from Altair to HAL non-stop
Jecel Assumpcao Jr
Mon, 26 Oct 1998 04:37:35 -0200
Given some of the recent discussion on this list, I would like to
add my own perspectives.
Is 50 year a long time? An interesting exercise is to read the
article "As We May Think" (http://www.isg.sfu.ca/~duchier/misc/vbush/),
now over 53 years old, and then look around at the very internet/web
where this is available.
How about the Xanadu project? It alternated several times between
a breeding ground for neat ideas and a place for serious hacking
(in the good old sense of the word). In the end, the much poorer
WWW is what we actually got. But would we have it without Xanadu?
Good things take a long time to mature. Many aspects of the Alto
computer (1973) were better than any PC you can buy today and if
there is anything comparable to Sketchpad (early 1960s) available
now then I would very much like to know about it.
Though I was already programming in Fortran/Algol and friends at
the time and had even built my own micros, it was reading Patrick
Winston's book "Artificial Intelligence" in 1980 that first got
me "hooked" into computers (this book was later divided into two).
A year later I learned about Smalltalk and GUI machines and decided
that while AI was a "nobler" goal, it would be better for me to
concentrate on bringing the Xerox Parc ideas into the commercial
world. There was the temptation of shorter term projects, like
Unix+SMP, and there was the threat of AI making my work obsolete
before it was finished (the Japanese were pouring billions of dollars
on their 10 year Fifth Generation Project - it would result and
computers that programmed themselves!). I have not regretted my
decision and given what I know now it was the right one.
What I am trying to say is that there is a place for DOS, a place
for the Mac, a place for the NeXT, a place for Windows, a place for
Linux, a place for Merlin, a place for Tunes and there will certainly
be a place for more intelligent computers. I shouldn't give up on
0.18 micron CMOS technology because quatum chips are coming down the
road and neither should any of these OSes be despised because of
the others that can replace them.
Tunes should neither become a simple patch for Linux nor should it
be an attempt at the ultimate system - the HAL 9000. It is in the
initial brainstorming phase and any coding happening right now is
really more in the area of simple tests to help check out ideas than
a serious attempt to bootstrap something more final. And that is
ok - even when I restarted Merlin in 1993 there were lots of things
I had no idea about how I was going to do them (I know now...).
Why is Tunes unique? All other OS projects are either very conventional
(Unix or Windows clones) or are closed (Spin, Exokernel, Taos/Elate
and even Merlin). So this is the only place where people who want to
do things differently can come together to swap ideas. It is no surprise
that there is still no convergence of these ideas. Just give it some
time - probably *much* less than 50 years ;-)
-- Jecel Mattos de Assumpcao Jr -- mailto: firstname.lastname@example.org