Subject: The Death of Xanadu
From: shapj@us.ibm.com
Date: Mon, 22 Nov 1999 11:57:41 -0500

At 08:05 PM 11/21/99 -0500, Brian Bartholomew wrote:
>It's more 'something about companies vs. individuals'.  The bigger and
>more publicly-owned a company gets, the more it sheds non-profit values.

Please note that this reply does NOT concern abisource, thus the subject
heading change.  Also, it is anecdotal. I have insufficient data to
generalize as broadly as Brian has [Brian also has insufficient data], but
there are *cases* where what he says is true.  I've synopsized the
background so as to get to the point more quickly, but I'll be happy to
expand it. Some of you may have read the Wired article some years back
entitled "The Curse of Xanadu."  It was inaccurate in almost every
particular, but accurate to the spirit of what occurred (to my mind at
least, others disagree).

Shortly after Carol Bartz came to Autodesk, a decision was made to refocus
the company.  As part of that decision process, all of the experimental
efforts then underway were examined, and a decision was made for each to
continue it, to spin it out, or to shut it down. The Xanadu Operating
Company, Inc. was an Autodesk subsidiary at that point, and the original
decision was to shut them down.  The Xanadu crew wanted to be spun out.
Autodesk agreed, provided that Xanadu came up with a CEO somehow.  I was
consulting for them at the time, had been working with them to improve
release management, and had been involved in founding HaL, so they asked me
to take the job. I agreed, with some reservations [friends in need is a BAD
reason to take over a company].

Autodesk initially offered $50k as bridge money.  This worked out to about
three week's salary, and is one of the less subtle "fuck yous" I have
experienced in my career.  I was able, with the support of Ann Hardy, to
get them to $250k, which was sufficient (if we worked hard) to bring in
other capital. Xanadu agreed to a $7M note paying back the Autodesk
investment.  Shortly thereafter, Autodesk forgave the note and took a
write-down on the investment. [Sometimes the tax system works for you; the
Autodesk CFO hadn't recognized this as a way to effectively retrieve over
20% on an otherwise bad investment).

Within weeks of the Xanadu spinout Ted Nelson decided to try to take
control of the company legally.  You can probably imagine about how far the
$250k went in lawyer's fees.  We managed to fight him off, but it was a
pyrhic victory; the company's funds were exhausted. To keep the company
running, we needed to generate revenue fast, and with minimal distraction.

The Xanadu community had never really cared about making money.  As far as
I know *none* of the outside stockholders considered their shares anything
more than an investment in advancing the world.  What the stockholders
wanted was a hypertext system.  As Xanadu had been incorporated for profit,
the law required maximizing the *fiduciary* interests of the stockholders.
After a lot of discussion, I found myself in the curious position of
needing to execute a strategy that the stockholders did not want but the
law seemed to require: licensing the code.

At the time, Ted Nelson was adamantly opposed to giving the source code
away.  Ted held the original trade secret rights on the "ZigZag" design,
which, while we were using it, was more a hindrance than a help.  If we had
been able to do that at the time, we would have done so, on the argument
that making the code public maximized the likelihood that there would be
some ultimate payoff to the investors.  As you know, Ted has recently
changed his mind.

In the end, we licensed the code to a company that would later be called
"Filoli Information Systems".  Filoli also hired most of the Xanadu team.
The company never used that code in a product, though they *were* able to
parlay it into a sizable venture investment in their own work.

The point is that the shareholders mostly disagreed with this decision to
license.  Frankly, *I* thought it was a bad idea. Unfortunately nobody had
a better one at the time.

On rare occasions a board must act according to the demands of the law
rather than the intent of the investors. I can imagine circumstances in
which this would require taking a product proprietary.  Ssh-2 is an example
of how this can be done effectively.


Jonathan S. Shapiro, Ph. D.
Research Staff Member
IBM T.J. Watson Research Center
Email: shapj@us.ibm.com
Phone: +1 914 784 7085  (Tieline: 863)
Fax: +1 914 784 6576