The position below is (with colors reversed) from analysis of a game I played 1,5 years ago. Back then, me and some others tried to find out if white could have won, if this position arose. It turned out it is just possible with the right plan, and that with the right plan there's no doubt it's a win. It involves tempo play and creating a winning plan, and of course a good knight vs. bad bishop ending.

What do you think? Too hard? Or maybe not suitable for this feature?

i think this is a very interesting position! can you include the plan? can a computer understand the plan?

I don't think a computer can find the plan. At least my Fritz didn't find the win. Only when you "help" him by going through the plan, at some point he realises it's actually a winning plan.

The solution consists of using the knight and king to advance the e-pawn to e5, then put the white king on e4, and get the knight back to d5 when black is in zugzwang. It takes like 10-15 moves to execute the plan and force the win of one of black's pawns.

so can you win this against a computer every single time?

Yes, certainly. I'm sure it wins, and there's hardly anywhere to go wrong, once you know the plan as white.

However, like I said, a computer will not see the winning variation. So when you play 1. Nc3 it'll likely want to tell the user he's making a bad move. Is that a problem?

the only problem is if you can't always win against the computer ;)

I forgot to mention the "triangulation" part. 11. Ke3! Bd8 12. Ke4! Be7 and now suddenly the knight has just enough time to go to d5: 13. Nd2 Bd8 14. Nb1 Bc7 15. Nc3 and now if 15... Bxe5 then 16. Nd5 winning the b-pawn. 15... Bd8 16. Nd5 leaves black in zugzwang, while 15... Bb8 16. Nd5 Ba7 is just too desperate to hold; 17. Nf6 (threatening 18. Nh7 to win g5) 17... Kf7 18. Kf5 is an easy win for white then.

{ 462MB, Rybka3.ctg, Home PC  }

Like I said, computers won't find the solution. So it's even a cheat-proof position!

that's comlicated

Maybe it's complicated, but once you found and analyzed the winning strategy, you will have no doubt that it's winning. So it's only complicated for those who haven't found the winning plan yet :)

My chess engine found this plan. Very nice position, this is how it went when my chess engine played it out:

One problem with a position where computers can't find the solution, is that the move by move feedback and "Best Move" button don't really work.

Will it is not always true.

Awesome puzzle!