World Championship 2021 match will start Nov. 24!!

[David Ottosen]

Radjabov won a game against Kasparov when he was 16.
Wei Yi was the youngest player to break into 2700.
Caruana was the next best thing.
Wesley So skyrocketed past everyone when he gained 100 ELO points in about a year and crossed 2800.

All these guys were young, have incredible games, tournament victories and achievements. Everyone clamoured "here's the next World Champion!"

But you know who else was "young, have incredible games, tournament victories and achievements. Everyone clamoured "here's the next World Champion!" ?

Kasparov. Anand. Kramnik. Carlsen. What made any of them stand out vs (say) Timman or Shirov or Gelfand or Short whoever? Nothing, until they actually won it.

There's a handful of people in the world who have the potential to go on that path. Some make it, some fall just short. Firouzja is one of the people on that path, and whether he has that indescribable special bit of magic to take him over the top is something no one can predict. Maybe you think he doesn't have "it". But really, what you're saying is "until I see them win the championship, I cannot consider anyone as a future world champion".

[Pargat Perrer]

We've cars, planes, and even rockets, however people still love running and competing. Engines have own championships, human own.

I've seen this analogy many times, and it isn't valid. Whether it's people running or flying in jets, the object is purely simple: go as much distance in as short a time as possible.

Can you say chess is that simple? If yes, you don't know chess at all.

A more correct analogy would be to compare people running a very long and very complex OBSTACLE COURSE, in which the number of obstacles, and the order they appear, is totally different each time. An example would be the necessity of crawling through a narrow pipe. Not only that but obstacles can appear WITHIN obstacles, much like in chess tactics appear within tactics.

We don't yet have machines that could handle every obstacle we can imagine that are manageable by humans. A small machine might get through a narrow pipe, but could not handle a rock-climbing obstacle. It would take some sort of humanoid robot.... but we don't even know if that is the best form. What if there is a machine we haven't even imagined yet that could handle every obstacle and combination of obstacles that humans can handle, AND could do it better than humans?

This is what AlphaZero amounts to. It takes all the complexity and obstacles in the way of playing good competitive chess and it overcomes them all, much better than humans.

Therefore AZ is very very special, it may be the best thing humans have created yet in history.

[Pargat Perrer]

[QUOTE=Egidijus Zeromskis;n216876]

Since it is going to be decided by human blunders anyway/QUOTE]

A strange appreciation of the game to look for blunders.

I totally agree. But that is the world of competitive chess! We are all guilty.

If all elite chess games were played perfectly and had no mistakes, and were all draws, who would watch them? Who would appreciate them?

It is the nature of competitive chess that it must have mistakes in order to be fully appreciated.

[Alex Ferreira]

But you know who else was "young, have incredible games, tournament victories and achievements. Everyone clamoured "here's the next World Champion!" ?

Kasparov. Anand. Kramnik. Carlsen. What made any of them stand out vs (say) Timman or Shirov or Gelfand or Short whoever? Nothing, until they actually won it.

There's a handful of people in the world who have the potential to go on that path. Some make it, some fall just short. Firouzja is one of the people on that path, and whether he has that indescribable special bit of magic to take him over the top is something no one can predict. Maybe you think he doesn't have "it". But really, what you're saying is "until I see them win the championship, I cannot consider anyone as a future world champion".

Good points! Of course all the "ones that made it" had to also have been "the next world champion". I am simply not sold on Firouzja yet, based on a good year. Many big tests ahead. Far more fail to become World Champion, than those who succeed, naturally. Carlsen continues to be at his best, and continues to push himself. It'll be great if he gets more serious challenges in the future.
I would still argue this is different, and it's different because of Carlsen. Fischer, Kasparov, Carlsen. These 3 world champions, at least in "recent" times, stand taller than all others because of their domination of the chess world. Kramnik beat an out-going Kasparov in an excellent match, and then did everything he could to avoid a rematch, while never being #1 in the world. Many of the other world champions, never had to de-throne someone who was head-and-shoulders above the field. The way Carlsen is. When Carlsen was 20, everyone already expected nothing less than for him to become World Champion. At this point, Carlsen was already #1 in ratings, and that does count for a lot. He was favourite to win the candidates, which he narrowly won, and he was favourite to beat Anand in Chennai, which he did.
Quite frankly, it'd be great to continue this conversation in a year. If Firouzja by Dec 2022 has won a few major tournaments, and narrowed (or passed) the gap between Carlsen, I think we'll be in for something really exciting.

