switch is about IBM's inability or unwillingness to sufficiently tailor the PPC line, which is used mainly in embedded devices, for PCs. Basically Jobs wanted a 3Ghz chip for the desktop two years ago and a low-power G5 for the laptop a year ago. Today he has neither. This transition will bring him both.

What I do think is that if there are ways to run Windows software on MacIntel hardware-- dual boot, fast emulation, a virtual machine-- then there is the possibility to remove one potential barrier a prospective "switcher" faces, which is the necessity of replacing all your software in one go and possibly losing access to some software that doesn't exist on the Mac.

It is not going to change the price of Macs. Macs range from $500 to several thousands. The generally higher price of desktops in some configurations is generally due to extra standard features Apple bundles in (firewire, etc) and Apple's industrial design-- it was never the processor, so I don't see that changing at all.

The Mini is Apple's one concession to the PC price wars, and I wouldn't expect them to make any more, it just wouldn't be profitable for them.


Rampant for over se7en years.