AT&T and Unix - The real issue

[Geoff Kuenning]

In article <47500006 at ccvaxa.UUCP> preece at ccvaxa.UUCP (Scott Preece) writes:

>I'd like to see AT&T recognize that if they
>got $500 from everybody who'd like to have the Unix sources they'd
>probably make more money than they have made from Unix.

I think this rather neatly summarizes the problem with the "ATT is cheating
us" theory -- there is basic misunderstanding of some of the economics
involved.  It's the old "you sell enough, the price keeps coming down towards
zero" approach.

In fact, AT&T couldn't afford to sell UNIX sources for $500 a crack, any more
than GM can afford to sell cars at that price.  The problem in GM's case is
parts;  the problem in AT&T's is (a) support and (b) paying for development.
Even assuming you charge a separate fair (and profitable) price for the
manuals and the distribution media, AT&T is still going to be stuck with a
lot of per-license support costs.  Okay, so they can charge for the support.
It's gonna cost a lot (though not $43,000).  But a lot of the rest goes to
pay for deveopment, which in this case means Bell Labs.  A neat place, but
rather expensive to run.  That $43K pays not only for UNIX, but for a whole
bunch of failed projects that might have been UNIX, might have been part of
it, or led up to it.  Or might still be -- Research Version 8 and C++ aren't
failures, to name a couple we know about.

Also, if AT&T charges $500 a crack for complete UNIX sources, you will see
two major side effects:  AT&T will have a much harder time keeping ownership
of the sources, and a lot of other software developers will find themselves
put out of business.

The first comes because there will be more people with access to the sources,
and because a court claim involving N copies of illegal UNIX will involve a
much smaller amount of dollars.  Okay, you say, so AT&T won't care so much
because of their volume.  You may be right, but it's an irreversible step,
and not one AT&T is going to take lightly and quickly.

The software developers are going to be hurt because UNIX puts a floor on
software prices.  I'm not talking about PC software for the mass market here.
I'm talking software for big computers -- Vaxen and upwards.  In those markets,
there just aren't that many customers, so you can't pay your development costs
on volume.  That's why big-computer software is expensive.  Prices of $10,000
for a binary license, $100,000 for a source license, are not at all uncommon
from garage-shop developers.  That's not based on greed -- it's based on
a realistic understanding of the costs of actually running a company,
especially a tiny one.

I already use a standard argument in negotiating prices with this sort of
company:  "explain to me why I should pay $10,000 to you when I can get all
of UNIX in binary for as little as $7200, and in source for less than $20,000,
and UNIX has a tool that does most of what yours does."  It is a hard
argument to respond to.

So when you scream to have the price of UNIX lowered so you, too, can get
rich writing sexy software, remember that when you try to sell your software,
I'm going to be standing there pointing out I can get all of UNIX for $500.
That sort of development would not destroy the software industry, but it
would change it radically.  In fact, if AT&T *did* lower the price that far,
other manufacturers would be very likely to scream in antitrust court.  Do
you think DEC can afford to sell VMS for $500 a crack?  Do you think they
can compete with UNIX if they don't?  [Okay all you :-) types, I didn't mean
VMS has to be $500 to compete with UNIX at $43K.]
-- 

	Geoff Kuenning
	Unix Consultant
	...!ihnp4!trwrb!desint!geoff