I debated about commenting on this one, but I couldn't resist.  In my
experience you can' go to far wrong by sticking with the "marketplac"
as long as yu undersand it.
 
Burroughs?  a nice tecnical piece of work, but you couldn't keep them up.
 
IBM mainframes?  Outdated, but for 25 years (1960 - 1985) they were the
best the marketplace had to offer for the large-scale user in terms of
rubustness and maintenability.
 
Windowa 3.1, etc. a good piece of work, considering the horrible
16 bit architecture if 286's abd previous Intels (and the requirement
of compatibility with them, until recently)
 
soda in cans?  bottles are available here in Maryland, I don't know
about California.
 
The Mac? Couldn't go anywhere because Apple wouldn't license the
hardware design.
 
In summary, strong marketing can't do it all; you have to have a decent
(not necessarily the best) product.  If marketing savvy could do it all,
we would still be on IBM's and HP would be a good instrument maker, no
more.
 
OK. I'm off my soapbox.
 
NMD