Alex F.

[Erik Malmsten]

Carlsen continues to be at his best, and continues to push himself. It'll be great if he gets more serious challenges in the future.
. . .
Quite frankly, it'd be great to continue this conversation in a year. If Firouzja by Dec 2022 has won a few major tournaments, and narrowed (or passed) the gap between Carlsen, I think we'll be in for something really exciting.

Alex F.

Carlsen has been working hard on his game for over 10 years, even learning from AI.. It will take FIrouzja a couple of years to catch up. I think Carlsen could play for a draw and draw all his games against any challenger, even computers. It takes unbalance to risk winning or losing.

What are the subtle understandings that make a 2700 grandmaster better than a 2600, a 2800 ahead of a 2700, and a champion ahead of a 2800? It is not just talent, memory, concentration, decision making, research, supervising a team, physical stamina (and relaxing when not playing), but also psychology, handling high levels of constant and surprising stress. A champion needs it all.

[Neil Frarey]

If you want to catch Carlsen ... catch him early in an unbalanced position.

Don't even try to fish in calm waters.

[Frank Dixon]

Responding to Erik Malmsten, concerning Capablanca and Alekhine:

1) It's true that Capablanca required Alekhine to raise the purse for the match, which wound up being at Buenos Aires 1927.

But that is how it was done back in that era -- challengers were responsible for this. And that is fundamentally unfair, since certain worthy challengers, such as Akiba Rubinstein, were not able to do this, so missed out on a chance for a title match. Rubinstein was a great player at his peak, but he was a quiet person, without a big media presence, or wealthy friends, and he could not find financial backers. He would have given champion Dr. Emanuel Lasker a very good fight circa 1912 and 1913, in my view. And David Janowski, while a strong player, got his chances at a title match since he had a financial backer, a wealthy Frenchman, Pierre Nardus, who was his personal friend. Lasker crushed Janowski in two matches before 1910.

2) Further, when Capablanca challenged champion Dr. Emanuel Lasker, at Havana 1921, Capablanca had to raise the purse for the match.

There was a lot of interest in Capa's native country, which had a wealthy upper class at that time, and Capa's family was quite prominent. The match could have been held earlier, but World War I (1914-1918) intervened. Capa won easily. He was at or near his peak at that time. He said that his peak came when he defeated Yugoslav GM Kostic in five straight games to begin their match, just before the world title match. Kostic then conceded.

3) After Capa lost a hard-fought match to Alekhine in 1927 (6-3 with 25 draws), he asked for a rematch, and there was interest in this in New York in 1928 and the first part of 1929 (the financial crash hit New York in October 1929, changing the situation completely).

Both players had strong ties to the city; Capablanca lived there for much of the year, was a member of the Manhattan Chess Club, and played in many strong tournaments there. He attended Columbia University for a year, but dropped out and concentrated on chess. By 1927, he was a Cuban diplomat, and the most famous Cuban in the world. Alekhine had written the tournament books for both the New York 1924 (Won by Em. Lasker) and New York 1927 (won by Capa) super-tournaments, and was well regarded in the city. He spent a lot of time there as well; GM Arnold Denker wrote in 1995 that he would spend up to three months at a time during the 1930s; the two were friends from that era. But Alekhine ducked the challenge from Capa; the champion could do that then.

Some of the documentation of negotiations through correspondence, between the two players at this time, as well as later, has survived, and been collected and published by the late Hanon W. Russell. I believe he had some of Alekhine's letters to Capablanca, but not the other side. No better evidence is needed for the change to a formalized challenge structure for the world title.

Instead, Alekhine accepted challenges from Efim Bogolyubow, who was not in the class of either Capa or Alekhine; Alekhine won easily in both 1929 and 1934, in matches played in Germany.

