Using an engine in arena

[sorryimlikethis]

It's been long speculated that a certain player has been using an engine, or bot, to play arena games this season.
Well, they have finally come forward and admitted to doing so via Discord.
But here's the kicker - they claim that nothing they are doing is against BGA rules..?


And also encourage other users to play with software that suggests the best moves to make.


What are the community's thoughts on this?
Names hidden for obvious reasons

[Jellby]

https://boardgamearena.com/legal?section=tosv:
"BGA will not tolerate any cheating and/or lack of fair play, whether through the use of cheating software..."

You can engage in a discussion on whether that qualifies as "cheating software" or not.

https://boardgamearena.com/faq?anchor=f ... edrankings:
"Some games are simply not suitable for online competition on Board Game Arena [...] This is due to various reasons [...] AIs specialized in the game are easily accessible and have been used to cheat."

Here it's clearly implied that using software to suggest moves counts as cheating.

Report the player and, if it really becomes common, then competitive features for Azul should be removed.

[flolem]

wonder how can we suggest the best move to do ?

But he say clearly he has a software that interact with BGA... for me it is cheating...

Report the player and, if it really becomes common, then competitive features for Azul should be removed.

This would be sad... Pretty all games with open information can be more or less cheated with some software...

[FrankJones]

Seems like blatant cheating to me.

Unless... is the person making the argument that he or she created the software, and that the program is not an AI-type software that thinks for itself, but rather, utilizes the creator's Azul skill based on how the creator plays Azul?

In that case, I still think BGA would object, but I could see some possible merit to the argument. For example, if I wrote a chess program that suggests moves, but the only information I fed the program was my own chess skills, the program would not suggest great moves, because I am not a great chess player.

Players on BGA are allowed to review their own games, and then can use that information to try to figure out the best moves in other games. Perhaps this person wrote a program that looks for similar games that player played, essentially creating a database of his or her own games that would be faster than manually reviewing games?

I mean, in theory, someone who has played 1000 Azul games could review old games one at a time until finding one that has a similar enough situation to the situation in a current game, and watch the game replay to see how the game turned out.

I would definitely like to see an admin weigh in one this matter.

[euklid314]

wonder how can we suggest the best move to do ?

But he say clearly he has a software that interact with BGA... for me it is cheating...

Report the player and, if it really becomes common, then competitive features for Azul should be removed.

This would be sad... Pretty all games with open information can be more or less cheated with some software...

Seems like it was a Master thesis project of an computer engineering student. From a scientific standpoint I find it fascinating to have somebody write a self-learning Azul AI and analyze if it plays moves that no strong human would play. In games like chess and go, such self-learning algorithms revolutionized the game play to some extent (these algorithms only get the rules as an input and not a single human game as training data).

But you cannot use BGA Arena framework to test such an algorithm. Perhaps BGA should allow such Bots playing only friendly play. Many very strong players would love to test and improve their skills against such an AI. And the AI could internally compute its own ELO number which could be presented to its opponents. Since only friendly games are played such a bot would not change BGA Elo.

[Mathew5000]

For example, if I wrote a chess program that suggests moves, but the only information I fed the program was my own chess skills, the program would not suggest great moves, because I am not a great chess player.

Do you think that would be allowed at a chess tournament?

Try this: write a computer program to play blackjack, feeding it only your own blackjack skills, and then use it to play at a Las Vegas casino. But in fact don't try that, because you will be arrested, because it is cheating.

[Andrei3009]

Seems like blatant cheating to me.

Unless... is the person making the argument that he or she created the software, and that the program is not an AI-type software that thinks for itself, but rather, utilizes the creator's Azul skill based on how the creator plays Azul?

In that case, I still think BGA would object, but I could see some possible merit to the argument. For example, if I wrote a chess program that suggests moves, but the only information I fed the program was my own chess skills, the program would not suggest great moves, because I am not a great chess player.

Players on BGA are allowed to review their own games, and then can use that information to try to figure out the best moves in other games. Perhaps this person wrote a program that looks for similar games that player played, essentially creating a database of his or her own games that would be faster than manually reviewing games?

I mean, in theory, someone who has played 1000 Azul games could review old games one at a time until finding one that has a similar enough situation to the situation in a current game, and watch the game replay to see how the game turned out.

I would definitely like to see an admin weigh in one this matter.

Even if the algorithm is based entirely on the actions of the person who wrote it, it is performed by a machine that, unlike a human, does not make mistakes. Stability is a very important factor in assessing a player's skill, especially when you can calculate a lot of things in the game.

For example, in Azul, on the last move, all information is open and you can find the optimal move by calculating all possible options and I can do it if I have 5-10 hours. However, with the help of friends, I can write a program that will do this for me. This certainly won't make me an expert, but it will seriously increase the chances of winning some games in which I wouldn't have calculated or just made a mistake.

[FrankJones]

For example, if I wrote a chess program that suggests moves, but the only information I fed the program was my own chess skills, the program would not suggest great moves, because I am not a great chess player.

Do you think that would be allowed at a chess tournament?

Try this: write a computer program to play blackjack, feeding it only your own blackjack skills, and then use it to play at a Las Vegas casino. But in fact don't try that, because you will be arrested, because it is cheating.

In a chess tournament over-the-board? No, this would not be allowed. Of course not. But that's not what we're discussing here.

I'm reasonable sure it is understood that people are watching strategy videos and reading strategy threads for the games they play. For example, Primptonite (I think that's the spelling) has a well-known Youtube video series on Azul strategy. Is it cheating to watch those videos?

Your statement about blackjack cannot possibly be accurate. What crime exactly would someone be arrested for in that circumstance? The casino would probably ask the customer to leave, but I don't see how an arrestable crime would have occurred. I mean, a person can walk up to a blackjack table with a book on basic strategy while playing blackjack. Some dealers will even tell a player what the "book" play is.

If I played blackjack with some handwritten notes reminding me of my strategy, do you really think the casino would care?

[FrankJones]

Even if the algorithm is based entirely on the actions of the person who wrote it, it is performed by a machine that, unlike a human, does not make mistakes. Stability is a very important factor in assessing a player's skill, especially when you can calculate a lot of things in the game.

For example, in Azul, on the last move, all information is open and you can find the optimal move by calculating all possible options and I can do it if I have 5-10 hours. However, with the help of friends, I can write a program that will do this for me. This certainly won't make me an expert, but it will seriously increase the chances of winning some games in which I wouldn't have calculated or just made a mistake.

Agreed, and I expect that such a program would still be against BGA rules.

I merely sought to clarify the distinction between:
Using AI to come up with its own move, which the player then makes, and
Using a database of one's own games to try to figure out the best move. As you noted though - there is a difference between manually researching one's own past games and writing a program that automates the task.

Regarding Azul - on the last move, I'll spend as long as it takes to find the right move
But, yes, it would me only my brain and my calculations being used, and it would take a lot of time.

I still believe the BGA official policy is hazy. I mean, people are encouraged to study and review games played by high-rated players. Many people use "book openings" for chess. Where is the line between cheating and studying?

[Mathew5000]

Your statement about blackjack cannot possibly be accurate.

Well go ahead and try it then! But maybe google it first: https://www.google.com/search?q=nevada+ ... ter+felony