In light of recent events I feel it is appropriate to discuss the future of online go tournaments. I am sure that while players want these tournaments they will continue. These are some questions I can think of that organizers of such tournaments and players may wish to ponder and perhaps resolve. First some facts.

1. Cheating occurs in online tournaments.

2. There is currently no indisputable way to detect cheating.

An observation:
Governing bodies responsible for online tournaments do not seem to be equipped to deal with a world where powerful go programs are easily available and can be run on relatively cheap computers.
In another thread viewtopic.php?f=10&t=15538&start=680 Javaness2 wrote:
"In this case (EWGC) the rules committee was not consulted at all, and indeed resigned when it found out about the matter. The appeal process was started during the tournament, but step 2 was never invoked since there was no appeals committee, and step 3 went to the wrong body. The exec should have just redirected it to the rules committee."

Some questions.

1. Can any reward be given to the winner of an online tournament?

2. Should online tournaments be anonymous, where players play with aliases, such as is normally the case when players play on a go server? (This would remove much of the temptation to cheat, since cheating would not really achieve anything meaningful.)

We just might have to go back to playing go for fun.

This may be a problem for a few people with overblown egos. I won't miss them.

_________________
Help make L19 more organized. Make an index: https://lifein19x19.com/viewtopic.php?f=14&t=5207

"We just might have to go back to playing go for fun." This seems like an unwarranted assumption. In a world with thousands of years of history of competition in various sports and games, I can imagine that many, if not most, people have never viewed Go (chess, bridge, golf, tennis, football [American or otherwise], or the Olympics) as just "for fun". And such a sweeping denigration of everyone potentially concerned about the issue of cheating with bots, just may say more about your ego than theirs.

_________________
Dave Sigaty
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21

Yes, if you evaluate the risk of cheating and are content with the possibility of it going to a cheat vs the motivation of the prize attracting people to the event. Having offline finals for the prize positions, as in the PGETC, can ameliorate the problem. Some examples:
1) KGS organises a monthly online tournament, the winner get 3 months of KGS+ (which costs $15). Someone could cheat and gets a gold crown icon for bragging (hmmm?) rights and some KGS+ for free. KGS+ is nice but not money so I doubt would attract cheaters too much, so as a KGS manager I would be happy to continue an event like this in the strong bot era, for now. They already require solidly ranked accounts to play, so a 5k couldn't use LZ and then win it, though they could make a solid 9d account using LZ prior to the tournament and then win it. If a load of new 9ds keep winning then re-evaluate policy.
2) KGS organisers a big tournament for strong amateurs with $1000 prize money (XY0908 sponsored something like this years ago). This is much more problematic in the strong bot era, so would at least require some anti-cheating measures like requiring webcams but that's still maybe not enough so this kind of event probably won't happen anymore
3) PGETC, the top 4 teams who win prize money play over-the-board finals, where it is harder to cheat. Maybe you make it so only places 1-3 win prizes in case 4 got there by cheating. As a participant in a team who will finish in place 10-25, the presence of prize money has no impact on me and I would still play the event if there was no prize money, though perhaps some of the top players are incentivised by the prizes. I imagine for many of them it is still a secondary (or lower) motivation, the desire to play games at a competitive level, win, do well for your country being more important. Simba said he is strongly patriotic and that is a big motivator for him, I'm not so much and whilst countries are an obvious organisational unit for teams would also play if the teams were based on the first letter of your name (yay, we get Andri Kravets and Artem Kachanovsky and Ali Jabarin!).
4) 2 years ago China organised the "Gold Cup" which had big prizes ($20k ? ) for amateurs. The preliminaries were online (I qualified through WAGC result). There was some suggestion then that some of the Chinese players got a strong friend to play for them (Hu Yuqing 8d, one of top Chinese amateurs failed to qualify, 4d me noobed up against one of the Chinese 5ds and lost by only half a point so that's odd as Hu Yuqing would pwn me). Running such online prelims is problematic in the strong bot era, and afaik the event isn't happening again. I don't know if that's because of worries about cheating now and/or because last time the Koreans rather embarrassingly won everything.


I think playing anonymously will increase rather than decrease the temptation and occurrence of cheating. If you are playing under your real name then your reputation is on the line. Look at facebook's real name policy for example, although facebook comment threads do have some unpleasant stuff, it is far less toxic than anonymous places such as youtube comment sections or 4chan. You can still give meaningful rewards (e.g. $1000, or ego boosts) to an anonymous winner. So prizes and anonymity are basically orthogonal, about the only relation is an anonymous winner can't so easily translate the bragging rights / ego-boost to real life, though of course I can tell my real life acquaintances I was that tartaric guy who won some online tournament, would they believe me. Also you cannot enforce anonymity, the player behind the alias could choose to reveal themselves if they want.

It seems to me that too many people are concerned with the specific issue of cheating with online bots. I think people need first to sort out their attitudes to cheating in general.

