Will computers ever solve chess? - Chess Forums - Page 189

godsofhell1235

Mar 21, 2018

0

#3761

ponz111 wrote:

godsofhell1235 wrote:

blacktower01 wrote:

never play 4.e5 on 3.Nf3

After 3.Nf3 Nf6 the most common move in my database is 4.e5 which has been played by many different GMs.

It may not be the objectively best move, but KG players aren't too concerned with this. They typically want an objectively borderline lost position, with a big white initiative and chances for both sides.

I am using the 2. ...Nf6 sequence as an attempt to refute that line of the King's Gambit. The current analysis of this line appears to be not very good. [i know the response is often 3. e5 by some strong players.]

But if white plays like this there should be no refutation I think?

godsofhell1235

Mar 21, 2018

0

#3762

USArmyParatrooper wrote:

Also, let’s say a human (or Stockfish) has forced mate in 153 moves, but doesn’t see it and instead plays a different move.

Was that an error?

The way we've been talking, if it spoils the win, then yes it's an error
If it's still a forced mate, but it takes more moves, then that's a bit gray
If it's a draw now, but in practical play it's almost certainly still a win, then even more gray
If it's a loss now, and the opponent has a forced mate, but it's almost certainly still a win in practical play...

etc etc.

I'm open to using whatever definition you want.

lfPatriotGames

Mar 21, 2018

0

#3763

USArmyParatrooper wrote:

Also, let’s say a human (or Stockfish) has forced mate in 153 moves, but doesn’t see it and instead plays a different move.

Was that an error?

I would say that's about the same as if there is a forced mate in 1, and it's not seen and a different move is played. Both would be errors in my opinion. If there is still a forced mate in 154 moves then at least it's not a critical error, but still an error because it's not best play. And I assume best play is what's needed to figure out if computers will ever solve chess.

vickalan

Mar 21, 2018

0

#3764

btickler wrote:

...Either way, it's pathetic.

If you don't agree with their conclusion, it might be better if you comment there rather than here. The forum at the StackExchange is subject to peer-review. This forum is not moderated by experts. You're not going to be able to change their conclusion by adding comments here. I'm sure they don't read this.

USArmyParatrooper

Mar 21, 2018

0

#3765

godsofhell1235 wrote:

USArmyParatrooper wrote:

That doesn’t at all conflict with what I said. I think there’s a miscommunication somewhere?

So you're saying the player's moves may be perfect (if they never change Hal's evaluation).

And you agree that if every move in the game is like this, then it's a perfect game.

But you also say "[we] cannot say anyone of those players played perfect chess" (if Hal would have beaten either of them).

---

When you say if Hal would have beaten either of them, what I'm imagining is you're saying during the game we substitute Hal for one of the players, and now Wesely So, or Carlsen, is playing Hal instead of the human opponent he drew with.

Do you mean something else when you say if Hal would have beaten either of them?

No. I find it incredibly unlikely that in these games Hal would not have found a forced win anywhere along the way.

USArmyParatrooper

Mar 21, 2018

0

#3766

godsofhell1235 wrote:

USArmyParatrooper wrote:

Also, let’s say a human (or Stockfish) has forced mate in 153 moves, but doesn’t see it and instead plays a different move.

Was that an error?

The way we've been talking, if it spoils the win, then yes it's an error
If it's still a forced mate, but it takes more moves, then that's a bit gray
If it's a draw now, but in practical play it's almost certainly still a win, then even more gray
If it's a loss now, and the opponent has a forced mate, but it's almost certainly still a win in practical play...

etc etc.

I'm open to using whatever definition you want.

Then there’s no way anyone can assess if a perfect game has ever been played. At least not until chess is solved.

IM pfren

Mar 21, 2018

0

#3767

The 3.Nf3 Nf6 King's Gambit line is thorougly analysed in two recent repertoire books.

https://www.everymanchess.com/opening-repertoire-the-open-games-with-black

https://www.qualitychess.co.uk/products/1/267/playing_1e4_e5_-_a_classical_repertoire_by_nikolaos_ntirlis/

Apparently the 3...Nf6 line is a serious threat to the gambit. My suggestion is to play in pure speculative King's Gambit style 4.e5 Nh5 5.Qe2!? which is not considered in both books. It was labelled as bad since 1942, after Keres lost a game with it against Alekhine, and the computer initially claims a large Black advantage, but careful analysis reveals that 1. things are far from simple, and white does have serious compensation, and 2.the whole position is a wonderful mess.

After all the stupid-looking 5.Qe2 is positionally consistent: it prevents a quick ...d6 which is thematic, and it also has a couple of well hidden tactical points.

USArmyParatrooper

Mar 21, 2018

0

#3768

By the way, my personal preference for “perfect” is literally the strongest moves. So, let’s say Hal is playing white, see no mating nets, but calculates 5 different moves lead to a draw with best play from black. Hal will pick the “strongest” of them, meaning the one with the most branches of moves that lead to a win ( The one that gives black the most opportunities to blunder).

I just conceded a slightly lesser version for the sake of argument, because I feel it changes nothing.

