RE: [tlug] Re: [CoLoCo] RESPECT MICROSOFT

["Stephen J. Turnbull" <stephen@example.com>]

Curt Sampson writes:

 > It's not just about market demand: it's about return on investment.
 > Sure, Intel could make money on memory. But it can make a lot more on
 > CPUs, because there's less competition. Micron can't make a CPU; only
 > a few companies can. So Intel can make more money by concentrating on
 > higher-margin markets, and expanding there.

OK, so Intel is mostly catering to Microsoft here, I suppose.

 > Microsoft made a conscious choice to try to move into the consumer
 > electronics market, and that is what led to the DRM demand. They could
 > have just as easily decided to move into consulting,

"Easily" in the "talk is cheap" sense?

Microsoft can't do consulting.  Nobody trusts Microsoft, they have
made any number of plays based on latching on to others' information
while preserving the secrecy of their own (cf the evil role they're
playing in the electronic voting market).  Furthermore, Microsoft has
always made its money in the most commodity of commodity markets; they
know nothing about tailoring product to the individual customer, which
is the essence of consulting.  Hardware is not a market they want to
be in for similar reasons, as well as bringing them into direct
conflict with Intel.

These actually are very hard markets for Microsoft to enter.

 > > In other words, Microsoft and Intel need to have the technology
 > > for use in embedded systems aimed at the consumer market
 > > anyway. As long as they have it, why not add it to the WinTel
 > > product lines, too?
 > 
 > Ok, now we're firmly back in an area where I do have expertise, as
 > do you. And I'm not buying this for a minute. As you know from your
 > extensive experience with XEmacs development, integration is expensive,

I don't do device drivers.  This is all about device drivers, no?
Correct me if I wrong, but AFAIK the same drivers with a bit of shim
are often used for both Linux and the BSDs.  Yes, integration is
expensive, but I suspect that this task is fairly easy as integration
goes.

 > probably as expensive or (IMHO) more expensive than coding. Plucking
 > code out of a set-top box or whatever and dropping it into Windows is
 > not likely to be a trivial operation.

Well, my assumption is that the set-top box *is* running Windows CE or
whatever the stripped down device manager version is called.  Windows
CE only makes a lot of sense if it presents the same APIs as its big
brother does, no?