| scratchpad |
[Dec. 4th, 2002|09:09 pm] |
reading Exploring Randomness, by Gregory Chaitin.
try to capture the train of thought:
1. let's say i'm a turing machine in search of the right people. it's clear that there's a big strategy difference between looking for "real" friends, i.e. people who may share your goals/interests, and search for the right information sources.
2. therefore, the nodes are bits in the program when you decide whether to stop or go.
3. in this environment there should be a halting probability of finding the right people.
4. pretty much at each node i have to make a random decision on halting, or, as google does, have some weghing factor, based on the kind of stuff i'm looking for.
5. this is not classical random walk, but it's close. shit!!! i wish i new about it earlier.
also, an interesting point on Occam's razor as an information compression tool (p. 18) |
|
|
