I noticed a problem that may be already known but can be annoying: it is possible to use the network in our IA (in Rift Platinum for example). I did the test with sockets in Java, I was able to mount a client / server that communicated with my PC, including the arena. It’s probably the same with other languages.
One can imagine a way to cheat his opponents by identifying and acting accordingly. Or deport the calculations to a more powerful PC …
In general I think it is better to prevent network connections.
Whoa! Is this really possible?
Because if this is the case, then cheater AIs can be developed, which help each other to get higher ranks in multiplayer battles!
That’s too bad… I’m starting to think that downloading source code from a remote site and running it dynamically is the only way to get a decent score in the optimisation puzzles!
I hope you’re joking because as of now, optimization puzzle are easy, and it’s not that hard to get a decent score with some research in order to improve your code.
If you need specific tricks in C/C++ i’ve already made a topic for golfing in c++, it might be useful.
Half joking here, in fact. My (Haskell) code is not that bad in itself, but till… Perl entries and alike seems to play in a whole another (space) dimension. Even if I remove all technical stuff such as the import directives and magically reduce every libray call to a single character… I still get 200 characters more than the top ranked! Since I don’t think that Perl has a special function whose name is a single japanese kanji roughly translating to “convertCloneInputStringToActionList”, I’m trying to think out of the box now
points are awarded by language, so you don’t need to be first of every language to gain points, just be first in haskell. If haskell doesn’t allow magic tricks, the top competitor in haskell won’t be able to neither.
