>   Well, for companies who think the only way to obtain something is to
> buy it for list price, sure.  But those morons get what they've got
> coming.  But even in view of that, $1000 will buy you a hell of a lot
> of UltraSPARC crunch power, even brand new from Sun.  (The Netra X1
> comes to mind...I have one here, and I love it)

there's been a similar story around the alpha for some years now.  but
here's the deal.  i've got a decstation 5100 (and a 5000), and a ka650
and a ka655, and a macIIx, and even an old zenith Z100, and i've got
vendorware for all of it, plus netbsd for most of it, and it's a lot
of fun, but none of it runs in "production."

production means among other things that it's part of a strategy whereby
i can pull parts out of other "production" machines to fix or expand it,
and it means i can copy compiled binaries over and run them.  usually
"production" also means it's something i'd buy more of if i needed it
(unless there's a migration in the works.)

so my 2xPiii@500MHz@1GB has two big monitors on a matrox G450, and i've
got three of these (home, ISC, and PAIX) and they are binary compatible
with the back office "cycle and file server" iron in those locations and
if i need more such iron i'll call billy bath and it'll show up and it
won't be an UltraSPARC or even an alpha -- even though if i needed more
iron for my hobbylab downstairs (and frankly my dears, who doesn't need
more hardware for their hobbylab downstairs?) a pentium would be thought
cheesy and awful.

sun and dec (compaq? hp?) have a lot of inertia to overcome in selling
their non-pentium machinery.  ISC has a lot of legacy alpha boxes, but
it's a little worrisome that 4100-class and even ES40-class boxes with
heavier configs than ISC's main servers are showing up on ebay for <$5K.

> >  (At least for generic "you" - perhaps
> > you(==Dave, or even ==Paul) know where to get "well-designed computer"s
> > cheap; if so, would you be willing to share the knowledge?)
> 
>   Sure.  The US Gov't surplus market.  Another excellent source is
> leasing companies...they frequently sell off-lease hardware for
> pennies on the dollar.  ...

or e-bay.

http://search.ebay.com/search/search.dll?MfcISAPICommand=GetResult&ht=1&SortProperty=MetaEndSort&ebaytag1code=0&query=dec+alpha

> > > Indeed, the Pentium4 is falling short of Intel's sales projections
> > > (or wishes?) by orders of magnitude, and Itanium is dead-on-arrival,
> > > and people are moving away from Microsoft Office in droves due to the
> > > forced-upgrade licensing crap that's about to be put into effect.  Do
> > > you REALLY think that side of the industry is going to last much
> > > longer?

corporations are sheep.  it's human destiny when operating in crowds to be
stupid in some ways, including just paying into the forced-upgrade licensing.
microsoft knows exactly what they can get away with in order to benefit from
yet preserve inertia in this case.  there's every reason to believe that
microsoft office and windows XP licensing will each grow by double digits
for each of the next five years.

> > Yes, I do.  (I would _love_ to be wrong; as I said above, I don't like
> > the x86, and would be delighted to see real choice appear in the
> > commodity computron market.)

then you should support apple in every way you can.  i bought a tibook,
whose keyboard i despise, but i'm going to learn it and run it and someday
get rid of my dayjob wintoy.

>   Software...The mainstream PeeCee software market (i.e. Microsoft)
> exists because there haven't been useful (at least according to
> perception) alternatives for Microsoft Office and the file-management
> GUI of Windows.

i wholly disagree.  it's because an MIS manager in the global 2000 has a
much bigger problem to solve than "which one is useful."  an army of
support people has to be trained on any chosen alternative, for one
thing.  the total cost of ownership of a given seat in the corporate
empire is known (budgetted) down to about a $25 granularity, and the
seats which aren't standard (usually including mine) can receive no
support unless there's a transition in which case the "new seats" will
have a cost, during transition, that's probably 2X higher than what you
would think of as "hardware cost."  and that's without amortizing all
of the back office and training costs, which add up to tens of millions
of dollars for just a domestic setup.

> But now there ARE...it was either the New York Times or the Wall
> Street Journal, I don't recall which, that just ran a big article
> discussing how many of Microsoft's major high-volume business
> customers...the people who buy $2-3 million dollars' worth of
> Microsoft Office licenses at a shot...are investigating alternatives
> like StarOffice and Linux with great haste, due to the licensing
> change that was announced last year and will go into effect in a few
> weeks.

i wish i could say "that's a load of crap" but we're in polite company and
it's the morning after and i no longer have booze as my excuse.  but no part
of microsoft's serious revenue stream is capable of making a turn within a
few-week radius.  and the TCO, even with microsoft's predatory licensing,
is still lower than the TCO of an open source alternative, unless an MIS
planner can look more than three years out, which mostly isn't allowed.  let
me say again that microsoft knows who they are selling to and they have done
_nothing_ to cancel out any inertia through their latest set of moves.

>   (I sure wish it weren't Linux, but I'll take it over Windoze any
> day!)
> 
>   Anyway...not trying to be argumentative...just presenting a
> different point of view.

let's imagine you're PAIX and you've got ~80 employees who have ~120 "seats"
if you count laptops and power users and whatnot, plus another ~40 "back
office" servers (file, print, web, firewall, what have you).  the TCO for
these, assuming your parent (MFN) just went into chapter 11 bankruptcy, is
only on a forward going basis.  you can go forward with win/XP and office/XP
and deal with predatory licensing.  or you can switch.  let's say you decide
to switch to linux or netbsd, with staroffice.  TCO is the a&d on licensing
*plus* a huge amount of one-time costs, probably including consulting, to get
trained and switched over.  *plus* you're taking the risk that you might have
"guessed wrong" and will have trouble finding qualified sysadmins a few years
down the road *plus* the difficulty in getting third party apps like SAP to
run in case *they* decide that they've guessed wrong in a few years.

TCO on a switch-to-macosX solution is about the same since the money you
didn't spend on consulting got spent on all new hardware, and you're likely
still running microsoft office anyway, and your down-line risks are minimized
because you know a lot of other sheep are turning into the same grasslands.

but nothing is as cheap as staying with microsoft and upgrading to XP, unless
you look at the 4-year and 5-year and beyond costs of predatory licensing.
it's likely that you are mostly concerned with a three year EBITDA projection
and that there's no room inside that horizon for these kinds of switchover
costs.

any reporter for the NYT who fails to take TCO into consideration is insulting
your intelligence.