The Excelsior Springs High School Chess Team Takes Second at the Kansas City K12 Championship
The Excelsior Springs High School chess team attended their second tournament ever on Saturday, Dec. 9, 2017 at Oakhill Day High School. As compared to their first tournament, the team had garnered a great deal of new players due primarily to the increasing popularity of chess in the English resource room. While they attended the first tournament with only 3 players (Noah Turner, Alex Garcia, and Jameson Mills), the team has since tripled in size, now including Turner, Garcia, Josiah Dillon, Sam Yelton, Lake Yelton, Chris Paige, and Michael Meyers. Also on the team, but not participating, were Mills and Marcus Cook. Having played with a large number of beginners (and without their top rated player, Mills), the team was not a favorite to win the tournament.
Dillon and Garcia warming up before the games.
Going into the first round, there were five players stepping into a tournament hall for the first time in their lives: Dillon, Sam, Lake, Paige, and Meyers. Despite this, the team played their first round respectably, winning three out of their seven games. Among these winners were Turner, Meyers, and Garcia. It was an unfortunate start for Lake Yelton, who lost his first round game against Daniel Wilson despite having a crushing position up a full piece and two pawns.
Round two was less fortunate, with only Turner and Sam managing wins, while five others took a loss. While the team was generally okay with this fact, the usally confident Meyers seemed visibly upset at his performance. "He went to nationals," he said. "This isn't even fair." Such is life in open tournaments, though; some players are bound to be much stronger than others.
Paige thinking at the board
If round two was not bad enough, the team got completely demolished in the third round. After six people had reported their results, the team had scored zero points. It was only after these six losses that Turner's lengthy game against Sadie Wilson finished in the only win for the Excelsior team. The game had many ups and downs, but ultimately culminated in an elementary zwischenzug (in-between move) under time pressure:
The final round, while perhaps not so great for the team, was a huge victory for Dillon, who scored his first win of the tournament. Other winners from the team include Turner and Lake. An observant reader may notice something interesting: Turner won all four of his games! This secured him first place in the individual section off of tiebreak with Daniel Wilson, who also escaped with a perfect score. Turner, having played a slightly stronger field, was awarded the prize. "I do feel kind of bad for Daniel, though," Turner said. "It seems kind of stupid that he didn't get any prize just because he played weaker players. He can't control who he plays."
The team overall did not score spectacularly, scoring only 10/24 as a whole, but as there were only three participating teams, Excelsior still took second place. "I don't think we did pretty well for the lineup," says Chris Hull, the coach of the chess team. Fortunately, everyone who participated scored at least one win, which is a decent turnout for the many beginners.
Deepmind AlphaZero Crushes Stockfish in a 100 game match
On Dec. 7, 2017, the chess world was turned upside-down by Google's debut of AlphaZero. AlphaZero is a program that uses artificial intelligence (AI) to teach itself various tasks. Last year, Google shocked the Go community with their program AlphaGo by beating its best player in a match, claiming superiority in one of the few remaining board games in which humans were previously considered largely superior. This time, the machine learning program took on the game of chess. After being taught the rules of the game, the program simply played games with itself, moving randomly at first, and tried to determine optimal chess strategy on its own. Within four hours of training, the program promptly crushed the top chess engine Stockfish in a 100 game match, scoring 28 wins and 72 draws.
Google Headquarters, the home of AlphaZero
The machine takeover of chess began back in 1997 when Deep Blue first beat the legendary world champion Garry Kasparov in a 6 game match. It was at this time that the mere carbon life forms were first forced to hang their heads in shame; they had been outsmarted by their own genius. Since then, there has been no option but to accept inferiority. Chess engines have only been growing stronger, and it didn't take long until humans had practically no chance to snag even a single win against the strongest chess engines. Nowadays, Stockfish 8, the strongest chess engine in the world, boasts a massive rating of 3424; a full 600 points above world champion Magnus Carlsen. To many, it seemed that engines were asymptotically approaching chess perfection.
AlphaZero's performance earlier this month absolutely destroyed this assumption. The result was nothing short of astonishing. To see an engine that was thought to be just short of perfection get dismantled so efficiently almost defies belief. This phenomenon would seem to open up many new possibilities to modern chess.
The methodology is one of the most interesting parts of AlphaZero. Rather than being "taught" chess classically by using human parameters, the AI literally learned all on its own. The machine utilized a form of reinforcement learning, which allowed it to learn from its mistakes given the result of the game. Moving randomly at first, the machine would gradually dissect what kind of moves led to a favorable result. In some cases, this led to absolutely shocking challenges to human chess theory. Take a look:
AlphaZero opts for an ending where it has sacrificed a knight for four pawns, but it plays with the pair of bishops. Despite Stockfish's material advantage, the immense activity of the bishop pair on the open board allowed AlphaZero so much counterplay that Stockfish was quickly forced into a passive position where it was simply ground down by AlphaZero's piece activity. This strong penchant for the bishop pair presents interesting questions for modern chess strategy.
In a broader sense, AlphaZero seemed to consistently show a neglect for material in favor of piece activity. According to Tony Rotella, AlphaZero seemed to display "an odd willingness to sacrifice even large amounts of material in exchange for the quality of the pieces." Watch how AlphaZero fearlessly sacrifices a full rook in order to tie up Stockfish's pieces entirely and obtain deep, long-term compensation:
This kind of sacrifice is one which would seem to elude both humans and the normal chess engines of today. With no direct tactics and a lack of any mating attacks, AlphaZero simply goes down a rook in exchange for more positional factors. When pointed out, humans can certainly appreciate the beauty of the grand idea, though these ideas never come up in actual play, as they're often just dismissed or overlooked completely due to the large material deficits. "It's almost scary," says Rotella. "To see the level of play that's achieved when you take out the rigid human ideas, it almost seems like we were just holding them back." The normal engine (like Stockfish, for example) is also rather oblivious to these kinds of decisions. While AlphaZero has managed to build a sort of intuition that allows for the evaluation of these long-term sacrifices, normal engines suffer from what is called the horizon effect. As they generally just calculate variations and evaluate them to the best of their ability, when the point of a move goes further than they can calculate, they'll simply miss the purpose and fail to find them. Thus, when the advantage is not so concrete, the moves can be hard to find, while an AI program like AlphaZero would have no problem whatsoever in finding such ideas.
It would seem hard to deny that a new era of chess has now dawned. Once products like AlphaZero are released to the public, the opportunities for new developments in chess theory are abundant. The 28-0 result against the strongest engine humans have to offer serves as a strong testament to the newfound power of artificial intelligence in modern chess.