Introduction

I was always aware that there is computer cheating in online chess. However, I never regarded it as an insurmountable problem because I always thought their number is too low in terms of the percentage of total players.

Also, it wasn’t such a big problem for me personally, since I almost exclusively played blitz and bullet in the past – disciplines with short time controls where computer cheating is much more difficult to pull off (although not entirely impossible).

However, some 2 months ago, I decided to take online chess more seriously and gave myself aim to play at least one longer game – rapid (10+0) or classical (15+15) – a day. Soon after I started doing it, I became aware of how many computer cheaters actually are there in online chess, as I started encountering them on a regular basis.

Without exaggeration, I think I encounter one cheater in every five games. I successfully reported over 30 of them, ¹  I even started reporting people I have never played against, just on the basis of their account. But new cheaters keep appearing every day and there is seemingly no end.

I got to a point where I have started aborting games against anyone who I consider even mildly suspicious ² – I decided it is worth being overly suspicious than spending anywhere between 20-40 minutes to play against the engine. ³

Just another day in the office when playing longer games online

Since this situation is incredibly frustrating and annoying and since one of the reasons why many cheaters get away with it is people not bothering/not being aware that they play against one, I’ve decided to write this article and share my experience regarding computer cheating in online chess accumulated during the last two months.

The article will be divided in two parts. First, I will talk about recognizing cheaters and describe signs you might be playing against one. Then I will talk about how I personally deal with them and what steps do I take after losing a game against someone I find suspicious.

I hope you will enjoy it and find it useful ⁴

How to recognize computer cheating in online chess: 12 signs a player might be a computer cheater

Big correlation to engine moves in multiple games

First and most obvious. Computer cheaters make very strong moves and usually crush you mercilessly.

It has to be pointed out that these aren’t ordinary crushes. In my online games, I regularly play against titled players. But even when I am up against a Grandmaster, I rarely feel I have zero chance from the start to finish – especially if we play multiple games. ⁵

When playing against an engine, the feeling is completely different. You have the feeling as if you were never really in the game. Their moves are just perfect and, needless to say, are unanimously correlated to top 3 (top 5) engine choices.

If you see your opponent had a small average centipawn loss, ⁶ you have to be vigilant that there is something fishy.

Finding beautiful and deep tactical ideas

Since chess is 90% tactics, ⁷ it is no wonder computer engine users excel in this area of play. Very often, they demonstrate very deep, long and forcing tactical ideas that are very difficult for a human (even of a higher playing strength) to find.

The previous game against SharpBishopChess is a good example, but this one is no less fascinating:

Demonstrating perfect conversion technique

Even though the difference between a human and the engine is apparent in all stages of the game, it gets even more pronounced when the stronger side obtains a clear advantage and has to convert it.

Upon obtaining an advantage, human players have a tendency to relax and start playing suboptimal moves. It is not a coincidence that there is a saying that „the hardest game to win is a won game“. Numerous games have been spoilt by laziness/over hastiness in a position with a clear advantage, ⁸.

Also, humans usually have a way they go about converting their advantage. We all tend for simplifications in such positions and avoid sharp continuations that might offer tactical counter chances.

Engines don’t have this problem, as the following game demonstrates:

Playing bad/weird openings and then continuing to play strongly

Another common trait of engine users is playing very bad/weird openings, landing in a suspicious/bad position with a lot of pieces on the board and then proceeding to outplay you.

I am not sure why this is characteristic. Perhaps some of them use engines without access to opening database ⁹ Or they just don’t bother because they know they will win anyway so they want to humiliate you.

But in any case, it is another sign to be aware of. Take my game against the user Marshall_The_Goat as an example:

Making very unnatural, „engine like“ moves

There is also a way engines play. There are some moves that are very difficult to find for humans (and some moves humans don’t even consider) – we all have certain aesthetics ingrained.

But engines don’t care – they will calculate and play the best continuation no matter how it looks.

A continuation such as the whole sequence ending up in Ng5 and Rf4-Rh4, trapping the queen, such as my afore-mentioned game against SharpBishopChess or seveshnikov.

This beautiful queen sac played against IM Eric Rosen is also a move worth noting:

Playing moves in regularly spaced intervals

Another typical cheater trait is playing the moves in regularly spaced intervals (say 3-5 seconds for blitz games), because it takes them time to feed the move to the engine, get its output and then play it on the board. A good example of this can be seen in the following video by Christoff Sielecki A.K.A. Chessexplained:

To be frank, I haven’t encountered this one myself all that much, probably because longer time controls provide the cheaters with more degrees of freedom.