Alekhine could have given Aron Nimzowitsch a chance; in the late 1920s and early 1930s; that would have been a fascinating struggle between two very strong personalities and great players. Nimzo died in 1935.

4) Alekhine kept Capablanca out of tournaments that he was playing for nearly a decade.

For example, Alekhine's massive triumphs at events such as San Remo 1930 and Bled 1931 didn't have Capa in the field. The next time the two met, following the match, was at Nottingham 1936 (nine years later!); Capablanca WON!! The organizers there refused Alekhine's request that Capa be blocked from playing.

In the meantime, Alekhine had lost to Max Euwe in 1935, in a very close match over 30 games in the Netherlands, in what was likely his best lifetime performance. When Euwe won, he gave Alekhine a rematch in 1937, also in the Netherlands; Alekhine had quit drinking for the match, and won fairly easily. Euwe raised the prize money for both matches. Remember also that Euwe was an AMATEUR; his career was as a math professor. Euwe wanted to bring the title matches under the control of FIDE, which was not attained until 1948, after Alekhine died in 1946. It was in a tournament format designed by Mikhail Botvinnik, and approved by FIDE. Botvinnik won the tournament.

5) I think the 'talent versus hard work' argument is largely overdone here.

Capa had never had a significant setback until losing the 1927 match. Clearly, his methods worked!! Many experts, even up to the present day, have opined that no greater natural talent than Capablanca has ever played.

Alekhine devoted himself to very close study of Capa's games, and found some weaknesses, which he exploited. Full credit. But Alekhine did nothing besides chess; Capa also was a diplomat for Cuba. I believe that Alekhine considered that he could play his best and still lose a rematch to Capablanca; that is why he avoided him, making specious excuses.

Alekhine also avoided a world title match with the 24-years-younger Estonian Paul Keres (born 1916), following Keres' win in the 1938 AVRO tournament, which was Euwe's attempt to formalize the challenger process and structure. AVRO had the world's top eight players in a double-round format; Keres won on tiebreak over American Reuben Fine. True, World War II was approaching. But Alekhine could have played Keres then; my belief is that Keres would have won that match.

6) Finally, I am going to bring in two quotes:
a) "I have known many chess players, but only one chess genius -- Capablanca." This was Alekhine's tribute when Capablanca passed away suddenly in 1942, at age 53. (This highly complimentary, incisive, and largely respectful essay, but with some dodgy rationale by Alekhine on their negotiations, is reprinted in '107 Great Chess Battles 1939-1945', collected and translated by Edward Winter, 1980, Dover publishers. With Capa dead, Alekhine tried to make up for some of the venom he spewed.).
b) "Alekhine sought lesser opponents, rather than give Capablanca a rematch." GM Robert Byrne, in the Introduction to 'The Immortal Games of Capablanca', by Fred Reinfeld, Dover publishers. GM Byrne was just getting into serious chess in 1942 in New York, at age 14, when Capa passed away. Reinfeld (1910-1964) also knew Capa from that era; he was one of the top players in the USA in the 1930s, before concentrating on writing.

[Egidijus Zeromskis]

I've seen this analogy many times, and it isn't valid. Whether it's people running or flying in jets, the object is purely simple: go as much distance in as short a time as possible.

Can you say chess is that simple? If yes, you don't know chess at all.

A more correct analogy would be to compare people running a very long and very complex OBSTACLE COURSE, in which the number of obstacles, and the order they appear, is totally different each time. An example would be the necessity of crawling through a narrow pipe. Not only that but obstacles can appear WITHIN obstacles, much like in chess tactics appear within tactics.

We don't yet have machines that could handle every obstacle we can imagine that are manageable by humans. A small machine might get through a narrow pipe, but could not handle a rock-climbing obstacle. It would take some sort of humanoid robot.... but we don't even know if that is the best form. What if there is a machine we haven't even imagined yet that could handle every obstacle and combination of obstacles that humans can handle, AND could do it better than humans?

This is what AlphaZero amounts to. It takes all the complexity and obstacles in the way of playing good competitive chess and it overcomes them all, much better than humans.

Therefore AZ is very very special, it may be the best thing humans have created yet in history.

In this line of thought let me know when engines will start composing chess problems without a human interference.

