On Thu, 2008-12-18 at 10:11 -0800, Kerim Aydin wrote: > Just goes to show great minds think alike while fools etc. Pondered an > arrangement just like this but worried about exactly that degeneracy (and > others). I thought I'd drop this fundamental corewars part: > "The data buffer for each of them is the opponent's program buffer." > I was thinking of instead doing a version where code buffer and stack > was protected/not shared, but the data array was shared. The object > would be (maybe) a type of capture the flag... each program has a > flag memory cell that zeroing would cause a loss, with an additional > loss for your pointer running off the end of the array. Ties would > have to be the same as losses to protect against the "just guard your > flag" over-defensive strategy. Could be interesting... in that case, most likely a two-dimensional grid would be better, and the programs should have methods to locate and shove around their opponent's data pointers. Maybe some sort of radar to detect flags and pointers, with pointers being easier to find? Again, though, it might be degenerate, unless we allow multithreading.
Actually, this is starting to remind me of the 2004 ICFP contest, <http://alliance.seas.upenn.edu/~plclub/cgi-bin/contest/ants.html>, although that's a lot more heavy-weight than what we'd be seeking to do here. (It had teams of ants moving around a grid, using their senses to grab food and detect and kill rival ants.) -- ais523
