[Kde-games-devel] New AI - direction to making some games more "serious"?

[Ian Wadham]

On Sat, 26 Jul 2008 02:17:57 am Dmitry Suzdalev wrote:
> Today I was approached by Inge on IRC and he wanted to know if kdegames
> would be interested in improving some games' AI to be more professional.
>
> For those of you who don't know Inge - he maintained KReversi and serveral
> other games and made numerous contributions to kdegames module in its kde3
> times :)
>
> Inge said that he has some well-crafted AI code at hands and he is willing
> to share it with kdegames if we all are interested in that.
>
A warm welcome back to Inge!  And it's very nice to see you again, Dmitry!

> Personally I think that we should try Inge's AI and try to make
> "professional vs newbie" be configurable.
>
I agree completely.  Yes we should have tough engines available, to retain
players' interest and enthusiasm, but also weaker engines are essential to
get people over the beginner's phase and generate that interest and
enthusiasm in the first place.  I really would like to see a "dumb" level
in KReversi, so that I could win sometimes ... :-)

Ideally, I think every KDE Game should be playable at some level by a
child of four or five ... and I have a couple of guys (grandchildren) just a
year or two away from starting to train ... :-)

AI also applies to hints, I would suggest.  A game that gives you good hints
could be a good teacher.  Some of our games could surely give better hints.

All the best, Ian W.

P.S.  Re games I maintain: KGoldrunner's "AI" is *deliberately* dumb.  The
difficulties lie in the level layouts themselves and how to exploit the
foolishness of the enemies.  So no changes there ...

But I intend to improve the AI in KJumpingCube, even if it is just to use
the strong corner and edge squares more and spread out the computer's
opening moves.

I would also like to explore algorithms for solving Kubrick.  Rubik's Cube
algorithms are a developing area on the web.