Re: Open Source Games and Cheating - a paradoxum?
On Thu, Jan 23, 2003 at 08:59:38AM +0700, Robert Lemmen <robertle@semistable.com> was heard to say:
> On Tue, Jan 21, 2003 at 04:20:05PM -0500, H. S. Teoh wrote:
> > Technically speaking, there is no way to tell if a remote client is a bot
> > or just a really, really talented and dedicated gamer. (OK, there are
> > heuristics to check for blatant signs like microsecond think-time, which
> > may exhibit itself as "jittery" movement. But every heuristic you come up
> > with can be defeated by a smart-enough AI.)
>
> according to turing a bot that you can't tell from a human is human
> enough to have the right to take part in the game, innit?
Perhaps, but can you write a computer program to test that? :P
One thing I read recently (in the NYT) is that someone -- Yahoo,
perhaps -- is working on a login system designed to defeat webcrawlers.
When you log in, it gives you a problem known to be easy to
generate/check on a computer, but hard to solve: for instance, an image
containing several overlapping words.
You could probably do something like that to stop a bot from logging
in automatically. In fact, I almost mentioned it myself in another post
on this thread, because it's a great example of an inadequate
countermeasure.
Think about it -- if it's just a login test, the human setting up the
bot just has to answer the question for the bot. The only way to really
screen computers out would be to pop up the question on a regular basis,
annoying every legitimate player. And this would only catch bots
running unattended -- if a human is keeping an eye on the game, they
could answer the question for the computer. (also, this doesn't catch
cheats that merely improve the human's ability, rather than cutting them
out of the loop entirely or breaking the rules -- eg, giving them more
accurate aim)
Daniel
--
/-------------------- Daniel Burrows <dburrows@debian.org> -------------------\
| Who is General Failure, and why is he reading my hard drive? |
\---------------------- A duck! -- http://www.python.org ---------------------/
Reply to: