February 18th, 2006
cartesiandaemon

2006/02/19 00:54:02

Hmmm. Probably some sort of cost function is desired. For instance, suppose the picadilly route takes time T+w1 (expectation T+E(w)=T+3) where T is a constant and wN is a random variable describing the time taken waiting, say w~N(3,1). Then the other takes, um, T+2 (expectation T4+2E(w)). The exact numbers don't matter.
The point is, if the train leaves in T+2, you probably have a better chance with the alternative route. If the train leaves in T+4, you may have a better chance with picadilly, because the variance is less.
Certainly, working all this out wasted longer than I'd ever save :)
PS. Ken MacLeod  would you recommend any starting book other than cosmonaut keep?

senji

2006/02/19 17:00:39

Well, I started with The Star Fraction (and I certainly wouldn't start with any of the other of that err quartet), or if you prefer something that pokes not entirely subtle fun at the genre then Newton's Wake might suit.

senji

2006/02/19 17:04:10

The point is, if the train leaves in T+2, you probably have a better chance with the alternative route. If the train leaves in T+4, you may have a better chance with picadilly, because the variance is less.
Personal observation of the Piccadilly is that it has quite a high variance of its own – probably because it has more stops in the shoppingandtourist areas of Central London, and hence more opportunity for passengerinduced delay.

cartesiandaemon

2006/02/20 17:44:21

Ah. Damn the insufficiency of statistics.
(Thanks for rec's.)


