How Hard Is It To Get A 'Perfect' Game?

@tcmfan #10 yes, low average centipawn loss over a long sequence of moves is a high indicator.

and also, yes, if the opp drops piece after piece, a short 0 0 0 game can be quite common.

And here comes a great example:

ru.lichess.org/IUYNW52I/white#38

Watch comp.analysis. Until I blundered my knight, I played perfectly vs 2619. And this stupid mistake is really of level 1000, not 1400. I can't imagine how I could do that most stupid move ever. I was not even under time pressure (which is number #1 reason of my mistakes usually)!

I just played one of these 0 0 0 games, only 8 centipawnloss. But it's only 20 moves and my play was very logical, just putting pressure one white's centre.

de.lichess.org/JoTXfy7V/black

@tcmfan #13 You deserved it, because you didnt jump at your opponent with the white pieces and instead played for draw :-p

Edit: also, after the opening you was worse by around 0.5 pawns, and that is by far not perfect play.

It's hard to answer because it depends very much of playing style. If you play positionally it's easier to get a perfect game than sharp attack playing. Also someone blundering early can result in perfect games.

In the context of cheating one single perfect game by occasion is not suspect but if someone plays like a lot of 0/0/0 games in a row or nearly perfect games I would be suspicious.

@blackzombie
I completely agree, in a positional game engine might consider your play as almost perfect, as long as you just play decently, while a GM would just laugh about your "perfect" play. tcmfan gave a nice example, he played very decently against a much stronger opponent, still it's clear that white was slightly worse early on and no GM would play like this. So how do we define "perfect"?

If "perfect" means stockfish didn't suggest better at some move, especially in CC, then doesn't mean much. Especially in a game where your opponent made a series of nasty blunders - you and every engine in the world will make a move to take advantage of the blunder.
If cheating is suspected, it's more (I would guess) where it plays some move that wins a pawn or position 5-8 moves out, but really doesn't change the outcome - real people miss that because the see a forced win and play accordingly.