But it is definitely something to keep in mind!

Taking time to make very obvious moves

Irrespective of whether they take in regular intervals or not – all cheaters do take time for all their moves.

Even the most obvious ones like a forced exchange or capture of a loose piece. For example, take a look at yet another game I played against user Marshall_The_Goat. It took my opponent more than 30 seconds to take the piece after executing a double attack. I actually didn’t realize he was cheating after that first one we have seen, but after this one, I had no doubt he was:

The following one against NT_99 is also worth noting:

Therefore, if someone executes a brilliant 5-move combination and then takes his time to play the final move and reap its rewards, you are more probably playing against a cheater.

A significant drop in level when faced with time trouble

I personally haven’t yet encountered it in longer time controls (although I did suspect some people in blitz because of it), but many cheaters start playing really badly when faced with time trouble.

The reason is simple – with little time, they don’t have the luxury of consulting their „silicon friend“ any longer. So they start playing themselves and reveal their „true“colors. The final result is sometimes comical, as the afore-mentioned video by Chessexplained demonstrates.

Having an anonymous account

This is probably self-understandable, but every single player I successfully reported for cheating had an anonymous account.

I usually check the profile before every game (even blitz ones) to see who I am playing against. If a player has a name listed, I also check his FIDE profile, just to see if he is legitimately a strong player of not.

With cheaters, you usually don’t have that luxury. You can’t see if they have a FIDE profile. Sometimes they even put some arbitrary FIDE rating, but there is no way to verify it.

But I guess that is kinda the point – nobody wants to risk their reputation by writing their name out in the open.

Having a recent/new account

Another thing I have noticed with cheaters – many of those I reported have made their account relatively recently and didn’t have a huge number of games.

Many cheaters don’t want to risk their „actual“ account, so they open a new one just for cheating purposes. Also, they all eventually get caught, so it is impossible to keep cheating consistently through a huge period of time and games.

Accounts made recently don’t have that constraint.

Having big discrepancy in rating between short and long time control (or no ratings in shorter time controls whatsoever)

Another thing I always look when dealing with potential cheaters is their blitz and bullet rating.

Many of them have a big discrepancy between ratings in short and long time controls. It is not uncommon to for me to play against a user with 1700 or 1800 blitz and bullet rating and 2200-2400 classical/rapid rating. Even though to speak of one is not to speak of another, I believe a certain correlation between these ratings has to exist.

Also, many cheaters don’t have blitz or bullet ratings whatsoever, which makes is it even more suspicious? What kind of player has NEVER been tempted to play at least one shorter game?

Different ways of trying to make the account less suspicious

Last, but not least, even though cheaters are very dumb, they are not completely dumb.

Some of them do realize that the account with high rapid/classical rating, big winning percentage, or big discrepancy in rating between bullet/blitz and rapid/classical is suspicious.

That is why some of them try various methods of making their accounts less suspicious. Here is an overview of different methods I have seen while investigating and reporting cheaters in my lichess classical career:

Using engine only in some games/only in some moments during a game

The most obvious thing. Even though the majority of cheaters is dumb enough to use the engine during the entire game, some of them realize this doesn’t help them to reduce the suspicion, so they use the engine irregularly – only in some games or – more commonly – only in some moments during a game.

This game against TheRicketsRick927  fooled me:

We will also see another game against polish_warrior_3000 later which was a good example of this. Also, both games against Marshall_The_Goat are exemplary of that (he skipped engine in the opening completely).

Intentionally resigning games to reduce their winning percentage

Since cheaters most often crush their opponents mercilessly, they usually have absurd winning percentages.

Some of them realize that and try to reduce it intentionally, by losing games on purpose. Some of them are smart enough to play an entire game without the engine and lose it in a human way, but some of them are so dumb that they play anywhere between 5-10 moves and then resign, irrespective of the position on the board.

I guess your time is too valuable to spend on actually playing chess when you could be cheating instead.

Losing casual rapid/classical games intentionally

Often combined with the previous point – many cheaters play casual games and then lose them intentionally. It both increases the total number of games they have played and includes loses in classical in their career stats.

