Hands off testing

[Mowers]

I would like to run hands off tests of the game, run it through 20-30 times and collate the results and make some observations. I have some suspicions on various balance elements that without some sort of stats would be meaningless.

I did this with a paradox game during beta and was able to identify a number of issues with the set up, it was a useful exercise.

I would like to do the same for this game now that its starting to become quite playable. What codes are used to allow a game to run hands off so that I can set it up in the morning on multiple computers and run it through to the end without any input from me.

Thanks
Mowers

[calvinus]

What do you mean for "hands off"? Only a AI vs AI playing?
If so, this is not allowed, sorry.

[Mowers]

What do you mean for "hands off"? Only a AI vs AI playing?
If so, this is not allowed, sorry.

Thanks, Interesting, How did you beta test the game without this function?

[calvinus]

I/we played hundreds of games...

[Hobbes]

I/we played hundreds of games...

I know the feeling

[Mowers]

I/we played hundreds of games...

You would save yourself an awful lot of time implementing a hands off tool Whilst it doesnt exclude the necessity for human games it does catch an awful lot of bugs and balance issues. Hundreds of games must have taken a considerable amount (months even) of your time and such a system may help narrow down the outstanding issues.

[Gray_Lensman]

deleted

[Mowers]

Not necessarily. Things like AI details and what if I do this what will result, etc. have to be tested on a turn-by-turn basis and each item analyzed for it's effect on gameplay. Hands-off-tools miss all the in between details and cause/effect relationships that is so necessary to a well designed AI. AGEod and Calvinus have done a fine job programming their games in their own manner and utilizing dozens, if not hundreds of dedicated volunteer testers, each looking for those performance issues that only a human can find in each turn's move.

In contrast, though a hands-off tool can generate many turns, the information gleaned is minimal at best when the designer needs to know when and more importantly why a particular move choice is made and the expected results.

Having beta tested a few games with a hands off tool I can say from experience that the information gleaned is not minimal at best. It's, as I said before, part of a bigger process but definitely worth including.

This particular peice of software may have benefitted from such an approach significantly. I dont know what went wrong in the development of this particular game so I say may rather than would.