Von Neumann's Universe George Dyson http://conferences.oreillynet.com/cs/et2005/view/e_sess/7032 At the O'Reilly Emerging Technology Conference San Diego, California, 15 March 2005 Impressionistic transcript by Cory Doctorow doctorow@craphound.com -- This talk is about how the universe of ETCON emerged -- the Von Neumann side. The first Von Neumann machine had 1/100,000 the power and capacity of this laptop -- which I can't get to work! [[Slide the first Von Neumann machine]] This is the reverse of today's machines: the memory is in cathode ray tubes, today we display info on CRTs, rather than storing it there. Von Neumann's whole class from Budapest came to the US and changed the world -- VN, Erdos, and others. Too bad we didn't have a Department of Homeland Security to keep them out. VN went to Princeton to study: a paradise for scholars. Princeton was about "the usefulness of useless knowledge" (eg Medieval names for badgers.) "What could be wiser than giving people who can think the leisure to do it?" [[slide: George and Esther Dyson in front of Fuld Hall 50 years ago]] Coming through the door, you'd find Einstein, Veblen and Von Neumann, all on the ground floor. Above Von Neumann: Godel. Godel did phenomenal work in 1931, in collaboration with Von Neumann (a work so profound it got Von Neumann to give up on mathematical modeling). Godel made the mistake of going to Austria to visit his family in the 30s and couldn't come back in to the US; the Reich wanted him to serve them and the US wouldn't help because he wasn't an American. Finally, Von Neumann lets the state department know that Godel was Aryan -- and they decided to help. The Germans decide to let him go to the US, via Siberia and Japan. When he's readmitted to the US, he is registered as a German -- an enemy alien. And there was no procedure to fix it "but in all probabiliuty it will be soon." You don't want to say "probability" to Godel, though! Godel spent the whole war under house arrest -- no wonder he went crazy. Finally, he becomes an American, and immediately he's drafted! Godel was convinced he was being poisoned, would only eat from plate brought o him by a trusted friend and only if she ate off it too,. But Godel's work led directly to Godel. Godel should have been in Cambridge breaking codes. Von Neumann went to Los Alamos and left: "I'm thinking about something much more important than bombs: I'm thinking of computers." By the end of the war, Von Neumann was already under contract to IBM. That's what gave IBM their head start (the other engineers didn't know this). First meetings for construction of a civilian computer was held at RCA -- they were supposed to build it there, but had a dispute over patent rights. Godel showed that you could take logical statements with numerically coded addresses and operate upon them with machines. They gave over the office next to Godel's to computer. The command-line that they developed still persists in computers today -- like a nucleotide. It was revolutionary to have engineers with soldering guns building things. They were put in the room next to he basement men's room -- and some of the historians hated this: "Mathematicians in our wing? Over my dead body." Real people doing real things wasn't totally accepted at Princeton. He didn't go to MIT to get engineers to build this, but von Neumann decided to go to Princeton. There wasn't enough room to house them -- they slept on their lawyers sofas -- they were the original hackers.. They also drank too much caffeine and they ate too much sugar -- it was proposed that they should have their tea under supervision. The stuff was built by children of researchers working summer jobs, replicating components. "Fred if you have time, please tighten all nuts" The institute was embarrassed that this real engineering project was taking place, so they locked all the reports into a garage. Von Neumann's reports were all public and non-proprietary -- they were freely shared with NCR, IBM, RCA, etc. The memory was really unreliable and sloppy -- the difference between a 1 and a 0 was very subtle. Getting all this stuff to work was akin to getting today's unreliable Internet services to work. The hackers' notebooks are full of bile: YAWN, CLOSING DOWN IN DISGUST, MANIAC LOST ITS MEMORY REGAINED ITS MEMORY, GARBAGE, CODE ERROR MACHINE NOT GUILTY, DAMN IT I CAN BE AS STUBBORN AS THIS THING, IBM IS PUTTING A TAR-LIKE SUBSTANCE ONT HE CARDS, MOUSE CRAWLED INTO BELT: RESULT NO MORE MOUSE. I HAVE NOW DUPLICATED BOTH RESULTS HOW WILL I KNOW WHICH IS RIGHT? Von Neumann's hobby-horse was artificial life. Did experiments to observe evolution of numerical organism within a machine, all in 5K! Aim: verifying evolution in an artificial universe. Real progress is the combination of diverse things. I got to the institute and opened six boxes that had never been opened. The archivist brought in another greasy box that contained on punch-cards for Barricelli's amazing artificial life simulation. Von Neumann knew that the tidal wave of computation would change the world and the world would never be the same -- and his people worked like they were on a quest. But after he died, the institute pulled the plug. Von Neumann probably died from inhaling plutonium at Los Alamos -- he died of bone cancer. Lost Von Neumann work: An organism is a universal automation which produces other automata like it in space which is inert or only randomly activated around it where are are no other universal automata (this universality is probably necessary to organize or resist organization by other automata) The problem is to see whether there will exist *boxes* of these cells containing n elements each, the # of states is then 2^n for each box: now we divide the 2^n states into k classes, and call each class a state of the box. These boxes will then perhaps be able to play the role of our present cells. -- Ulam to von Neumann, 9 November 1952