The emergence of AI go programs is a perfect illustration of the proverb: "Be careful that you might what you wish for." In other words, every new development brings in its train a host of unexpected problems. Since cheating is endemic in real life - and maybe even biologically necessary - we can always expect cheating to be among those unexpected problems.

But how do we differentiate properly - not just in some vague morally intuitive way, but properly - between the guy who plays every move by asking a bot for the next move and the guy who learns a trick move from a bot (or a book) that he would never have discovered himself and wins a game because of it. Or in tennis, how do you distinguish properly between a player who puts itching powder inside her opponent's sports bra in the locker room and one who, off-puttingly but unawares, grunts every time she hits the ball.

People just don't respond to problems in an overarching or consistent way. If you a have a leaky roof, do you put pails out on the grounds that it doesn't actually rain that often and a new roof is expensive - and might you be dead in a couple of years. Or do you fit a new roof?

Not everyone thinks a problem is a problem. In go I have seen major arguments erupt because someone objected to his opponent not playing first move in his upper right corner. Is impoliteness even a form of cheating? Is misinformed objection to such a spurious allegation a counter form of cheating? In tennis again, is objecting to grunting also a form of cheating? Is trying to get the psychological upper hand ever a form if cheating - e.g. playing to the crowd and trying to get their sympathy? Is grabbing a seat so that your opponent has the sun in his eyes cheating? Is trying to put the clock on the "wrong" side of the board for the opponent and hoping he won't object a form of cheating?

Is playing strictly according to the rules as Robert Jasiek famously did in a major amateur tournament cheating? Very many people did think so.

And if you answer yes to any question of whether it is cheating, what punishment do you administer? How can you be consistent or fair? Even murder can get you only ten years in prison now. Is that fair on the victim's family? We used to hang starving peasants for stealing a loaf of bread but ignored the lord of the manor who raped their daughters under droit de seigneur. A drugs cheating role model in the Olympics might get banned for just one year, but what about his responsibility for the guy in the gym who copies him but overdoses and dies? Is it really fair to give whopping punishment to someone who does something against the rules because the Minister of Sport says it's good for the country - and you won't get your grant otherwise? Or to someone from a poor country who is trying to make a level playing field for himself with athletes who get the support of rich American universities?

We can all multiply these examples many times over. I haven't got the specific answers and I don't for a moment believe anyone else has either. But I do believe that we can approximate to a solution by establishing what the majority feels about the issue as a whole (which is, in effect, what happened in the RJ case). In many sports or pastimes this is partially formalised by setting up an organisation where to have to promise to uphold agreed values (e.g. not to cheat). It doesn't necessarily stop cheating but it does provide a way for the majority of members to show their disapproval, and while the "micro" treatment of banning or ostracising a defaulting member may seem tame at the time, the upholding of the majority "macro" view can be extraordinarily powerful in the long term. At any rate, judging by the plethora of organisations everywhere, most people appear to think this group-morality approach is worth doing.

It seems impossible to make this work for a disparate group of worldwide amateurs operating anonymously online. But, rather than quibbling about statistics or how strong you think someone is, would it be more fruitful to try and find a way for group morality to express itself?

What this might mean for an event such as the one that sparked off the whole debate is that it must begin NOT with some organiser offering to run a tournament and freely allowing interested people to play (the usual approach to go tournaments) but with a published code of conduct which interested players must explicitly sign up to before being allowed to play. The code of conduct could say not just that members must not cheat but that they must also avoid giving any impression of cheating, and that if a majority of members believe cheating has occurred or is likely to have occurred, a defaulting member can be banned (and/or named and shamed). That would not necessarily have stopped the alleged cheating we have been talking about, but it would have allowed the swift and decisive action the organisers tried to take once they suspected it. It may turn out to be unfair in reality but at least the alleged perpetrator knows in advance it is a possibility and he has the opportunity to take measures to avoid such suspicions.

And the majority would be happy. That seems to be the criterion that works in every other activity. We don't seek perfection. We live with an approximation to it. If that approximation is not close enough, the majority can vote in another way: they can just give up that activity. A shame for those who want to continue, but they are after all the minority.

So what I recommend as a fruitful way for the discussion to progress is for a sufficient number of people to make their views known about what should be in a pre-published code of conduct for each kind of event.

To get the ball rolling on that, my own view can be partly inferred from what I said above, but I would add that I think the sanctions should be strong (and the appeal procedure minimal or even non-existent for money-free amateur tournaments). I also think a requirement to provide a real name and publishable photograph is not unreasonable, at least if there are significant prizes or glory involved.





Until

Fascinating, while you are here, could you tell me about this so called RJ case, or point me in a direction where I can learn something about it?

Ultimately I think the solution is, as someone already kindly pointed out, to forget about cheating. Just enjoy the game and play for fun.

