> I see it playing out this way:

And at what time does start licensing of Mac OS X for non-Macintosh
hardware?  Or are we going to have a repeat of my 1985 discussion with Chris
Espinosa (Apple Employee #8) on that subject (short version):

"Me: Why don't you license the Mac OS for other hardware?

Chris Espinosa: We'll never do that; we're a HARDWARE vendor! The software
is just to sell our hardware.

Me: Software developers don't care about hardware. We want our software to
run well and broadly. The other GUI platforms may suck, but if GUIs become
popular, they'll keep plugging away.  No one will let Apple own the entire
market for GUI computers."

By the time Apple did get around it licensing the MacOS, years later, it was
too late and poorly executed.  Non-Mac GUIs sucked for a long time (heck,
MS-Windows STILL hasn't quite figured out color models, and ICC wasn't part
of every version until last year), but hey ... time passed, and Apple died,
only to be CPR'd by the company that killed them in the first place.

MacOS X represents a second chance for Apple to learn from past mistakes.
The core is *nix, and the whole could be ported to other processors and
machine architectures. There is synergy with the burgeoning Linux/FreeBSD/et
al community. There are applications which can migrate forward from classic
MacOS, as well as from *nix.

Programmers care about solutions to programming problems, and markets large
enough to support development.  More progammers were using Pascal than C in
the early '80s; many switched to C because it was similar enough to learn
easily, and provided better performance by being more closely matched to the
hardware (Ira Ruben had to make lots of extensions to Pascal to make it
suitable for systems work).  Many switched to C++ from C because it provided
a more productive environment.  Many switched to Java from C++ because it
provides both portability and productivity.  The masses have NOT switched to
Smalltalk because it is not perceived as a suitable tool for generating
turnkey applications, and because it is not compatible with their knowledge
of C/C++/Java.

There are already efforts porting .NET to non-Windows server platforms.
Don't know about any for non-Windows clients, but I'm sure they'll come.

I do think that people will balk at the constant cost "subscription model",
which is pay-for-use, but who knows?  And has for your scenario that has
Windows dying off in favor of free software, there are plenty of arguments
that the free software model will fail, too.

The Open Source model may very well collapse; it does not generate revenue
for the majority of practitioners, and most Open Source programs are awful.
Most open source projects are started by someone who wants something, and
when they get to a certain point, they stop.  Many die on the vine, without
ever ripening.  Are there exceptions?  Of course.  Linux, Apache and SAMBA
come to mind, as do other projects coming from academia.  And the notion
that a company may pay its staff to develop/enhance an open source program
to solve its problems is reasonable.  We did a lot of work on a particular
open source server because we needed it to work properly in order to solve
our problems.  Mind you, we did not continue to fund development after it
got to a certain point; we were solving our problems, not being altruistic.

The Squeak license is one of the smartest compromises between Open Source
and allowing commercial programs. The problem is funding work.  There are
issues apparent today with respect to a lack of funding. For example, who is
going to pay Tim to produce the StrongARM jit? For that matter, who is going
to finally polish J3, and make it a general part of the VM?  Who is going to
fund work so that a Squeak PDA can interoperate with Desktop PIM
applications (e.g., Outlook, Notes, etc.)?  Who is going to fund someone to
sell potential funders on the advantages of using Squeak?

Another problem is that, by and large, most open source programs are written
by the in-crowd for the in-crowd, with little regard to the masses.  The
vast majority represent the 80/20 rule, and no one wants to do the 20% dirty
work neccessary to make a program really usable. The mantra is "the source
is available, do it yourself."  Sorry, but that doesn't fly except to
programmers, and *nix admins. In this sense, most Open Source programs
remind me of color slide film.

Color negative film is not as good as color slide film, and the photographer
cedes control of the final image to the print processor, but negative film
is easier to work with (wide exposure latitude). No matter how much slide
film shooters may extoll its virtues over print film the photographic masses
neither capable of using slide film, nor interested in using it; slide film
represents 2% of film sales.

For that matter, lack of quality, and lowered consumer expectations aren't
unique to the computer industry. Recently, Kodak killed their best color
(negative and slide) films in favor of faster, grainy, lower resolution
films more suited to the masses. And although today's consumer digital
imaging is attrocious, it is growing faster than anything. Average image
quality is getting worse and worse, but the buying public doesn't know the
difference. Another 10 years will pass before digital imaging is as good as
the finest emulsions available today.

<<shrug>> Who knows?  It will be interesting to see where things go over the
next 5 years.

I do think that if Squeak could compile and run Java in the VM, that you'd
see a huge increase in Squeak interest.

	--- Noel