Nobody, including ponz111 can possibly know if a full and complete game was played perfectly.

ponz111

Mar 21, 2018

0

#3769

USArmyParatrooper wrote:

By the way, my personal preference for “perfect” is literally the strongest moves. This is ambiguous as hell.

So, let’s say Hal is playing white, see no mating nets, but calculates 5 different moves lead to a draw with best play from black. Hal will pick the “strongest” of them, meaning the one with the most branches of moves that lead to a win ( The one that gives black the most opportunities to blunder). This is also ambiguous as hell. One would have to know the other player better than that player knows himself to pick out a move or moves which would give him the most opportunities to blunder.

I just conceded a slightly lesser version for the sake of argument, because I feel it changes nothing. Actually it changes a lot. You changing the definition of "perfect" means you have decided to talk past the debate--it is a form of equivocation. Equivocation is a logical fallacy where a term is used in 2 or more different senses within a single argument.

Changing an agreed definition of "perfect" means you can no longer legitimately claim that perfect games are rare as you know another definition of "perfect" was agreed on.

Nobody, including ponz111 can possibly know if a full and complete game was played perfectly. Not by your changed definition of "perfect" as for sure we cannot read the player's minds to figure out what moves would be most likely cause an error.

However under the agreed on definition of "perfect" i can and do claim i have played a game perfectly.

By the way, under your definition of "perfect" even if chess were solved--a perfect game could never be played by humans.

godsofhell1235

Mar 21, 2018

0

#3770

USArmyParatrooper wrote:

godsofhell1235 wrote:

USArmyParatrooper wrote:

That doesn’t at all conflict with what I said. I think there’s a miscommunication somewhere?

So you're saying the player's moves may be perfect (if they never change Hal's evaluation).

And you agree that if every move in the game is like this, then it's a perfect game.

But you also say "[we] cannot say anyone of those players played perfect chess" (if Hal would have beaten either of them).

---

When you say if Hal would have beaten either of them, what I'm imagining is you're saying during the game we substitute Hal for one of the players, and now Wesely So, or Carlsen, is playing Hal instead of the human opponent he drew with.

Do you mean something else when you say if Hal would have beaten either of them?

No. I find it incredibly unlikely that in these games Hal would not have found a forced win anywhere along the way.

Yes, well, when it's symmetrical and boring there's not much to do even if you are a supercomputer

godsofhell1235

Mar 21, 2018

0

#3771

USArmyParatrooper wrote:

godsofhell1235 wrote:

USArmyParatrooper wrote:

Also, let’s say a human (or Stockfish) has forced mate in 153 moves, but doesn’t see it and instead plays a different move.

Was that an error?

The way we've been talking, if it spoils the win, then yes it's an error
If it's still a forced mate, but it takes more moves, then that's a bit gray
If it's a draw now, but in practical play it's almost certainly still a win, then even more gray
If it's a loss now, and the opponent has a forced mate, but it's almost certainly still a win in practical play...

etc etc.

I'm open to using whatever definition you want.

Then there’s no way anyone can assess if a perfect game has ever been played. At least not until chess is solved.

We've already agreed on this point.

But I still say it's likely for the reasons that by now I've given multiple times.

godsofhell1235

Mar 21, 2018

0

#3772

pfren wrote:

The 3.Nf3 Nf6 King's Gambit line is thorougly analysed in two recent repertoire books.

https://www.everymanchess.com/opening-repertoire-the-open-games-with-black

https://www.qualitychess.co.uk/products/1/267/playing_1e4_e5_-_a_classical_repertoire_by_nikolaos_ntirlis/

Apparently the 3...Nf6 line is a serious threat to the gambit. My suggestion is to play in pure speculative King's Gambit style 4.e5 Nh5 5.Qe2!? which is not considered in both books. It was labelled as bad since 1942, after Keres lost a game with it against Alekhine, and the computer initially claims a large Black advantage, but careful analysis reveals that 1. things are far from simple, and white does have serious compensation, and 2.the whole position is a wonderful mess.

After all the stupid-looking 5.Qe2 is positionally consistent: it prevents a quick ...d6 which is thematic, and it also has a couple of well hidden tactical points.

Eh, the king's gambit is such a headache.

I had a coach tell me to play 2...Bc5 and don't try to compete with the theory of the KG player.

I'm not sure what I want to do. I think I've only faced it once in my last 100 games so I guess it's not so important.

5.Qe2 is interesting though. I like to have interesting games in blitz with either color.

USArmyParatrooper

Mar 21, 2018

0

#3773

godsofhell1235 wrote:

USArmyParatrooper wrote:

godsofhell1235 wrote:

USArmyParatrooper wrote:

That doesn’t at all conflict with what I said. I think there’s a miscommunication somewhere?

So you're saying the player's moves may be perfect (if they never change Hal's evaluation).

And you agree that if every move in the game is like this, then it's a perfect game.

But you also say "[we] cannot say anyone of those players played perfect chess" (if Hal would have beaten either of them).

---

When you say if Hal would have beaten either of them, what I'm imagining is you're saying during the game we substitute Hal for one of the players, and now Wesely So, or Carlsen, is playing Hal instead of the human opponent he drew with.