Playing a huge number of games against the same opponent (friend) and losing some of them, but still gaining rating

I caught one or two cheaters playing 10-20 classical games against the same opponent and essentially trading rating with him. Most of the games ended up in quick resignations /checkmates and both cheaters actually gained rating at the end of it.

It was a clear attempt to include some classical games in their collection of games to make the account seem more legitimate.

Playing a low number of games in all time controls/variants

Some engine users know that an account with only classical games is suspicious, so they try to play some games in all time controls and even variants of the game.

It is not too hard to detect if you actually click on the profile and see the number of games. If you bother to check the actual content of the games, you will most probably see a lot of short games and early resignations, as well.

But if you don’t bother to click on their profile and just check if they have any ratings by hovering over it, ¹⁰ you might get fooled, so some of them like to implement it.

Playing the first game on their own/with moderate engine use, asking for a rematch and then using the engine

Last, but not least, I recently played a pair of games against a clear engine user.

In the first one, he played a horrible opening, then after my imprecisions found a very good defence, managed to equalize, reached a completely drawn endgame and then fell for the checkmate of the knight against an edge pawn. The game was actually pretty funny:

Since I assumed he was playing by himself, I accepted his rematch offer.. and got crushed like a bug, as we have seen earlier (the third game I have given in the article).

I guess it can be compared to a drug dealer. He first “hooked me” in the first game (although I am not sure he actually wanted to – he probably aimed for a draw and then relaxed prematurely) and then punished me in the second one.

Disgusting.

How to deal with computer cheating in online chess in 4 steps

Now that I have described how to recognize a computer cheater, I can get on with how to deal with them. Even though the answer: „Report them“ is self-evident advice, I’d like to expand on it a bit. I have pinpointed 4 important steps of dealing with computer cheating in online chess.

Don’t make premature accusations. Give your opponent the benefit of a doubt (everyone can play 1 good game)

I have recently read the book Talking to Strangers by famous writer Malcolm Gladwell. The topic of the book is the human tendency to evaluate unknown people completely wrong, which often leads to misunderstandings and conflicts.

In one of the chapters, Gladwell writes about Harry Markopolous, an American investor and financial fraud investigator, who warned people about the Bernie Madoff pyramid scheme as early as in 2001. He tries to explain why how exactly Bernie Madoff managed to keep his fraud going for so many years and why nobody listened to Markopolous even when he came out with his claims. ¹¹

Gladwell mentions that we humans have a truth bias – a certain tendency to expect the truth and to convince ourselves that even the most unviable events are actually possible. He also points out that this mostly a good characteristic and that world would be unable to function normally if we suspected everyone.

He also emphasizes that Markopolous is one of those rare people whose brain is wired completely differently. He was incredibly suspicious in all spheres of his life and Bernie Madoff scheme was just one of the examples of how it manifested itself.

Now, you might be wondering why I am blabbering about Gladwell, Madoff and Markopolous. The reason is that both extremes – tendency to expect the truth and tendency to suspect everyone – are bad when dealing with computer cheating in online chess. ¹²

One of the reasons I think many of the cheaters manage to get away with it is the fact that other players simply don’t suspect them. Some of the accounts I reported and got banned were so blatantly obvious that it left me wondering how on Earth did none of their previous opponents did not react earlier. Of course, I don’t recommend anyone to become a narcissistic, overly competitive rating-whore like myself, but I do think that cheaters would have a much harder time if an average Joe was slightly more vigilant.

On the other hand, I also think I have become too suspicious, borderline paranoid. I think it is important to give your opponents the benefit of a doubt and realize that anyone can play one good game. Acting too rashly, sending reports without any investigations and publicly accusing people of cheating just because it „felt like it“ are more likely to start a witch hunt than to help anyone.

I think it is important to keep everything private and under the table and to be rational when it comes to dealing with such a situation – not to go „full Markopolous“.

Take your time to investigate their profile and analyze their games. Search for signs and patterns mentioned above

I think this is the most important step of dealing with cheaters most people are not willing to make. ¹³ Before reporting a cheater, it is necessary to gather as much information/evidence as possible by going over their profile, analyzing their games and searching for signs and patterns they might be cheating mentioned above.

I usually do it by checking obvious stuff such as when was the account made, whether it has a name and what are the ratings in all time controls and variants.

Then I go and analyze a lot of their games  (5-10) by myself. At first, I look at their average centipawn loss (which is available immediately in the lichess environment) and then check the nature of games.

