3 ms·
The subtitle of this article "How the game theory of John von Neumann transformed the 20th century." isn't addressed in this article unless briefly in passing.
by kristianp 4y ago
The subtitle of this article "How the game theory of John von Neumann transformed the 20th century." isn't addressed in this article unless briefly in passing. Poor editing (if that's the word for it).
This part is intriguing:
> In his Theory of Self-Reproducing Automata (published posthumously in 1966), von Neumann went even further, describing the conditions under which, with no more than eight parts (four structural and four dedicated to logical operations), complex creatures could emerge capable of executing any type of computation and even of replicating themselves.
Von Neumann machines are a fun part of science fiction. The phrase "Hegemonic swarm" from Iain Banks comes to mind.
- joe_the_user 4y agoYeah, I think that's because the subheading basically isn't true. Game theory is more of footnote on von Neuman's many amazing achievements than a capstone. It influenced cold war thinking certainly, but the situation of the US and USSR didn't really fit the zero sum, repeated games that Von Neuman originally formulated studied. Essentially, the US and USSR faced a "Mexican standoff", "mutually assured destruction"(MAD), a non-repeated, non-zero-sum situation you can claim has a mathematical solution but which ultimately is more of a psychological problem.
- eru 4y agoIf you want to model MAD via game theory, it would certainly be a repeated game. They faced each other day in day out year after year. I agree that war, both cold and hot, is a non-zero sum game. That's why things like temporary ceasefires can work at all. Or Geneva conventions.
- selestify 4y agoAlso, game theory doesn't imply zero-sum or repeated games. It can most certainly apply to "a non-repeated, non-zero-sum situation" as well.
- eru 4y agoAlso repeating (two-player) zero-sum games doesn't do anything: In zero sum games, you minimax and that's it. You don't even care what the other guy is doing. Whether you repeat or not, it doesn't matter.
- liquidsmoke 4y agoBefore the book with Morgenstein, Von Neumann single-authored a paper on game theory, which formalizes all of multiplayer reinforcement learning. Where Turing asks if a machine is intelligent, if it plays the Turing Game well, Von Neumann asks how such a machine should play to do well at the Turing Game, or any other parlor game for that matter. On another note, Von Neumann is surely put on a pedestal. I believe such sanctification to be natural. What is unnatural, is that in more recent years, whenever the name of Von Neumann is brought up, someone inevitably pops in the comments here to describe our hero of science as a genocidal maniac. On the MAD development, you have to realize that the USSR also had very smart mathematician minds working on this. So yes, it is as much a physical arms race as a mental/psychological one.