By the way, in case anyone doesn't know, it is TOTALLY cheating to use an endgame tablebase to analyse active games. And playing a perfect endgame with a rating of 1700 or whatever is extremely suspicious. I rarely play a perfect endgame.
It's ok to analyse arbitrary positions, assuming it's also ok to use an opening database. Most blitz games, it would not be allowed, but most daily games it would be.
Black to move, white mates in 297
Practically it doesn't matter if it is theoretically won or drawn when neither the attacker or defender will play perfectly. If it is a close case, such as 2 pawns vs 1, then the evaluation is likely to flip several times.
It is sort of amusing when you hear stuff like "mate in x" when clearly a human would not think of the positions as how many moves are required to checkmate.
Kh6
but I can't play that.
haha sorry about that, my fault.
I'll have to study all the comments about this game later. For now...
65...Qf6+
Vick has failed to achieve a DRAW!
Ki5
Maybe, but not proven yet.
66...Qe5+
Kj6 
Definitely.
67...Qh8+
i7
68...Qxh4+
Ki6
69...Qf6+
I think this is a draw, if I play correctly. But I'm not 100% sure.
Kh5
Here's a useful endgame analysis tool...
http://www.k4it.de/index.php?lang=en&topic=egtb
It allows up to six pieces, so you can analyse all K+Q+P vs K+Q endgames. If you start looking at them, you notice that most are drawn... usually the attacker can only win if he can trade off queens while keeping the king close to the pawn.
It's very difficult to turn a pawn advantage into a win with just a queen. The more pawns, the easier for the attacker to win, because it's harder for the defender to find perpetuals and it's easier to fork king and pawn. 3 vs 2 is too close for me to call, it could be a win by force for white but he needs to be accurate to avoid giving you drawing chances.