Have engines refuted any established openings?

Can you name some? Not looking to argue, just curious.

This refutation is known as the Fischer Defence. It is not actually taken seriously or played by today's grandmasters. The original story behind this line is that Fischer and Spassky were once playing a tournament and Spassky (a well-known King's Gambit practitioner) beat Fischer using the KG, winning first place. Fischer became so angry with himself that he tried developing a refutation. This article is not actually from 2015 but was originally written in 1961, long before engine analysis became available.

I think the 3. ... Qa5 Scandinavian might be an example. The Scandi has never been treated with much respect and doesn't get played much but in many lines, Black has a very playable position and engines don't really see all that much wrong with it.

Thanks.

Do you spy on everyone's opening preferences?

I kinda get the feeling that not everyone in this thread defines "refuted" the same way ...

@29

refute = to prove to be false by argument or evidence (Webster)

@26

But you cannot claim it as a refutation to the Kings Gambit.  Even if you could prove a forced win after 1.e4 e5 2.f4 exf4 3.Nf3 d6, which I highly doubt you can as you have never proven such capabilities before and at least half of what you say is outright wrong, that does nothing to claim the Kings Gambit being refuted, it would only refute 3.Nf3.  You've still got 3.Bc4 and 3.Nc3 to deal with!

The Kings Gambit is crap and highly dubious, but you cannot claim refuted until you. An prove it!  Same goes for other dubious openings like the Latvian Gambit, the Fajarowicz, the Budapest, the Alekine, the Scandinavian, the Elephant Gambit, 5...Nxd5? In the two knights defense, etc.

One game that Black wins is not proof.

A lot of evidence is circumstantial, which can be thrown out, just like in a court room!

It would be like saying that 1...e5 (Black's best response) is refuted against 1.b4 because of one game where I smashed my opponent in 17 moves this past weekend over the board.  That would be one of the dumbest things you can possibly say.

In the laws of statistics, you do not have a valid sample size until you have at least 30.  Do a search in your database for the position after 1.e4 e5 2.f4 exf4 3.Nf3 d6, nothing else.  No year range, no specific results, no rating minimums or maximums, nothing but that position.

Now, take your results, and find me 30 consecutive games where the result is 0-1.  I would bet good money that there are not 30 consecutive 0-1's in that list!  We are talking things like Megabase, not skewed databases that come with a video on beating the Kings Gambit.

I would personally define "refuted" in Chess as follows:

• The opponent is able to get a +2 advantage in all variations (I choose +2 as that's generally an advantage large enough to win the game. Adjust this as you want.)
• There is little chaos or uncertainty in the positions where the +2 advantage is established (example: the opponent is up two pawns and you have no compensation, and the opponent does not need to play a long chain of "only moves" (or anything like that) afterwards to keep the +2 advantage)
• The refutation is forcing and/or comprehensive (i.e. you do not have the option to play something differently or the refutation covers any alternatives that you have in mind)

By my criteria, I do not consider the King's Gambit to be refuted as that opening is such a mess, however I do believe that certain simpler openings like the Colorado Gambit can be refuted under these conditions.

I think that you are reading my comment wrong. I do not claim that the King's Gambit is refuted, rather I was explaining to another commenter that the Fischer Defence is not considered a good refutation of the King's Gambit.

+2 is a very high bar. For most positions, 1.5+ is plenty enough.

the problem of definition always plagues talk of refutation, i agree some people have very narrow (and if you ask me dumb) notions of refutation. Some people argue that any white opening thats even and not a clear slight advantage is refuted (like the word suboptimal doesnt exist).

but the purist definition of forced loss (or clearly forced loss despite being beyond calculating horizon) is too narrow and would only apply  to mostly gambits and very dubious stuff, almost all from black. 

