Have you ever cheated at go?

[Bonobo]

Well, since I didn’t get an answer, I assume that using an online Joseki dictionary in a correspondence game is NOT cheating … even more so since I still decide for myself how to proceed and don't ask anybody what I “should” do.

[dfan]

Well, since I didn’t get an answer, I assume that using an online Joseki dictionary in a correspondence game is NOT cheating

It depends on the rules of the server. For example, here are the DGS rules. If the server doesn't list rules about external assistance, then it doesn't seem supportable to call it cheating.

I come from the chess world where these sorts of rules have historically been very explicit. For example, consulting opening books and playing out variations on a physical board have generally always been considered to be inherently part of the correspondence chess experience. For this reason, consulting joseki dictionaries in a correspondence game seems perfectly reasonable to me, but again, if the server disallowed it, I wouldn't do it.

[gowan]

I don't recall having cheated and I seriously doubt the benefit of occasionally looking up a joseki. It reduces the mistake rate with 1 or 2. When the opponent deviates from the joseki, you still need to know how to "punish" it and what the punishment is actually worth (many think their opponent should lose when deviating from joseki, while it often is just a 2 point difference)

I'm from before the time when bots were useful. I would find it very interesting to play a game with the help of a bot. I think it may give a Hikaru-Sai-like experience, which makes you a better player. But I'd never do it without consent of my opponent.

Obviously, there is no point in carrying an online rank under a pseudonym which actually reflects Leela's strength. You must have a very low self esteem to go down that path.

I once conducted an experiment with playing using a program to help me. My friend and opponent was almost 300 Elo points rank stronger and he played with no assistance. I used a fairly weak program, an early version of Many Faces of Go, which at that time was maybe 100 Elo stronger than I. Essentially I became the evaluation program. MFOG suggested moves and I chose the move. The result was that I could play 300 points higher in rank. I often rejected what MFOG suggested and played some moves not suggested by MFOG. It was fun to try that but I wouldn't do it in a serious game because my goal is to play better all by myself. This wouldn't work using something like Leela or even stronger monte carlo type programs because they are so much stronger than I that I couldn't function as an evaluation unit.

[tapir]

This topic has some severe self-selection bias ...

Imo, in pseudonymous online play any consistent conduct is fair. If someone wants to spend his time as user interface for a bot, who cares? Rating exists to sort this out. This might be a good learning experience for him, but I would expect him to use a different account when playing himself.

As long as our Hikaru has different accounts for Sai and himself, I am fine. In fact, I would be interested in someone openly trying to be the Leelabot for a year and (honestly) reporting the results this had on his own performance.

[Bill Spight]

TImo, in pseudonymous online play any consistent conduct is fair. If someone wants to spend his time as user interface for a bot, who cares? Rating exists to sort this out. This might be a good learning experience for him, but I would expect him to use a different account when playing himself.

One thing that can make it difficult to improve is that playing the right play will often take us into unfamiliar territory, where we play poorly. One way that bot assisted play can help is to let the bot be our guide into and in the unfamiliar territory.

For instance, a lot of players do not tenuki enough. They might come up with their own candidate move and then, before making a play, see what the bot recommends. If it says to tenuki, make the bot's play. Then there will be new challenges to face. Or a player may be insecure about making life. Then let him look at the bot's suggestion only when he is worried about the life. Another idea is to look at the bot's suggestion only after making your play. Just doing that should help you to improve, because you get immediate feedback.

OC, any of these ways of playing with bot assistance should have their own accounts.

[HermanHiddema]

I really think it's not a black and white issue, and context is important.

For all of you asking where to draw the line, which things are *really* cheating, ask yourself the following question: If I were a top player playing an official game in a major tournament with significant prize money, would the referee consider this behaviour cheating?

If the answer is yes, then your behaviour is, in that context, cheating.

Now ask yourself: If I were playing a casual friendly match with a close friend, would they consider this cheating?

If the answer is no, then your behaviour is, in that context, not cheating.

The examples mentioned here, looking up joseki or playing out sequences, would definitely be considered cheating in official matches. But probably plenty of your online opponents are fine with them. So they are a bit of a grey area.

Personally, I avoid that grey area, but in casual play I won't mind if my opponents go there. I also wouldn't mind if they consulted a stronger player. In a casual game, anything that increases their enjoyment of the game is fine by me.

[Ian Butler]

Personally, I wouldn't mind if my opponent cheated (I don't play tournaments, though, there I might ) because that's his business. Sure, I'll lose, but I'm playing someone a lot stronger so I hope to learn from it. The only downside in that case is that you might feel bad because you got outclassed by a "weaker" player if his rating is lower than yours. But even then I would suspect a sandbagger so I wouldn't feel that bad after all.

[zac]

This topic has some severe self-selection bias ...

I would be interested in someone openly trying to be the Leelabot for a year and (honestly) reporting the results this had on his own performance.

y
This is an interesting thought. You can think about the moves that you yourself would make, and try and make sense of why Leela plays the way it does. Because you have to be actively playing out the moves, inputting your opponents moves, I can imagine that it could have some positive effect on skill. Whether that time would be better spent playing your own games, reviewing your own games, solving problems, playing through pro games etc., who knows? It would still be an interesting experiment

[lowiq60]

When I bought crazy stone three years ago (what a huge waste of money by the way - I was not smart enough to predict the impact of deep learning on go), I made a KGS account to test the strengh of the program at Level 7 and Level 8. Of course I manually relayed the moves without telling my opponents that I was playing for a bot. I wanted to know exactly what level of an opponent I would be training with, and testing it on KGS is the only way to answer the question. I'm not sure if I cheated but I did make my opponents participants in an experiment without their consent...

