Okay, so the whole thing is, at some level,
what is the legacy of Thinking Machines, and what is the legacy of the Connection Machine and the ideas of AI — and of computers that were extant 30 years ago. How are
they today? Have they changed? What are they moving into the future?
Thanks Brewster Kahle for talking to me about this, as you were probably after
Danny the most significant scientist working on the machine, is that true? Or
can you talk about your role on the Connection Machine? –So yes I started working with Danny at MIT early on on this project when he was a 24 year old
graduate student. –How old were you? –I was 20, and I wanted to build
the Library of Alexandria Version 2. I wanted to build the global brain, and Danny had this idea on how to make a machine that would be fantastic. And I said, oh, how hard could that be? I’ll spend a couple years and build that machine. Well, it was a lot harder than that, but it… What a fantastic trip! Danny had this… He wanted to build a machine that would be proud of him. Which I thought was really an interesting way of viewing the whole thing. These were the heady days
of AI and computers when we could see the future. We knew it was gonna happen, and this Carl Feynman, Richard Feynman’s son– –Who’s also in this picture here, on your left.
–Yeah, in this picture, and I think he was the
one that said to me, if we’re… if we’re going to bring up the next
generation, if these machines really are going to take over from us, then let’s
have them read good books. Let’s have it, let’s teach them well. Let’s
bring up good machines. And so we really thought of it in that kind of heady way
as there was a generation change happening. –A generation change in
the machines, or in the people? –And what year was this?
–Oh, in the species! We were bringing up the next generation of evolution. –Of Homo sapiens?
–Well, Homo sapiens will be about as relevant as zoo animals. Homo sapiens were… We get to try to help bring up the next generation of what comes next, and the things that are gonna go to the stars? They’re likely to be more relating to Connection Machines than they are going to be relating to
human biology. –Can you put a year on this? When were you starting these conversations with Danny and with Carl? –These are sorta 1980, 1981, 1982 at MIT.
And then it came time to spin off the Connection Machine
from MIT to go and build it. And Sheryl Handler was this wonderful woman
that she completely caught the dream and understood where this whole thing could
go, and tried to provide an environment for these, you know, looking back on us,
cocky kids, that sort of had this dream, and she said, I could really help make
this go. And getting involved with Paley of CBS to go and fund this – really is a
24 year old Danny’s kinda wacko dream! Yes, there was Marvin Minsky who seemed more
established, or at least older, but it was still kind of wacko. Let’s give
this thing a shot! And the pulling together of amazing people. There were more…
There was a better faculty at thinking machines than there was at MIT that I
was exposed to. So, Richard Feynman, Stephen Wolfram, just to drop names, I
mean, Doug Lenat — they were all coming through and trying to help shape what it
is this next generation would be. The idea… But also there was this sort of new, but there was also old.
So Danny would go back to the thinking of Turing and was friends, got to meet and hang out
with Claude Shannon, the inventor of the bit. Right, the concept that there’s a one
and a zero, and these are the new atoms that are going to build the next generation world. This required going backwards in time,
to go and find the thinking that made the computers the way they became,
because only those people in the very beginning saw the diversity of possibility. And yes we ended up with one von Neumann based machine, with one computer central processing unit with external memory, but there are lots of other possibilities. By Danny’s reading, these old original papers were very inspiring to me, and this concept allowed Danny to see things in a very
different way… like what if we made the memory smart? So the problem… the computer
wasn’t the central processing unit. The computer was the memory. And it
wasn’t even then just the memory, it was the interconnectedness of it all. Neurons
were part of this whole conversation of how do we build machines to really take
a major leap forward? The idea of, like, we’re not building a multiprocessor with
two or four processors or eight; let’s go with a million! –So you think that what was
unique, or one thing that was unique, was that instead of saying, what do we have
now and how can it be better, that you guys actually went back to the origins
of computers and said, how were they thinking about it? And how can we take a
branch that then was not taken because of the technology that they had, which
we can now take? Is that what you’re saying? –Yes, absolutely. It was looking, it
was going backwards to be able to go forwards. And it was just stepping back
and taking a much bigger view, and it was thinking big! That we could do
these things that, okay, at the end of the day, we didn’t achieve a lot of them! But
it’s set a course and an ambition that has directly, indirectly inspired this
forward thinking that has become, oh gosh, the Internet, Google, cluster computing, the world that you know now, in 2016.