Quotes From Kelvin

[Marc Stiegler <marcs>]

Date: Tue, 6 Dec 88 10:27:56 PST
From: acad!throop!kelvin (John Walker)
Message-Id: <8812061827.AA00291@xxxxxxxxxxx>
To: acad!xanadu!marcs, file
Subject: Comdex trip report

Thanks for the detailed trip report.  I'm glad that things went well at
the demo; every time Xanadu performs professionally and the others
stumble the relative flakiness indices are readjusted.  It will take
time to equilibrate, but at least we're heading in the right direction.

I haven't seen KMS and I know nothing about it, but it seems to me that
we should ask ourselves one key question about its positioning and
potential: is this a Sun-only (or Sun/DEC/Apollo-only) product?  If so,
I don't think it's likely to become established as any kind of significant
standard outside the community that buys such machines, which is
insignificant in the larger world.  I do not denigrate the high-end 
early-adopter market or wish to write off any segment of the universal
market, but I think there's a strong parallel between CAD and hypertext
here.

The aerospace companies were not only the early adopters of CAD--in large
part they developed it.  CADAM is a division of Lockheed
and was originally an in-house system.  McDonnell-Douglas' CAD system
started similarly, and CATIA was Dassault's private 3D system before
it was marketed by IBM, and those aren't the only examples.
Notwithstanding enormous high-end penetration by those systems, we were
able to take over the CAD market starting at the low end, not just by
addressing people who couldn't afford those systems, but eventually
entering the aerospace companies after having become established
as the de-facto standard in the outside world.  The large companies
don't want to spend any more than they have to, and if provided with a 
low-cost solution that does the job and is compatible with the outside
world represented by their subcontractors, they will adopt it.

Note, by the way, that data interchange among CAD systems is generally
a much more difficult and painful problem than interconversion of
word processing or other document data.  If any of you are repelled
by SGML, I invite you to spend an afternoon (or week, or month, or year--
I'm easy) studying IGES.  And remember that IGES was the result of the
Air Force and a bunch of big companies ganging up on the CAD companies
and saying "standards or else", not an attempt by the CAD companies to
obstruct competition.

If KMS sees themselves as a Sun product (or a S/A/D product), at least
through 1989, I think we're quite likely to just walk in and snatch the
mass market from beneath them simply by having a product which runs on
the platforms that utterly dominate the market, and promoting it as the
platform-independent solution to collaborative work and data interchange.
If KMS captures EVERY Sun user, their market share will remain tiny
compared to what we can achieve with a small initial penetration of the 
PC and Macintosh markets.

I am not arguing for adopting a "don't worry" approach to KMS or any
other competitor, shipping or not.  Frankly, in a market where the hype
level is rising as rapidly as it is for anything remotely called
"hypertext", Xanadu is dangerously far from market and seemingly often
unconcerned with near-term delivery committments.  We are in a contest for
survival and ownership of the marketplace which Xanadu should control and
does best, and we should approach that battle with the deadly seriousness
and urgency that it deserves and requires.

Every argument over terminology, office layout, operating system or
language arcana; every 16 hour bull session about science fiction,
molecular engineering, or anarchy is, unless a last ditch sanity-preserving
retreat, allocation of precious resources to something that does not
get Xanadu to market earlier.  When we were developing AutoCAD, and for
that matter, for most of the last 7 years, our key development team
"went ballistic" to the exclusion of essentially everything
else.  Why?  Because we were scared to death while at the same time aware
that we grasped a once-in-a-lifetime shot at the Main Chance.

At the same time, we were not, and I think you must not be, afraid of
getting your hands, minds, and souls dirty by working on whatever is
needed to get the product in the hands of as many people as possible.
I'm sorry, but you aren't going to get much sympathy from me about
having to hack OS/2 Presentation Manager, Phar Lap 386 DOS Extender,
Kludge-O-Matic IN-SEINE-NET, Macintosh Multifinder, or what ever
bete noir you can't abide this week.  After writing assembly-language
drivers for dozens of the most ghastly display devices ever conceived
by the mind of man, after porting code such pinnacles of engineering
as the Hewlett-Packard 150, the DEC MicroVAX I, the DEC Rainbow, and
Roland the Mindless B-Size Plotter, not to mention assorted plasma
panels, IBM RTs, and sonic digitisers (I'm not making this up), all in
search of more customers or to meet a market need, perceived or not,
I can't help but feel the current landscape is pretty clean and
inviting.

Xanadu must dare to get dirty, and must provide products, including
professional-quality front ends, that enable the widest variety of
customers to get running on Xanadu as soon as possible.  If we do this,
and our products meet the need, and are well-supported and aggressively
enhanced, then Xanadu will have bright prospects regardless of what
happens in the Sun market.

I agree that we our major communications effort should be deferred
until we have a product, but not so much from the fear of paving the
way for a competitor as from the belief that the market (wisely, IMO)
simply discounts all promotion for products they can't buy.  I would not
interpret this as precluding demos of what we have as we go, as long
as people understand what they're seeing and not, of course, spilling
any strategy or front-end plans that we consider proprietary and best
kept under wraps until formal roll-out.

Feel free to redistribute this message as you wish.