Do you mean something else when you say if Hal would have beaten either of them?

No. I find it incredibly unlikely that in these games Hal would not have found a forced win anywhere along the way.

Yes, well, when it's symmetrical and boring there's not much to do even if you are a supercomputer

In that case, go up against Stockfish just play symmetrical and boring. And by the way a computer that has solved chess would be way, way beyond anything we can conceive of as a supercomputer.

USArmyParatrooper

Mar 21, 2018

0

#3774

And the comprehension issues continue...

edit: Not you goh.

godsofhell1235

Mar 21, 2018

0

#3775

USArmyParatrooper wrote:

In that case, go up against Stockfish just play symmetrical and boring. And by the way a computer that has solved chess would be way, way beyond anything we can conceive of as a supercomputer.

As we've covered before it's when both players choose to play this way.

And I've beaten SF with knight odds in a G/20. It wasn't hard when it let me play a french exchange and trade everything on the e file.

When it played something not retarded I was slaughtered.

godsofhell1235

Mar 21, 2018

0

#3776

(On my present day computer that might not work though, but anyway, not a big deal)

DiogenesDue

Mar 21, 2018

0

#3777

vickalan wrote:

btickler wrote:

...Either way, it's pathetic.

If you don't agree with their conclusion, it might be better if you comment there rather than here. The forum at the StackExchange is subject to peer-review. This forum is not moderated by experts. You're not going to be able to change their conclusion by adding comments here. I'm sure they don't read this.

Sorry, but it's not my job hold everyone's feet to the fire planetwide. I'm only doing that here, in this thread.

That forum is certainly not moderated by "experts". It is moderated by volunteers and/or entry level developers. Nobody is paying a 6-figure salary to to any expert developers to moderate a silly forum that is making its revenue off of Google ads, because that thread you posted, based on number of views, has only made stack exchange, at best, about $3. Not really a good business model. It also is not "peer reviewed"...not by any reasonable or rigorous definition of the term. Random Internet readers and posters does not constitute "peer review".

Also, your quote is misleading, as usual. The term "pathetic" was clearly referring to you and your chosen usage (mis-usage, more accurately) of this link. Since I has already quoted the main poster's conclusion, without disputing it, your first statement/phrase is at best incompetent, in that you don't even understand what's going on (certainly not something we can rule out with your history here), and at worst willfully misleading and unethical, in that you know full well that there actually is no "conclusion" or consensus on that thread. You are just pretending that it agrees with your stance, like you pretend everything else here.

USArmyParatrooper

Mar 21, 2018

0

#3778

godsofhell1235 wrote:

USArmyParatrooper wrote:

In that case, go up against Stockfish just play symmetrical and boring. And by the way a computer that has solved chess would be way, way beyond anything we can conceive of as a supercomputer.

As we've covered before it's when both players choose to play this way.

And I've beaten SF with knight odds in a G/20. It wasn't hard when it let me play a french exchange and trade everything on the e file.

When it played something not retarded I was slaughtered.

We’re about to go in circles.

troy7915

Mar 21, 2018

0

#3779

The lack of logic in Ponz’s claim that 1. e4 e5 2. Nf3 Nc6 3. Bb5, draw agreed is a perfect game is rather obvious, for he starts from an assumption which he presents as a fact.

He starts with the assumption that the perfect game ends up in a draw, with best play for both sides. Why? Because ‘ most GMs tend to agree on that’ ( who cares?! ) , ‘ personal experience’ ( who cares ?! ) and ‘other factors’ .

This assumption that a perfect game is a draw is presented as a fact. Then, of course, it doesn’t matter if the draw is achieved after 10,000 or 3 moves. This is his logic behind presenting those 3 moves with an agreed draw as ‘perfect’.

The whole proposition is founded on this assumption, that a perfect game is a draw.

But needless to say, nobody knows that. At some point, even he says, ‘ a perfect game is most likely a draw.’. Which brings out certain doubts from remote corners of his mind. But doubts or not, the foundation of his claim is a fallacy: nobody knows what a perfect game ends up in.

Now, starting from this new fact, that a perfect game can actually end up in any of the three possible outcomes, that beginning of the Ruy Lopez suddenly loses its previous strength as a perfect game: now it has 30% chances of being elevated to the ‘perfect’ status. Potentially, it can end either way, perfect or imperfect. But we are not labeling games based on their potential.

Because it cannot be proven that a perfect game is a draw, the first three moves of the Ruy Looez, draw agreed upon, are not, at the present moment, ‘perfect’. As of now, they are ‘imperfect’.

The same goes for 1. e4 e5, draw agreed upon, or comically—if it turns out that the perfect game is a win for White, 1. e4 and Black resigns, because...he doesn’t like his opponent’s face!!

USArmyParatrooper

Mar 21, 2018

0

#3780

Troy, all valid points and it goes even further than that.

How does ponz111 know that after 1. e4 e5, white doesn’t have Mate in 104 on the board? He doesn’t! He’ll say something like, “I know from my experience”