My guess is that because of the difficulty/expense of getting all the players to one physical location we are going to want to have some tournaments via on-ine connection.

Which means perhaps we step back and brainstorm WHAT measures/conditions could alleviate the cheating issue.

I'll start this off, "trusted observers". Each on-line player is to have present trusted observers who will verify that the player was the player, no computer (OR human) assistance, etc. Cheating is of course still possible but would require collusion.

I gave a bit of thought to that question after the online cheating scandal on IGS several years ago by the Chinese player with the handle, Sprint. It seemed to me that you could have a number of tournament locations that would not be too difficult or expensive for the players to attend. All games, online or FTF if both players were on site, would take place with a local tournament director/observer, possibly with the help of other observers.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

That would have been my number two suggestion << it's more or less a special case of my number one >>

Anybody have any other suggestions? We're still brainstorming.

I mentioned that possibility in another thread: https://lifein19x19.com/viewtopic.php?p=232615#p232615

_________________
I am the author of GoReviewPartner, a small software aimed at assisting reviewing a game of Go. Give it a try!

Perhaps an informed and instructed witness is sufficient?
It would require that the organizers make detailed instructions available to witnesses, but no need to go to the local go club.
Witnesses need not necessarily be good go players, I think. Any go club member will suffice.
Just someone (subcribed on beforehand, using real name and contact details) would suffice? I'd prefer a computer literate above a (high) dan player to keep an eye. Witness shall be obliged to attend at least XX part of the game live and check for unauthorized running programs (to be defined on beforehand by organizers and as instructed) and report to the organizers.
For prize winning tournaments, a 10% fee is for the witness.

Just an idea.

Online university courses also have to worry about cheating. Some institutions are using proctoring software. You install a program on your computer before you take the exam that records your webcam, screen, and internet activity. The teacher can see your screen during the exam, as well as your face to tell if you are looking at some other resource.

Computer + human team tournaments may be a potential option to consider. Since people can cheat with a computer, make it so that using computer assistance isn't cheating.

Maybe efforts could be made to provide all players with the strongest version of whatever go AI is available.

Perhaps less interesting, but I guess that's how things are now.

_________________
be immersed

All of this talk of observers and playing in monitored conditions rather misses the mark IMO. Maybe it applies if you are playing for the IGS online world amateur with $10,000 prize. But taking the PGETC as the primary example (seeing as that's what started all this), I wouldn't want to travel to some monitored place to play my games every few weeks (and where exactly, our go club meets in a cafe/pub/college canteen, and I live close to a club, others might be far away). It's just not convenient and too much of an imposition on the vast majority of honest players. Applying such conditions to only suspicious players (e.g. CM) would seem more reasonable.

I think online tournaments will just become prize less - or at least small prizes only. Maybe in Europe they will also become unrated.

_________________
North Lecale

But if a non monitored player is accused of cheating, what happens?

Evaluate the accusation to see if it has any merit. If so analyse that game (and others if those suspected too) for similarity to bots and if level of play is very unusual, both with reference to decent control groups (we can't do this well yet). If that provides sufficient evidence of cheating then forfeit and punish, else require future games of the suspicious player to be monitored. If their good results continue then that is evidence they weren't cheating before, if they stop it's evidence they were.

I would rather the PGETC continues with me being able to play at home, and lets say 2% of players will cheat and get caught and 2% will cheat and not get caught, than either: cancel PGETC or continue but I and everyone else have to go to some secure monitored location to play every game. I would be willing to buy and run a webcam if that was thought useful, so long as it wasn't too much of a faff (it wasn't fun being IGS technical support when I was captain). I don't know how widespread my view is, maybe others think cheating is already or will be much more prevalent, or are less willing to accept some may happen but continue with the event nevertheless. Perhaps a survey of participants would be a good idea.

I start with the handicap of not really understanding why anybody would want to waste three or so hours of their precious lives by creating a fake game with no money involved, but if I was forced to find reasons, I expect two would come out at the top of the pile.

One would be to ensure winning a match (e.g. not letting mates down).

The other would be a desire to play with their shiny new toy. I suspect this may be the dominant reason, but in any event both these reasons would seem to cover most cases.

Now at some point, probably not too far away, everybody will have this new toy. That means it will no longer be possible to ensure winning as it will just be AI versus AI, and, more significantly, the dull old toy will have lost its glamour. Just as no-one now tells us "Hi, honey, I'm on the train" in a loud voice (meaning "Hi fellow passengers, I've got a shiny new iPhone"), no go player will specially want the world to brag that he's got LeelaZero. Having AI thus becomes a bit pointless, except for entirely private study.

There doesn't seem to be much danger of an arms race in this field, so online cheating could more or less wither on the vine, while private study takes off.

How does that possible scenario fit what has happened in the chess world?

Look what event just popped up in my inbox:

https://www.pandanet.co.jp/event/iwag2018/index_e.htm