I do not expect you to keep reading beyond the latest paragraph but I'll keep typing anyway.
First of all, you're making an unnecessary shift. Were mentioned two previous transitions, both were painful but necessary; this one is not only even more painful but completely superfluous. In a period where the PC market falls into a seemingly bottomless bit, sales of Apple's computers are rising. You're gaining market share. Want to gamble everything when you have nothing to win and everything to lose? Some would claim neither IBM or Motorola could deliver for the notebooks but is Intel's offering any better? I'll elaborate on this fallacy later on.
I do not believe either fat binaries or dynamic recompilation will solve compatibility issues between architectures:
I just learned that dynamic recompilation won't work if the executable requires a G4, G5 or Altivec (among other things). Great, meaning most assembly code must be rewritten for x86 as well as many optimized binaries. How about device drivers? Must third parties write two versions of their drivers to be compatible with Macs? Weren't they already scared away from the platform because of lack of market share?
Aren't you worried that dynamic recompilation, working only one way, will leave current PPC users in the dust? I am worried that in a year from now third parties will start releasing Intel-only binaries because it's less hassle. It is particularly worrisome since many will start porting their applications to the Mac because of this transition. Right now I wouldn't think of buying a PPC Mac. I believe your sales will crash starting from now. Besides, that's not the only reason why it will happen: potential customers approving the switch will obviously wait while those ambivalent or disapproving will still prefer to stay away from an architecture that has been announced will become obsolete.
You're claiming that recompiling PPC binaries on the fly for x86 processors is fast enough. My personal experience with such recompilation prevents me from believing such a thing. It's going to be awfully slow, probably just as much as Java, or any software that runs in a virtual machine. Apart from that, loading time will be atrociously slow. Caching recompiled code might be an option, but then it will break any applications that internally generate code (like most emulators) or modify their own code segment (many useful anti-debugger tricks are based on that principle), while requiring even more disk space.
The use of fat binaries is an inelegant bloat. That means twice the required disk space and twice the download time (at the very least). As for developers, that also means twice the assembly code (as mentioned above; for many applications, that means almost twice the development time), twice the compiler glitches (as a programmer, I can attest there will be quite many), severe endianness issues (data on file must be read by platforms with different byte ordering, meaning many patches), and more frustrating surprises.
What's the point of choosing Intel as a partner anyway? A company having so little vision that it has bet its future on the infamous Itanic and is now left with no long-term alternative? You claimed that IBM couldn't deliver its 3GHz G5 but isn't Intel having the same problem? Where is the 4GHz Pentium 4? Can you really believe its notable absence is due to a "marketing decision" as they claim? You claimed that Intel chips are faster but then how about your previous claims about the Megahertz myth (which is truly a myth anyway)? You claimed that Intel could provide faster, low-power chips for laptops but aren't they using low-performance Pentium M's, Centrinos (performance being especially bad when it comes to integer computation and many multimedia tasks like DivX encoding, and what to say of the fact that the benchmarks giving it an edge, even for day to day applications use Pentium 4's with at most 1Mb L2 cache while newest ones have 2Mb)? Because Pentium 4's in notebooks heat so much that the fan must blow continuously -- forget thin, quiet laptops. By the way, how about your shift to 64-bit machines? What's the point of switching Powerbooks to Pentium M's, a 32-bit processor, since your original complaint is that you couldn't put a G5 in, a 64-bit processor? Or if you plan to use real P4's in Powerbooks, I wish you luck. Speaking of 64-bit, don't you mind at all that Intel, once owning its own architecture, is now following AMD's lead by copying x86-64? By the way, aren't you a bit worried about the fact that Intel is losing ground to AMD in the desktop processors market (since last year) just like nVidia is against ATI in the graphics chipsets market (thes two companies have a lot in common by the way) and that you may have to divorce from them even before marriage has been consumed? Oh, and how to explain that x86 processors are so much better than PPC (that can't be further from the truth) given that everyone else is switching to the latter architecture, even Microsoft with the XBOX 360? Do you really feel the need to run counter-mainstream at all costs, enough not to learn of the mistakes of others?
You claimed that Mac OS X will never run on generic PC's. You wanna bet? How about I predict that whatever scheme you have in mind to prevent this will be defeated almost immediately by hobbyists, may it be engraved into the BIOS or otherwise? That software like VMWare (or better, upcoming Linux kernel versions) running an operating system on top of another may easily support the Mac? If it's so much about the OS like many Mac users claim, why would they bother buying Apple hardware when they can just copy the OS and crack it? Or do you plan to introduce such insane amounts of DRM into your operating system that it will alienate even legitimate users?
Aren't you awfully worried that Wine will be ported for the Mac? And then that switchers will keep using Windows applications on their new machines? Far from being good news as a transitional help, that means a far less reliable platform overall. Gone will be the days of the secure operating system with solid applications -- a Mac will just be another PC, with the same awful reputation than Wintel machines (actually, it will even run Windows -- even if not officially supported). Give a warm welcome to mail worms (or do you seriously think people will abandon Outlook, even those switching to Mac OS X?), Internet Explorer security issues, and more.
On top of that, all of this is bound to make the system more complex, therefore inherently less reliable and secure. And what to say about the memory usage, which is obviously bound to rise significantlty? Isn't that going to deter performance even more than it already does?
I guess you didn't think of the fact that making so many transitions every now and then may also scare many potential customers and partners away, if only because they worry about you making more of such reckless transitions or simply out of frustration. Or perhaps having to pay an extra $999 for your Developer Transition Kit will do the trick.
Let's summarize the above: this transition means a likely dark period of low sales, plenty of nasty software issues for both users and developers, hardware issues for laptops (supposedly the main reason for the transition), your reputation sinking rock bottom over time for all the reasons mentioned above, a complete absence of vision just as characteristic of your new partner, and overall a move just so devoid of meaning that I couldn't believe the rumors until I read the press releases. In short, so long Apple. It was good while it lasted.
About the author:
I am a low-level, embedded computer programmer who started like many on a Windows desktop, then switched to Linux, then switched to the Mac. And plans to switch back to Linux before long.
If you would like to see your thoughts or experiences with technology published, please consider writing an article for OSNews.