Forgotten History: The Modern War of Chess AI!

  This is part 2 of the series, talking about the Story of Chess AI. The previous blog posts talked about the pre-2000 Era, where IBM’s Deep Blue dominated Humanity. However, in 2008, Tord Romstad, Marco Costalba, and Joona Kiisk created something. And that thing is called Stockfish!

So let’s dive right into it!

THE CREATION OF STOCKFISH

In 2004, a college student named Tord Romstad created an open-source chess bot called Glaurung. The bot never exploded into the competitive scene, however, it laid the framework for the eventual creation of Stockfish.

4 years later, two more college students came to the team and created what would be known now as Stockfish. Stockfish was especially powerful because it was far more aggressive than any other chess engine, considering the most forcing options first.

When it was first created, it wasn’t the most powerful chess engine by any means, being ranked 3rd or 4th when it was first created. However, due to its forcing options, this aggressive chess engine very quickly became the most potent chess engine objectively. But for how long?

Google’s War Against Stockfish

The previous chess bots had learned to play and master chess by using loads of human data. But Google’s approach was very different. Google had created a neural network, where their robot, Alphazero played millions of games against itself. 

This allowed it to improve at exponential speed. This brand new neural network was so powerful that it could defeat the legendary Stockfish, which had been champion for over 6 years in a row, with just 9 hours of training. The legendary games between Stockfish and Alphazero infamously ended with Alphazero winning 99% of the games.

Alphazero was interesting as it hugely favoured activity over material, sacrificing multiple pawns, sometimes multiple pieces to gain a colossal activity advantage. Sadly, AlphaZero’s development was stopped by Google, and once again, Stockfish became the reigning champion.

LEELA CHESS ZERO

Google’s Bot, AlphaZero became such an excellent success that Gian-Carlo Pascutto used AlphaZero’s reinforced learning to create Leela Chess Zero (LCZero). LCZero was first announced in January 2018, and within a few months, it had already reached the Grandmaster level.

LCZero, unfortunately, did not have as big of a legacy as AlphaZero, however, in 2019, it won the Top Chess Engine Championship, scoring significantly higher than Stockfish. LcZero was known to have exceptional strength in the Fischer Random category.

THE PRESENT + THE FUTURE

Recently, Chess.com released their own chess engine, named Torch, which is currently the second most powerful Chess Engine. It is unknown currently whether Stockfish will be dethroned or not, but, we can say that Stockfish has lamented its legacy, as one of the Greatest Chess Bots in History.

I Started a Chess Movement in 24 Hours.

ChessMinds Prized Blitz Arena Free Entry 3+0 180 Minutes $50

Amped Chess

Magnus v Abasov & Fabiano v Pragg! Salimova v Goryachkina | Open SF & Wo...

PogChamps 5 Finals: LIVE From LA!! CDawgVA v Franks-is-heres & QTCindere...

Forgotten History: The Ancient Past of Chess AI (Part 1)

  In this current day and age, the use of chess computers has become so trivial, many have taken it for granted to be able to play against a 3700-rated machine. People may not have realised this, but Chess AI has come a long way.

You may have noticed from the title that this will be a 2 parter since this story is so long, and I didn’t want to leave out any details. So, strap in as we are going to go through the history of Chess AI, from the ancient past to today.

THE TURK

The Turk is the first ever recorded Chess “robot”. This robot was showcased worldwide, defeating some of the greatest players. It played against many well-known figures, such as Napoleon Bonaparte and Benjamin Franklin. For a robot made in 1770, this was very impressive.

The Turk was incredible in another way. During its game against Napoleon Bonaparte, he tried to make illegal moves. The Turk was smart enough to be able to detect them and turn back the move, something NO chess AI can do TO THIS DAY!

This robot was amazing, as there was a small window into the robot, for the players to see inside! However, it had one BIG problem. It was a complete LIE. There was actually a small box, where a person could hide, and play the moves for the Turk.

THE FIRST REAL CHESS AI

The first engine that was able to play the game of Chess is the Los Alamos Bot. Created in 1956, it was the first bot to be able to play a simplified version of chess. In the 1980s however, the development of Chess-AI rose massively. Belle was the first bot to reach the Grandmaster level.

However, Deep Blue was a much bigger revolution than any of these bots are. Yes, the bots before it were at the Grandmaster level, but humanity was still better. After Deep Blue, Bots became objectively stronger than humans. Many do not know this however, there were actually 2 versions of Deep Blue.

The first, lesser-known version and game played by Deep Blue, was played in 1996, against Garry Kasparov. It’s less known as the results ended with Garry, 4, and Deep Blue, 2. However, one year later, history was written when AI became stronger than the strongest human.

CONCLUSION (PART 1)

This is where the story ends for most people. Ask a chess fan you know, and they will end at this point, the point where Deep Blue defeated Garry Kasparov. But this isn’t the end of the Story. Because something happened in 2008. And that thing… is called STOCKFISH.

FIDE World Cup 2023 | Semifinal Open | Final Women | Game 1

FIDE World Youth U16 Olympiad - ROUND 9

Pinkamena (1373) vs A. Yunker (1730). Chess Fight Night. CFN. Blitz

GM Abasov, Nijat vs GM Vidit, Santosh Gujrathi𝐅𝐈𝐃𝐄 𝐖𝐎𝐑𝐋𝐃 𝐂𝐔𝐏 2023 𝗹 𝗛𝗶𝗴𝗵...