Here is my pragmatic definition of it:

"A refuted opening is one where there is either a  Forced loss down the road OR the path to a draw is long arduous,, unclear and unpleasant. "

in practice this means among two reasonably strong opponents , player 1 will never play refuted opening X agaisnt player 2 if player 1 knows player 2 knows refutation of X. if they are unbiased 

(i add the last part, because there is a player on lichess i have played over  40 times in 15 10 , that has played the sniper 1.e4 g6 2.d4 bg7 3.nc3 c5 10 times agaisnt me , and i show him that all 3 lines after 4.dxc5! practically lead to a refutation, our record is 9.5-0.5 and he still keeps trying to play it agaisnt me. I guess that one draw gave him hope. my memory isnt perfect)

Sometimes, some daring players purposely play refuted lines sparingly knowing the odds their opponent will know that specific refuting line is slim to none. I recently recall seeing  a rapport  game with the 3.nc3 variation of the king's gambit agaisnt a GM where the GM played the refuting line accurately until like move 14, then blew the next move, and that gave enough room for a comeback and Rapport won.Under my definition, its still a refuted line, Even Rapport isnt crazy enough to do an encore agaisnt the re-booked traumatized opponent.

@39

"+2 is a very high bar. For most positions, 1.5+ is plenty enough."
++ GM Larry Kaufman considered +0.7 a win.
The logic is that 1 pawn is enough to win: queen the pawn.
'Other things being equal, any material gain, no matter how small, means success' - Capablanca

@34

"One game that Black wins is not proof."
++ It is evidence. Then the burden is on white to find an improvement.

"1...e5 (Black's best response) is refuted against 1.b4 because of one game where I smashed my opponent in 17 moves this past weekend over the board."
++ Then the burden is on black to find an improvement.

"Do a search in your database for the position after 1.e4 e5 2.f4 exf4 3.Nf3 d6, nothing else."
++ Black wins 43%, draws 29%, white wins 27%. That is proof.

"No year range" ++ There should be a year range. If an opening is refuted, then all games before the refutation are irrelevant.

"no rating minimums or maximums"
++ There should be a rating minimum. The outcome of a game correlates more with the rating difference than with the opening. Games between lower rated players or in fast time controls contain so many errors, that the influence of the opening is negligible.

"find me 30 consecutive games where the result is 0-1" ++ No, that is not a way of proving.

https://www.chess.com/forum/view/chess-openings/have-engines-refuted-any-established-openings?page=2#comment-73377799

I mean +0.7 is very much varying from position to position, are we a pawn down? Is the problem structural weaknesses or lack of developement, 0.7 with an NNUE engine would give whote 35% winrate against an equal opponent, now at the GM level, if they know what they are doing, this is probably enough, but the opponent often times needs to be booked up, and so espessialy for surprise weapons or aggressive gambits 0.7 is perfectly playable in my opinion (classical chess ofc)

not all 0.7 are created equal, they are some real comfy 0.7 and then they are some 0.7 that are grueling and even a 1.1 in the right complex position is better than the dried out 0.7 with no winning chances. 
also there has been some computer inflation we need to address. in the older days, engine evaluations around 0.6-0.7 is when they start getting real concerning (usually where 1.e4 a6 2.d4 b5, started being evaluated), 0.6-0.7 on stockfish 16 is not that bad at all, i usually begin worrying around 1.1-1.2 just on pure eval and even then if its one computer line virtually no human will play, according to database. i may let it slide. They are some lines that are 0.8 i woudnt touch though just on taste, and realistic chances and other preferences. 
tldr: 0.7 is not what it used to be, 0.6-0.7, now is more like 1.2-1.5 depending on the position

Yes, I agree. Results of chess games can be considered as a form evidence, not proof. I think that they can also inform about the human element of the game: some positions can be completely equal according to a computer but still be hard for a human.

Current opening theory is mostly about evidence, not about proof in the strict non colloquial sense of the word which is reserved to mathematics and logic. There's proof for example for different endgames that have been solved and for example positions that lead to a forced checkmate and have been shown to do so. If you lose a rook in the opening with no apparent compensation, it doesn't prove your position is losing in the strict sense. I would of course believe that it's losing in light of the material difference.

Completely agree.