there was a game i was watching a dude with 2k+ some elo aganist stockfish 8 and it was pretty common but the guy played very offensively so he lost a rook and got his pieces to the king site and then it looked like a normal game but then he somehow got down by only -1 piece and traded the stockfish's queen around step 30 for a rook i said wtf. i double checked if it was atomic or some antichess stuff 3 check but no it was a normal game i guess classic so wtf almost everybody says its impossible from a few years to defeat stockfish 8 for any humans so....i checked the level and it was 8 and i couldnt belive it he has like 100 days on the site not marked as cheater and he played few games before that he lost to comp and was thinking for like 1 sec and then for like 1 min but the who is cheating thinks between moves for a similar amount of time. so i belive the guy is not a cheater but how is this possible for him to defeat stockfish like this? he was 2200 centipawn above stock and in some 70 80 steps loll. is it possible? the guy played like alpha zero the engine (i think that engine is a hype anyways because its weak but it was running on some super computers and stock didnt have opening book enchantments aganist it and basically it s a weak engine but plays pretty well on a strong computer) is it possible for one to somehow copy the alpha zeros strategy and defeat stockfish like this? im amazed on the game what do you say about it.:
i know of people who are winning aganist stockfish by time in ultrabullet and winning in weird games like antichess and crazyhouse or cheaters but not in a common classical game. not by 2200 centipawn.
In chat, Soultego talks of playing this line vs SF many games and found SF misplays with 12. ... gxh6. He played several games and lost, even with the mistake by SF, before finding the best continuation. He then showcased this game giving the winning continuation (after one takeback.) A few other nice sacrifices are made. This is the type of opening, with the offered sacrifice which SF mistakenly took, that humans search for to defeat the silicon monsters.
I knew a line in the Smith Morra Gambit that the older programs mishandled. By repeatedly playing the game, I eventually found the winning line. Fritz would always play the same inferior move until a newer version corrected the mistake.
OP ... ? Suggest you learn the meaning of centipawn. You keep referring to the number "2200" for the player. The average centipawn loss was 15 for White. acpl for short. You questioned a half dozen times if the player was "cheating". I'd suggest not making such references , especially with your obvious lack of knowledge on the topic. Great game to post though. I'm betting a lot of players will be trying out this line vs SF and see if it also plays the losing move 12. ...gxh6
SF obviously is very confused here. It plays 11. ...h6 giving and evaluation of 0.4+ and after 12. Bxh6 (remains at 0.4+) and after gxh6? it gives 2.5+ for White. Does not make the best move by taking the Bishop. It's almost as if a button was pushed "play current move being analyzed" (the obvious one of 12. ...gxh6) and the engine was not allowed any further analysis to find 12. ...Ne4
so i meant centipawn like this: 1 pawn is +1 *stuff* on the board and idk how to say it if you are up by 1 point then u can see +1 and idk whats it so i said centipawn like +1 if u have 1 more pawn thats 1 pawn and 1 pawn is 100 centipawn witch is 1 pawn and the guy had +22 points with the 2 queens it said +22 so its like he had +22 pawns on the filed in wroth and stuff
yea i mentioned cheat ofthen its a pretty hot topic everyone is mentioning it cheaters may not get cheat sign and some players get cheater stuff even if tehy are not cheating
cheating is not unique stuff its just like an engine playing not a human so see stockfish doesnt think for 2 sec and then 200 sec it sometimes think for 1 sec very rarely for 3 sec and its a little bit off difference. so cheating is not a difficult topics i have knowledge on that topic. i guess stockfish analyses options witch are good so bad options are not analysed and stuff so if u sacrefise ur knignt queen and stuff the fish is not prepared for that so if u make moves that would look stupid the fish doesnt react to them as well as it for good moves. i seen only this guy beat the engine so far
Your conversation is impossible to follow. There is a definition for the term "centipawn." Your "made up" definition does not resemble the accepted meaning in any way, shape, or form.
The term is used in conjunction with another term "loss", giving acpl .... average centipawn loss, which is an evaluation of all moves made in a game.
Suggest using a spell check program. 1/3 of your terms are misspelled, the syntax is unreadable.
sorry about that
my english is bad you know xD
Did you call AlphaZero a weak engine? Wrong, mate
I just gave it a shot playing the exact same moves, and on the first try, it played the exact same line.
Although I had to throw my winning advantage away at the end because I sort of feared being reported hahahha
So funny that someone has actually figured out a way to beat it this easy :P
I guess this line is just so underdeveloped hahahhaaha.
BUT on a serious note, can someone answer this; I wouldn't get reported for remembering the game and playing it out against SF would I?
SF (AI level 8) doesn't care if you use assistance against it.
"2200 centipawn above stock"
Does this mean you added up all the centipawn losses in the game? Never heard of that.