It usually takes me 30-60 minutes to do this. But it is definitely worth it – not only because I convince myself that the person is actually cheating and avoid starting a witch hunt, but also because I am able to provide moderation with much more context and more convincing arguments. ¹⁴

Report the cheater and let the moderation do their job

After you have done your homework, there is nothing else left to do but report the cheater, block him and wait for the moderation to respond.

Don’t be like me and report the cheater several times – the moderation is doing their job the best they can, ¹⁵ so filling several reports about the same person doesn’t really help.

Laugh in the cheater’s face and taunt him if he gets reported and goes on to insult you below your Youtube videos

Of course, if you followed these instructions closely, the cheater will have no ideas who reported him or what hit him.

However, if you are a sore, salty loser like the author of these lines you can also accuse a person of cheating in the post-game chat, report him and get him banned and then have some fun when he posts angry comments below your Youtube videos because he knows exactly who caused his online chess career to end:

I don’t usually endorse the use of the „mum“ jokes, but when it comes to computer cheating in online chess, I have no inhibitors or empathy, whatsoever. Good riddance.

In any case, I hope you have found this article helpful and that you will manage to deal with computer cheaters in online chess more successfully after reading it.

The post How To Deal With Computer Cheating In Online Chess appeared first on Chessentials.

Throughout history, chess was revolutionized several times.

In the late 19th century, the first World Champion Wilhelm Steinitz provoked his opponents to attack him vigorously and laid the foundations of positional chess.

In the middle of the 20th century, the sixth World Champion Mikhail Botvinnik started devoting an enormous amount of time to physical and chess preparation before the tournaments and matches, announcing the era of strict chess professionalism.

In the 1970s, largely under the influence of eleventh World Champion Robert James Fischer, the opening theory went through massive changes. New opening systems such as the Hedgehog were introduced and new ideas were discovered in ancient opening systems, demonstrating the inexhaustible nature of chess.

Nothing else, however, has revolutionized chess so much as the appearance of computer chess engines. Nowadays, even the biggest beginners are familiar with the terms „Stockfish“ and „Rybka“ (especially when they are shouting their names in online chats while kibitzing the games of top chess players). Even a chess computer on your mobile phone is stronger than a grandmaster nowadays. It is not surprising players have started using them as a tool and learning from them.

Even though we all embrace computer chess engines as something normal and use them on a daily basis, most of us aren’t familiar with the challenges and problems previous generations had to face in order to create one. Who devised the first computer chess engine? How long did it take for computers to become stronger than humans? What is the difference between classical computer chess engines and Google’s Alpha Zero?

In this article, we will answer these questions and take a closer look at the history of computer chess engines.

THE 18TH CENTURY: THE TURK

In 1769, French illusionist Francois Pelletier was performing an act in front of Maria Theresa of Austria, at the Schönbrunn Palace in Vienna. Among the onlookers was Hungarian inventor and author Wolfgang von Kempelen. Inspired by Pelletier’s performance, Kempelen immediately started building the invention which would later become one of the most notorious hoaxes in the entire history.

Next year, in 1770, at the exact same place, Kempelen exhibited the Turk – the first chess-playing automaton in history. Its complicated construction consisted of several compartments with various operational mechanisms. It incorporated a life-sized model of a human, dressed in traditional oriental clothes (hence its name).

Turk was capable of playing chess on its own and beating human opponents. It was even able to recognize illegal moves and force its opponents to take them back. In one exhibition, it even solved the difficult knight-tour-around-the-board puzzle.

There was just one problem. It was all fake. The Turk was not the one making all the moves. A human hidden inside it was.

You see, during the construction, Kempelen envisioned several hidden compartments, big enough to fit an adult. He even made sliding hallways, through which a person was able to move from one compartment to another. It was important to preserve the illusion – before each exhibition, Kempelen would invite the audience to examine the automaton, during which the operator of the machine would remain hidden.

And it worked? Kempelen toured Europe exhibiting his invention until his 1804. After his death, the machine changed several owners and continued playing chess until 1854, when a fire in Philadelphia sealed its fate. It is amazing no one realized the fraud during the Turk’s lifetime.

Thus, the first computer chess „engine“ was a major commercial success.

However, if we disregard several similar automatons, the development of chess engines was stalled during the next 100 years. The real breakthrough happened only in the late 1940s and early 1950s. And just like with many other technological advances, a major historical event fulfilled the role of the catalyst.