[Pargat Perrer]

In this line of thought let me know when engines will start composing chess problems without a human interference.

AlphaZero could do this easily, if fact it would be amazing if the people controlling AZ would train it to do this. All that would be required is for there to be some sort of scoring for already existing problems. AZ would learn what makes some problems better than others. Then after some millions of training problems with scoring to tell it what is good and bad, it would create its own.

Remember, AZ taught itself chess without human interference. I believe it can do problem creation too if it knows a scoring to it.

EDIT: in my previous post I mentioned that the best form for solving random obstacle courses might not be humanoid, even though it seems a humanoid form seems best because we humans can handle all sorts of obstacle courses.

Well, it should be noted that AZ is not a "standard form" of computer. Although some AZ-like engines might be available now for home computers with GPUs in their hardware, maybe even without the GPUs, but the true AZ, the one that really hums, requires much more hardware than a home computer has. So part of the invention of AZ was the realization of the new form of computer it would require.

[Peter McKillop]

[Regarding composing chess problems] ....it would be amazing if the people controlling AZ would train it to do this. All that would be required is for there to be some sort of scoring for already existing problems. ....

Remember, AZ taught itself chess without human interference. ......

Why can't AZ teach itself to compose problems, including developing its own scoring system (*if* that is what AZ needs), without human interference?

[Pargat Perrer]

Why can't AZ teach itself to compose problems, including developing its own scoring system (*if* that is what AZ needs), without human interference?

I am not sufficiently knowledgeable in the internals of AZ to answer that question, nor do i know for sure that AZ would need a scoring system.... if it was fed millions of chess problems, all with the same number of plies N to a solution, here's what I *think* it would figure out on its own:

1) every problem must have one White King, one Black King, and at least one other piece, possibly many other pieces, with one side's number of pieces and Pawns never exceeding 16 in number, all placed on an 8x8 chessboard, with neither King in check and all Pawns excluded from 1st rank and 8th rank.

2) the rules of chess (if it weren't already 'knowing' that from previous training)

3) the goal of producing a checkmate in N plies of legal chess play, with only 1 such line producing the mate, and any generated position that cannot do that is rejected.

It *might* be able to figure out that every piece on the board must have a purpose. That is a really fascinating question, because that is part of the beauty of chess problems. I think the reason I mentioned the scoring system is that if the scoring system were totally *accurate* and consistent (with accuracy being somewhat subjective? i.e. beauty is in the eye of the beholder), then it would work to make sure it was creating problems that would score high on the system of scoring.... which it would figure out on its own. If the scoring was random, it could not be expected to produce beautiful problems (garbage in, garbage out).

And if there were no scoring system provided, then AZ would have no way to know that any scoring system was even desired, so it would not figure that out. It would create random problem positions that we might not think were special at all. Just a lot of mate-in-N-plies positions, probably with totally unnecessary pieces included. In fact, it might take one such position and then vary it by adding / removing unnecessary pieces in almost infinite variety -- although maybe not, because the input problems that it trained from wouldn't do that. So it would likely figure out that that was not desired.

But that question, would it figure out that each piece must have a purpose? Wow, that is fascinating!!!

[Erik Malmsten]

I would think the computer process would be more like the Nalimov Endgame Database. Calculate all the mates in 1 with 3 pieces, then 4, 5, 6 etc. Then mates in 2 feed into those positions. Then mates in 3 etc.

[David Ottosen]

I am simply not sold on Firouzja yet, based on a good year. Many big tests ahead..When Carlsen was 20, everyone already expected nothing less than for him to become World Champion. At this point, Carlsen was already #1 in ratings, and that does count for a lot.

This is fair and I'm playing a bit of devil's advocate. When I think back on the last 40 years or so, when you think of those people who were considered "chosen ones" before winning the championship, you think of Kasparov, Anand, Kramnik, and Carlsen, where it seemed almost pre-destined that they would someday be the champ. Even the other examples I gave I'm not sure anyone ever really saw them that way.

Do people see Firouzja that way? I dunno. His best bet is probably to catch a hot streak and catch Magnus when he's bored or stressed or whatever about defending the title.