By John Gruber
Discover fractals and create your own fractal art with Ultra Fractal.
John C. Dvorak’s modus operandi is to instigate. He is a button-pusher, seldom if ever trying to inform, preferring instead to inflame. And he’s pretty good at that. A decade ago, he was the back-page columnist for the now-defunct MacUser magazine; almost every month, the MacUser letters-to-the-editor page contained at least one angry message asking “Why do you publish this guy’s column?” The answer, of course, was because it was the sort of column that inspired people to write letters-to-the-editor.
Dvorak is a pundit, not a reporter. When he makes a prediction, it is usually based on nothing more than his own conjecture, not actual sources. And looking at his track record, his conjecture usually has more to do with what he thinks will be controversial, rather than what might actually happen. (E.g. he’s often predicted that Apple was about to go out of business, a prediction which never ceases to get a rise out of the easily incensed.) There’s nothing wrong or dishonest about that, but it’s something you need to keep in mind with everything he writes. To the best of my knowledge, he’s never had a serious scoop regarding Apple — a significant prediction that turned out to be right — and he’s been on the job for at least two decades.
Recently he’s been trumpeting the idea that Apple will move the Mac platform to Intel processors. His column on this subject from last month was widely linked to, everywhere from Mac-oriented weblogs to Slashdot. Having struck gold with that one, Dvorak returned to the topic yesterday with a column titled “How the MacIntel Will Change the Market”. (Note the use of “Will” in lieu of “Would”, subtly casting his conjecture as fact.)
This really isn’t worth debunking point-by-point. Just remember that Dvorak is pulling all of this out of his ass. He has no sources, on- or off-record, claiming knowledge of such a project. The only actual facts Dvorak has in favor of this Apple-switching-to-Intel rumor are:
Everything else is conjecture, with lots of hand-waving. I’m not going to say it’s impossible that Apple might switch to Intel processors at some point in the future, but if they were to do so, they’d need to overcome several mountainous obstacles, none of which Dvorak addresses.
Fans of the idea of Mac OS X running-on-Intel often point out that the low-level code in Mac OS X is processor-neutral. NextStep ran on multiple architectures, and Darwin on x86 is actively maintained. This is undoubtedly true: much of Mac OS X, including anything Darwin or Cocoa, could probably be ported to another processor with a simple recompilation in ProjectBuilder. The way bundled applications work in Mac OS X, a developer could include both PowerPC and Intel binaries in the same application package.
Great. But what about the remaining software? Like, say, everything Carbon, which includes most of the major commercial applications people actually use. Or the Classic environment and classic Mac OS software, for that matter? That stuff is not going to be recompiled or rewritten for another processor platform.
Dvorak (and other fans of this rumor) like to point out that Apple successfully switched processor platforms before, from Motorola’s 68K to the PowerPC. But the reason that switch worked is that Apple wrote an amazing emulation layer for the PowerPC, under which old 68K Mac software ended up running nearly as fast as it did on actual 68K chips. When you bought a new PowerMac, your old Mac software just worked.
The only way Apple could switch from PowerPC to Intel as seamlessly as they switched from 68K to PowerPC would be if they were to implement a PowerPC emulator for this hypothetical Apple-Intel platform. Not just a working emulator, but a fast working emulator. I just don’t think this is possible, and I’ve never seen a credible report claiming it is.
And without a fast PowerPC emulator, all current Mac software would be dead. Sure, companies like Adobe, Microsoft, and Macromedia might release the next major updates to their applications with native support for the new processor, but what about the versions people already own? Who would buy a new Mac that couldn’t run the thousands of dollars worth of software they already own?
Dvorak posits the idea that Apple could create machines containing both Intel and PowerPC processors, but this seems rather sketchy, at best. (Not to mention expensive and hot — you think your 12" PowerBook is hot now?).
This point cannot be emphasized strongly enough. Apple is a computer hardware company. Selling hardware is how Apple generates most of its revenue. Their operating system software may well be the best aspect of their computers, but that does not make them a software company. Anyone who claims that Apple could simply switch to being a software company and make up for lost hardware revenue by selling additional software doesn’t understand how the company operates.
During the brief period of time when Apple licensed the Mac OS to other manufacturers, their revenue tanked. Too many people bought cheap clones from PowerComputing and Umax instead of higher-priced Macs from Apple, and the licensing revenue didn’t compensate for the lost hardware revenue. The situation may well have been good for Mac users, but it was terrible for Apple’s bottom line.
No matter how badly people clamor for it, Apple is never going to release a version of Mac OS X that runs on standard Wintel PC hardware. Whether it’s possible or not, it isn’t going to happen. A frequent comment regarding this rumor is something like “I’d love a version of Mac OS X that ran on my PC.” Sure you would, you cheap bastard. Apple’s Switch campaign is an attempt to get PC users to buy thousands of dollars of Apple hardware, not hundreds of dollars of Apple software.
Many people want this to rumor to be true, as they envision faster Macs that cost less. Meaning, cheaper computers from clone manufacturers. Yes, that business model has worked wonderfully for Microsoft, but Apple is not Microsoft. A fundamental aspect of Microsoft’s business model is that there’s only room for one Microsoft.
As I’ve written before, Apple’s primary goal for the Mac is very different than Microsoft’s goal for Windows. I wrote:
The only way to see the Mac as unsuccessful is to compare it to Windows on Microsoft’s terms — market share and raw profit. And that’s exactly how analysts and the PC press cover the Mac.
What they miss is that the Mac’s primary purpose is to be better. Windows’s primary purpose is to be ubiquitous. Both platforms have been successful in achieving these goals. That’s not to say they’re mutually exclusive. Apple would of course love to achieve higher market share. Love love love. And Microsoft doesn’t purposely make Windows uninintuitive. Well, maybe they do. But it’s not as bad as it used to be.
So while Apple would love to clip a few percentage points from the Windows user base, they’re never going to mount a full assault on Microsoft’s Windows hegemony. The vast majority of Wintel PCs are sold as disposable business plumbing — the adding machines and typewriters of the 21st Century. Apple wants no part of this low-margin market.
Thus, even if Apple were to switch to an Intel processor, they would not be switching to the standard Intel PC architecture. They’d continue making proprietary Apple hardware, but which happened to have an Intel processor on the motherboard. You’d still need to buy an Apple computer to run Mac OS X.
Dvorak alternatively ignores and confuses this issue. His latest column advocates Apple releasing both proprietary Intel hardware (based on the Itanium 64-bit processor) and releasing a separate version of Mac OS X that runs on standard PC clones. Oh, and he also wants Apple to continue manufacturing PowerPC machines. So he wants three Mac OS X platforms: PowerPC, Itanium, and regular PCs. And somehow this is supposed to make sense.
Part of Apple’s appeal is that their product line is clear. You want fast, buy a PowerMac. You want cheap, buy an iMac. Portable? PowerBook, fast; iBook, cheap. Apple moved to the PowerPC by dropping the 68K, alleviating any chance of confusion over whether to buy a PowerMac or a Quadra. If they were to switch to a new processor again, they’d need to do the same thing and drop the PowerPC. But that leaves the problem of what to do about existing PowerPC software.
Dvorak resorts to confusing convolutions to distract from the glaring holes in his purported plan. Part of Dvorak’s sucker argument is the notion that it would be easy for Apple to make the switch. Don’t be fooled. Possible? Maybe. But it would be both technically difficult and financially risky — more so even than the switch from Mac OS 9 to Mac OS X.