THE 1950s: THE BEGINNING OF AN ERA

During World War II, Axis and Allies leadership spent a lot of time hiring top scientists and gathering them into teams. The purpose was clear – winning the war. Greatest minds of the 20th century devoted their wartime years to developing technology and weapons that will help overpower the enemy. The science was in service of the war industry.

In spite of that, the war led to a startling amount of breakthroughs. Scientists laid a foundation for a number of new fields. In the context of this article, computer science is of particular interest. Two famous names played the crucial role in its development – Claude Shannon and Alan Turing.

Shannon is widely recognized as the father of the modern information theory, while Turing is the father of the modern computer (Turing’s machine!). Their contributions to the field of computer science are well known. What is perhaps lesser known, is that Shannon and Turing are also fathers of chess computer engines.

After finishing their PhDs on the eve of the war, Shannon and Turing both started working on cryptanalysis, in USA and UK, respectively. They closely followed each other’s work and even met in person back in 1943, during Turing’s two-month stay at Bell’s Lab in New York, where Shannon was working. After the war ended, they both took an increased interest in programming a chess computer.

(Main source describing the relationship between Shannon and Turing: Life In Code And Digits – When Shannon Met Turing)

In 1949, Shannon published an iconic paper titled “Programming a Computer For Playing Chess”, in which he described an algorithm for the chess-playing machine. Simultaneously, Turing was developing his own chess playing program. His work on “Turbochamp” started back in 1948 and finished in 1950.

(Link to Shannon’s article)

It is noteworthy to mention Turing programmed Turbochamp on paper, without access to an actual computer. He tried testing it on Ferranti Mark I – the first commercially available computer – and failed. In the end, he tested it manually (!) in a friendly chess game in 1951.

(51 years later, none other than Garry Kasparov tested “Turbochamp”. The program didn’t stand a chance, but the greatest players of all times recognized how tremendous Turing’s achievement was – source: Alan Turing Created A Chess Computer).

It is not yet certain whether Shannon and Turing worked independently or were inspired by each other – the content of their conversations in 1943 is not known. However, even if they didn’t collaborate, their contributions were immense. Subsequent generations of computer chess developers all “stood on the shoulders” of these two giants.

In 1951, Turing’s colleague Dietrich Prinz managed to implement the algorithm on Ferranti Mark I and created a program capable of solving mate in two. In 1956, a team of scientists led by Stan Ulam (one of the inventors of the h-bomb)created a program capable of playing chess on a 6×6 board. And finally, in 1957, IBM engineer Alex Bernstein created the first automated program fully capable of playing a complete game of chess.

The era of computer chess has officially begun.

THE 1960s AND 1970s: SMALL STEP OR A GIANT LEAP?

The first computer chess engines were rather weak and primitive. During the 1960s and the 1970s, however, their strength increased rapidly. Two major factors contributed to this qualitative leap:

• Improved and more sophisticated algorithms

In the 1960s and 1970s, algorithms for the computer chess engines were significantly improved. The foundation was set by the genius of John Von Neumann, who developed the MiniMax algorithm, perfectly suited for the game of chess (it minimizes the score of one player while maximizing the score of another). In the decades that followed, MiniMax search was improved with advanced heuristic techniques and “iterative deepening”, which gradually increased the depth of the search by MiniMax.

• Improved and faster hardware

The greatest restriction for Turing and other pioneers was computing power. In the 1960s and 1970s, however, hardware speed increased exponentially – according to Moore’s law, the computing power doubled every two years. It allowed the implementation of advanced algorithms – the search time for MiniMax was significantly decreased.

It was not only the matter of raw computing power, though. In the 1960s and 1970s, computer chess gained prominence. The first all-computer chess championship was staged in 1970 in New York. Shortly afterward, in 1974, first World Chess Computer Chess Championship was staged in Stockholm. The appearance of the “scene” inevitably led to the foundation of specialized companies which started developing customized chess playing software and hardware. Which allowed computer chess engines to get even better.

How good “get better” was? How good were computer chess engines compared to humans at a time? Experience showed – good enough to beat an amateur, not good enough to compete with a master.

For instance, in 1967, MacHack VI became the first computer chess engine to beat a human opponent. It’s playing strength revolved around 1300 USCF rating. But already in 1976, a significant leap happened – chess computer engine Chess 4.5 won the Class B Section of the Paul Mason tournament in Northern California. In 1977 it also won Minnesota Open with the performance rating of 2271 and beat a Class A player Stenberg – rated 1969.

