Randomized surfing

Fragment of a discussion from Talk:Phantom
Jump to navigation Jump to search

> If one is using a traditional "orbit until selected GF" method for instance, as soon as you're stopped or begin to decelerate, you've given away exactly where you intend to be for that wave.

Disagree with that. If I decelerate, that means full stop or reverse will take me to a safer point than full forward. But things could change next tick and I'll go back the other way, and again after that. I wouldn't say "orbit until selected GF" is an accurate summary of the "traditional" surfing algorithm.

My latest attempt at random surfing actually randomized the stats that my normal surfing algorithm works from, so it shouldn't have been subject to any tells re: what path I take to get to the "target/random GF", because that didn't really exist.

Voidious (talk)19:05, 11 November 2013

I have been thinking about random movements, but I was not able to devise an algorithm which have a linear displacement grow with time. Every time I put, roughly speaking, a choice of stop, back, forward. I see that bot perform diffusive/brownian type motion which as math teaches has displacement grows as sqrt of time. So the simplest of all guns: head on has the highest chance of success. From other hand it is very unlikely, that such bot will reach the ends of the guess factor ranges. Which yet again make it easy target due to limited ranges in a GF metric.

My current bot has such randomness buil in, affected by the danger map, yet it still likes to hang in one place, which makes it an easy target to GF guns (I think) and may be even DC guns.

Beaming (talk)04:45, 12 November 2013