In the very early days of Leela Zero I did the same thing, only I was manually relaying moves of (at the time) a glorified random number generator (and lost many many games to IdiotBot!)


On an unrelated note, why do people still check josekipedia or joseki dictionaries? Haven't go AI's already showed that our understanding of the opening is deeply flawed, and what we accepted as joseki's are likely not even "joseki's" in the traditional interpretation of the word?

[dsatkas]

I use josekipedia from time to time in my turn-based games on OGS/DGS

[Kirby]

i'm surprised by how people don't seem to consider using a joseki dictionary to be cheating. when i play an opponent, even a casual game online, i'm assuming that the moves are coming from their head.

i guess herman has it right when he says that it depends on context. but i never thought of a tournament game and a casual game on kgs that different in terms of "what is cheating" - to me, it was always just a difference in how serious i was about trying to win a game. some people here say that even using a bot isn't cheating in an online game.

so if using a bot isn't cheating, and using a joseki dictionary isn't cheating, then what type of behavior *would* be considered cheating? whatever fits within the TOS is free territory?

[gowan]

i'm surprised by how people don't seem to consider using a joseki dictionary to be cheating. when i play an opponent, even a casual game online, i'm assuming that the moves are coming from their head.

i guess herman has it right when he says that it depends on context. but i never thought of a tournament game and a casual game on kgs that different in terms of "what is cheating" - to me, it was always just a difference in how serious i was about trying to win a game. some people here say that even using a bot isn't cheating in an online game.

so if using a bot isn't cheating, and using a joseki dictionary isn't cheating, then what type of behavior *would* be considered cheating? whatever fits within the TOS is free territory?

RIght now it seems to be generally accepted that people no longer have to do routine arithmetic, using a calculator such as is built into cell phones is common. Even advanced engineering type math is now done by programs like Mathematica or Maple. I can imagine that in the not too distant future it will be possible, even common, to have AI brain assists implanted on human bodies. If that happens then things like "thinking" games (go, chess, etc.) will often be played in a hybrid way. From my perspective that sort of thing is not significantly different from watching pros play now; we could get a lot of pleasure from it. But it isn't playing go as we know it now.

[Pio2001]

Come on ! Using assistance while playing go is like using a car to win a 100-metres race.

[BaghwanB]

Maybe I'm just inventing this distinction to justify my own habits, but I draw a strong line between what's allowable in real-time (F2F or KGS for example) vs. long-term (like DGS) playing.

In real time, it is just you and me. No aids allowed at all since we are both committing the time and place to play out a game with each other. We bring what we have inside ourselves to the table and that is it. My position on long-term on the other hand developed over time. Back when I was studying hard, I could come across some text or playout that was very similar to something going on in an active DGS game. If I just learned something good, should I not do it since I didn't know it at the start of the game? So this lead to me feeling "directed study" during a game was OK and then generalized it to all reference materials (and play aids outside of other people directly) were fair game during a week or month long game.

I think the key is to be explicit in what's allowable outside of F2F play where I've only been offered the use of play aids once (as opposed to a handicap) and I think everyone expects it to be you and just you. Even though DGS has stated their policy I still added it to my bio that I'll use references during play. But I'd still be annoyed if I found out a KGS opponent was flipping through books or web pages or even just doing variations in another window much less running a program and aping moves.

Again, the above could be why 98% of my online play is on DGS so I might just be self-justifying, but that's my current perspective.

i'm surprised by how people don't seem to consider using a joseki dictionary to be cheating. when i play an opponent, even a casual game online, i'm assuming that the moves are coming from their head.

i guess herman has it right when he says that it depends on context. but i never thought of a tournament game and a casual game on kgs that different in terms of "what is cheating" - to me, it was always just a difference in how serious i was about trying to win a game. some people here say that even using a bot isn't cheating in an online game.

so if using a bot isn't cheating, and using a joseki dictionary isn't cheating, then what type of behavior *would* be considered cheating? whatever fits within the TOS is free territory?

[Uberdude]

Another aspect for real-life play, is it cheating to wander around and look at other boards? I do this a lot; I get restless and find it interesting to watch other games, but maybe I would win more and get in time trouble less (I don't wonder on my time, but could use opponent's time to read/plan more) if I focused more on my game. I think it's not cheating. But what if you then use the information from other games to help you in yours? I think that would be. I don't think I've done this consciously, but it would be hard to police, though it seems somewhat plausible. For example at the Amsterdam tournament this last weekend I considered playing a new joseki variant but ended up playing the old one, not because I couldn't remember the new mainline but was unsure how to continue in some branches if my opponent deviated, and then when I went for a wander I saw one of the top boards playing the new line so although that wouldn't have helped me as I had remembered that line (and wouldn't help with the branches) it would have been a big help if I'd forgotten it, and could have provided some reassurance. Or another potential situation, imagine you've got somewhat common shape like a carpenter's square and forgot the usual continuation, and another board had it too and in your wander you saw the 2-1 attachment sequence as a nice reminder of how to make the ko.