However, the computers weren’t on equal terms with stronger humans. In 1968, International Master David Levy bet 3000$ he would be able to beat any chess computer engine in the next 10 years. In 1977, he won his bet in a match against chess computer engine KAISSA. He also beat the newer version of MacHack in 1978.

Therefore, at the end of the 1970s, could compete against humans, but couldn’t play on a master level.

But then the 1980s arrived.

THE 1980s: CHALLENGE TO HUMANITY

By the start of the 1980s, computer chess engine programming has become a lucrative business. Personal computers have become widespread in households. The interest in customized and specialized software – including computer chess engines – exploded. In 1982 alone, computer chess companies topped 100 million in sales.

It is not surprising, thus, that computer chess engines continued improving. In 1980, their programming became a serious competition. Edward Fredkin, professor of computer science at Carnegie Mellon University, introduced the Fredkin Prize. He offered monetary prizes for various achievements in the world of computer chess programming (5000$ for the first engine to reach Master level, 10000$ for the first engine to reach Grandmaster level, 100000$ for the first engine to beat the World Champion).

The race was on. The number and strength of computer chess engines continued increasing.

Fredkin’s colleague from Carnegie Mellon University led the way. The most notable project and the spearhead of an entire generation is Deep Thought – the first chess computer engine which reached Grandmaster Level. In 1988, it shared the first place with Grandmaster Tony Miles in US Open Championship. In 1989, it easily beat International Master David Levy four games to zero.

The downfall of humanity seemed inevitable. In the search for its Champion, it turned to the reigning World Champion and arguably the greatest player of all times – Garry Kasparov. First, in an iconic simultaneous exhibition in Hamburg in 1985, he played against 32 (!) strongest chess engines – and beat all of them (!!). Then, in 1989, he faced Deep Thought in a two-game match, which he also won.

(Source: Kasparov And Thirty Years Of Computer Chess)

Therefore, in his encounters with engines in the 1980s, Garry became the “chosen one”.

Alas, in the 1990s, he too would fall.

1990 – 1997: ENGINES TAKE OVER

Thus, in 1989, humanity won the battle, but not yet the war. In the 1990s, many more clashes between humans and computer chess engines were held. Top chess engines even competed in elite chess tournaments – with mixed success. Several “human vs machine” matches were organized, but none of them became famous as the match in which the engine defeated the World Champion in 1997.

In 1989, researchers involved in the Deep Thought project were hired by IBM. They started developing a more powerful version of the engine.

The name of the project was Deep Blue. In 1996, only 7 years after the Kasparov – Deep Thought match, round two of „man vs. machine“ was held: the first Kasparov – Deep Blue match. In the very first game, Deep Blue shocked Kasparov and became the first chess computer engine to beat a World Champion in a classical game. However, Garry composed himself and won the match with the result 4-2, prolonging the inevitable once again.

The failure didn’t dishearten Deep Blue programmers. During the next year, they continued to improve the engine’s strength. They even hired Grandmaster Joel Benjamin as a consultant to help them build the opening book. After thorough preparation, they challenged Kasparov to a rematch again – and as they say – the rest is history.

In the celebrated match, Kasparov won the first game. But the engine’s counterintuitive 44th move confused him. It shook his confidence and he missed serious drawing chances in the 2nd game. Then Deep Blue crushed him in 6th game with a knight sacrifice early in the opening. This sealed Kasparov’s fate and earned Deep Blue team the Fredkin Prize.

(Source: Deep Blue Inventors Win Fredkin Prize)

After the match, he claimed IBM cheated and that a human was actually making the moves (Joel Benjamin). Other sources speculated a bug in Deep Blue’s code brought him victory.

(Source: Did Deep Blue Beat Kasparov Because Of A Glitch?)

In any case, 1997 symbolically marked the end of human domination over the chess computer engines.

However, humans didn’t yet resign.

1997-2006: HUMANITY’S LAST STAND

After Deep Blue’s victory, everyone expected the gap between the humans and machines to widen further. However, over the next six years, until roughly 2003, humanity made its last stand.

For instance, in 2000, computer chess engine Deep Junior participated in the Dortmund Elite Supertournament – and scored just 50%. In 2002, new World Champion Vladimir Kramnik drew his match against computer chess engine Deep Fritz (despite a catastrophic mate-in-one blunder). In 2003, Garry Kasparov drew two matches against Deep Junior 7 and X3d Fritz.

