03 Brewster Kahle: The Connection Machine was the first search engine


So the fear of machines, or the symbiosis,
this entanglement with machines and software, essentially, especially in young kids
growing up with all of this entanglement as part of their entire life. That it will
somehow divest us of some part of our humanity, turn us into people who can only live with the technology– and what happens when there’s something like the big denial of service attack that happened a couple– what, a week ago? I still haven’t heard who did it! And why?
–We don’t know. –And it shut down, I don’t know, Twitter? It shut down this DNS company and shut down Twitter and some other ones, and we still don’t know why or what or who, and is that just a practice for a larger one? And what happens when someone else,
another outside power, shuts down our entire Internet? What do we… Can we
function under those circumstances? Oh yeah. Well, in many ways, Danny was right by calling it the Connection Machine.
It wasn’t the process; it wasn’t the memory; it was the connections. And the
connections grew out of the box into the Internet. On the ARPANET, in the Internet.
And so it’s this interconnectedness. Between people, between machines, between organizations. That’s the real power of it all. That was an amazing insight. I mean, who out of a computer would go and stare at the wires, and go and say, it’s
the wires that are important. Or it you look at the American economy, who would go and say, it’s the roads. Right? I mean it’s just… it took a different way
of thinking of it all. And what happens when we turn this sort of thing off? And
how dependent are we on it? And I’d say, extremely! I don’t know how long our food supplies would last if any of this stuff really was majorly disrupted. And now, as
we now know because of Edward Snowden, governments are actively not only
surveilling they’re actually out to control what’s
going on. So these are state actors, not just script kiddies, or spammers from Eastern Europe trying to make some bucks off of causing you to buy fake viagra. It’s where you have a whole other level of interplay that’s going on, and … So the blessing and the curse of it is,
we are building a global brain, we’re all part of it. But it requires a trust. It requires a web of trust. And mostly in the last 20 years? 2016– the Internet Archive just celebrated 20 years of a web archive of showing how people have come together to build, and share, in ways that when I was growing up people just wouldn’t do. It was just this sort of McCarthy era sort of: “Keep your head down, be careful what you say, just know who your friends are, and who they aren’t.” But then there was this, “let’s build
something bigger” that came over people and it was all of this interconnectedness of computers and it’s the Connection Machine writ large. And
there is in fact quite direct connections between the sort of Connection Machine
and the Internet of machines, of people that evolved. But we still haven’t solved
some of the basic problems, and we’ll see how we get through it! Do we actually keep sort of our meaner instincts under control? –So, a couple of things just there: one,
if you could talk about what do you see as the direct legacy of the Connection
Machine, and the Internet of today. And the other question: what should we be doing as citizens to shape the future, so that it is positive for humans, and not, say, a Big Brother scenario, and not a scenario where essentially so much is built into algorithms that we no longer can trace back and find out, well, who is deciding that, you know, I should go to
jail, or my child should go into foster care, or all of these decisions that might somehow become– –Algorithmic. –Pardon?
–It’ll become algorithmic. –Yeah, and that’s a level on top of the existing bureaucracy where then people
say, well, you know, it wasn’t my decision; the machine learning algorithm is
the one that evaluated that. And I can’t question that, because I have no
way to reach into that system. –So the Connection Machine and Thinking Machines evolved into becoming what we think of as the modern Internet. And in an
interesting way. So there’s the Connection Machine, which by the time we had the
Connection Machine Version 1, we could load up all of the published works in
these huge databases, all magazines and newspapers at Dow Jones would come
together to make a search engine, the very first modern search engine was built
out of a Connection Machine, Model 1. And then we built a Model 2 to be even
a better search engine, and then we needed to make it so that people could contact it.
And I did a project called WAIS [Wide Area Information Servers] that was at Thinking Machines, and Danny went and talked with Alan Kay at Apple, Bill Dunn at Dow Jones, KPMG, the professionals, and we put together a project. And I ran this project; it was a precursor
of the World Wide Web. It came before the world wide web; before
Gopher. It was the first client-server system where you could turn on a computer, early early Macintosh, to go and ask questions– in English! In English! This
was radical– and contact a machine that was thousands of miles away, ask it a question; it comes back. Now let’s ask this other computer a question and come back. And the idea of, “That is what a role of a Connection Machine was for” was born at Thinking Machines. And so the first client-server, what then really evolved into the web… it was being developed separately in Sweden – uh, in Switzerland, but definitely was a mosaic of NCSA before Netscape was a mosaic of Ways,
World Wide Web, and Gopher. And those systems were put together to make what became the World Wide Web. And so the thinking at Thinking Machines was far enough ahead because we could see the future because we built these machines, that were million dollar machines that we
knew were going to be[come] tiny. But Danny always had this idea of, “Let’s keep thinking BIG.” It’s like, oh, it’s great that these machines are now the size of watches.
But what’s the machine that’s big again? What’s the warehouse size machine? And
that traversal was actually some of the direct computer architecture. We started
with these tiny tiny processors: 64,000, the Connection Machine Model 1, and
more or less the Connection Machine Model 2, have these larger processors. By the Connection Machine Model 5, there were even larger processors starting to
be operating independently from each other. By the time of the next generation,
which was sort of Yahoo or Alexa Internet, the Internet Archive, hotmail– those were all built out of self-similar computers stacked on top of each other.
That was the architecture of the future. When Alexa Internet, a company that I
started– we had built this sort of Connection Machine like thing to scale–
we were bought by Amazon.com because they needed a solution; they were still
built out of these old clunky things, and they said, how do we do this? More or less, you know, direct cause and effect is always a little hard to exactly figure out, but the Amazon Web Services, EC2, S3, it all looked like that same BRICKS style that we brought in. So the Connection Machine all the way
through, to the Googles, the Amazons, now the Apples, come through
that legacy of thinking of how do you interconnect machines? How do you
interconnect people? That was done back in the 1980s at Thinking Machines.
–Yeah, and Danny mentioned in the interview that I did with him, that
Sergey Brin who started Google got his start as a student programming the CM-2
the Connection Machine Model 2, and then took the programming paradigm
that you guys had developed for working on the Connection Machines and made a
version that he called MapReduce which became the basis of Google Search, and
also incorporated Danny’s second company Metaweb in 2010. So it seems like there’re some strands where there’s a very direct lineage from the
Connection Machine. But it’s interesting to hear that you’re saying that the way
of thinking that lead to the web was also something that, perhaps in
parallel was happening in different places, but you guys were certainly a
part of it at Thinking Machines.

Leave a Reply

Your email address will not be published. Required fields are marked *