It seems unlikely to me that scribd are actually looking for a solution to this problem, or an isomorphic one. Have you any basis for thinking that?

There are many standard approaches to the general class of two player board games (they are clearly aware of these, with an alpha-beta based bot).

I would guess that what is going to differentiate winners from losers in this specific game are the exact specifics of the setup. So if they are looking for a solution to an isomorphic problem, it must be a very similar problem - as the general type of challenge is well understood.

It just looks like a neat game to me; I doubt its a cleverly disguised problem from 'social reading and publishing' that they are trying to get a cheap solution to..?

That doesn't really make any sense. Scribd is trying to trick you into solving an undergrad-level AI problem for them? Sounds unlikely. Designing and writing the game probably took more effort than writing a bot would!

Coincidentally, I really like the game design. I took an AI course recently and we had some really similar using Pacman and Python, and that was fun too, but I would much rather use JavaScript and compete online than use Python and the annoying framework code they gave us.

Your comment is totally removed from reality. Did you even read the page? Do you think that Scribd is in the business of picking up fruits from grids?

Wouldn't be so quick to down-vote his comment. I don't think scribd is stealing code or that submissions will be directly applicable, but there's some cross-over between a game like this and searching/crawling a db/docs/network. It's a great way for them to find devs with skills they need.