Alas, this small ray of hope was quickly destroyed by computer chess engine Hydra. In 2004, it beat GMs Evgeny Vladimirov (3-1) and Ruslan Ponomariov (2-0). Then, in 2005, it destroyed Michael Adams, a member of the world top 10, with a devastating score: 5.5-0.5.

And in 2006, Deep Fritz drove the final nail in the coffin of humanity by beating the World Champion Vladimir Kramnik 4-2.

2006-2017: THE GOLDEN ERA OF COMPUTER CHESS ENGINES

The 2006-2017 decade is the golden era of the computer chess engines.

On an everyday basis, computer chess engines infiltrated all areas of chess – from broadcasting and analyzing to playing. They became an integral tool for any chess tournament player. They raised a generation of strong youngsters – the so-called computer generation. When smartphones appeared, their Android and iOS versions were immediately developed, to the horror of any tournament anti-cheating committee.

World Chess Computer Championship continued growing. Due to the prestige and money involved, some controversies ensued. The most infamous case is the disqualification of the computer chess engine Rybka, the consecutive winner of the four editions between 2007-2010, due to code plagiarization.

(Source: Rybka, The World’s Best Chess Engine, Outlawed and Disqualified)

Apart from the World Chess Championship, a new competition was introduced in 2010 – Top Chess Engines Competition. In contrast to the WCC, TCEC features longer games played using high-end hardware, which leads to a higher quality of chess.

The human-machine matches proved the gap is widening. Over the decade, a number of handicap matches were held, in which engines gave various odds to strong grandmasters. The results were depressing for humanity. For instance, in 2014, computer chess engine Stockfish defeated GM Daniel Naroditsky who used the assistance of an early version of Rybka. In 2015, computer chess engine Komodo played 6 odds games against Sergei Movsesian, a former top 10 member and crushed him easily. And in 2016, Hikaru Nakamura, a current member of world top 10, feel down to Komodo as well.

(Sources: Man Versus Machine Historical Archive and Komodo Beats Nakamura In Final Battle)

In any case, during the 2006-2017 period, classical computer chess engines dominated. Until the end of 2017, nothing revolutionary happened. Their strength changed consistently, but slowly. Everything seemed familiar and nobody expected any radical changes, any major breakthroughs.

Then AlphaZero happened.

2017-ONWARD: A NEW REVOLUTION?

On 5th December 2017,  a group of scientists from Google AI company Deep Mindshattered the chess world.

In the paper titled “Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm“, they described the development of a new chess engine, AlphaZero, based on a completely new approach. Instead of alpha-beta searching and linear approximation function for position evaluation used by traditional engines, AlphaZero uses non-linear approximation function based on a deep neural network and Monte Carlo Simulation. As a consequence, instead of brute-forcing through the analysis, AlphaZero is able to “self-learn” chess.

The team tested the strength of the new engine in a 100 game match against the strongest classical chess engine available at the moment – Stockfish 8. The result of the match was staggering – +28-0=72 in AlphaZero’s favor. But what was even more impressive was AlphaZero’s style of play – it was much more intuitive and human-like than the play of the traditional engines.

The reactions to the paper were mixed. Many chess players were amazed – the mighty Stockfish suffering such a defeat was indeed a miracle. But on the other hand, there was a lot of skepticism. A lot of criticism has been directed toward the playing environment of the match – Stockfish ran on inferior hardware and played without an opening book. One interesting article on Medium was very careful to label Alpha Zero a scientific breakthrough in AI.

However, on 7th December 2018, another paper was published by the Deep Mind team, titled “A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play“. In it, they presented the results of another match between Alpha Zero and Stockfish. Out of 1000 games, A0 won 155 games and lost only 6. Even though the older version of Stockfish was used, it is an impressive result.

Irrespective of whether skepticism is justified or not, Alpha Zero made a significant impact on the chess world. They forced everyone to start pondering about what is coming next. They might signify the start another revolution.

One thing is certain: with the entry of AI, computer chess engines will never be the same.

REFERENCES AND ADDITIONAL READING

The Best Schools: Brief History Of Computer Chess

Bill Wall: Early Chess Computers

HPC Wire: Deep Blue Inventors Win Fredkin Prize For Computer Chess

IBM: DeepBlue

The post History Of Chess Computer Engines appeared first on Chessentials.