if grandmasters critique computers

style of play, saying they make this move such and such, why cant they beat the computer then.

Because one big mistake will lose the game even if 50 minor mistakes are made by the other side, and humans don't have the near-perfect consistency to avoid big mistakes like computers do.

So computers can play strategically questionable (or bad) moves, but to exploit a few small mistakes, a human will have to play the rest of the game making no mistakes bigger than that.

Games in the ICCF try to take advantage of both types of strength. The human guides the game with their understanding (no calculation needed to know some moves are not good) and the computer checks the (relatively) short range considerations by calculating.

That would have been a good opportunity to offer the GM to play the engine from that position.

Of course he may have offered you take lessons from your computer if you trust it so much ;)

(Sometimes moves are bad from a practical standpoint.)