Cheating Allegations in AGA City League 2021

FWIW If this game had been reported on OGS as botting and fell to me for initial review (which many do) I would have said “I can see why you are suspicious, but this one game does not have enough evidence to convict, or really make it worth referring to the Dan mods for further consideration”, based on the usual checks that we do. I would have gone on to suggest to the reporter that if they want to pursue the matter, they should keep an eye on the the suspect’s games and report any others that are suspect.

Ironically, this person (6D bot reporter) made the same mistake SO MANY of our OGS Bot reporters make: they let the person know that they are suspected.

This is SO shortsighted. It takes time and often multiple games to get sure enough to convict someone of botting. And someone who bots isn’t going to do it in only one game ever - they will keep doing it, so there is time to nail them - as long as they are unaware.

So saying to that person “Hey, you’re a cheater”, or saying that to the world, before giving an investigation a chance is nuts. That’s the end of it … now they will cover up, and you will never convict them unless they left enough of a trail already.

So BTW if you want to convict this 1k, have a look at their earlier games as well…

… and if you want to report botting, for heaven’s sake keep it quiet while the investigation happens.

16 Likes

What this tells me, is they didn’t get a satisfying answer and proceeded to take part in the league.
So, less justified to make a stink. If they got no answer and they thought it was a deal breaker, they shouldn’t participate. Whining after the fact isn’t a good look.

The pro still doesn’t offer any proof other than “a pro knows”.

Even if they are right and the 1k cheated, they are wrong on how they go about it.

Both FB posts are focused on outing the other player in the harshest way possible, because they didn’t get their way with AGA.

The 6D seems young and I can sympathize, but the pro should know better.

2 out of 2, they didn’t get the answer they wanted and they lash out. The kyu is marked, but for me they are marked too, no sense of proper handling whatsoever.

2 Likes

Strictly speaking it is not possible to prove, unless you have video footage of the person cheating. So asking for a proof is like asking for the impossible.

In my opinion anyone who says that they didn’t handle it well, should also state how they should have handled it instead.

Regardless of how thorough an investigation you do, there will always be voices saying “that is not proof of cheating” (and strictly speaking, they are right). For example see Robert Jasiek’s post in this topic on lifein19x19 https://lifein19x19.com/viewtopic.php?f=10&t=17528.

Does this mean we should simply accept that we are powerless against cheaters? I think not. The difficult question is “How much evidence is considered enough evidence to convict?”. I am assuming that Remi Campagnie and Yoonyoung Kim hoped that the AGA would find the evidence conclusive. Instead the response was their team is forced to forfeit the rest of their games. I for one can’t fault them for reaching out to the public, instead of quietly accepting that.

3 Likes

I think the (Go *) world is only just now starting to learn how to deal with these things.

I’m not aware yet whether there are established protocols in the Pro level organisations, or amateur leagues.

At OGS I do know that we ourselves are having to work this out “as we go”.

One thing for sure: it sounds like they need some system that reassures the accuser that their suspicion is being treated seriously, and also (most importantly) allows for TIME TO OBSERVE the suspect.

It seems like this may have happened more successfully in the Kimmji case. From what I recall of that, they monitored and looked back over old games.

In contrast, it sounds like (as an outside observer) that this accuser got told “well, we looked, and they are not guilty end of story”.

That’s not going to be very satisfying, for sure. IMO a better approach would be “thanks, look we can’t convict now, but your report is taken seriously and we will monitor. Please give us time”.

This would have the double benefit of actually giving time to watch the suspect and time for the accuser to “get over it”, cool off a bit!

This is all easy to say of course from the outside - I have no idea what constraints they are up against.

But I sympathise - it’s not easy to deal with this problem, and it’s very very easy to criticise.

(That said, I do hope they are waking up and actively working on protocols, so these things go more smoothly in the future!)

*: Does the Chess world have this sorted, I had the impression they are “ahead” with this problem?

9 Likes

I made a whole post about it. So.

They raised the Webcam issue, then got in, then raised it again when they lost. That’s not caring about cheating, that’s caring about losing.

According to your logic, it’s ok to accuse everyone and let the ones that stick, stick.


EDIT

No proof and not enough proof are two very different things and blurring the line between them isn’t a good look either. It almost looks like you make your judgements based on what you want the outcome to be.

From the responses on the FB post (I’m on my laptop now)

*

*
This is AGA’s analysis

and what the pro says

“what I felt was ‘this is very high level performance, especially spending 20 minutes and beating 8d level on Fox…not following the graph, the movement (haengma) high level players can feel it. No Significant mistake, I feel like watching at least pro or yunguseng (insei)’s game. For a 1kyu to improve so fast in one year and half after having stagnated at that level for almost 10 years… all I can say is that we will welcome a new pro this year…”

She “feels” it, therefore it’s true. Not how things work, you know.

1 Like

Let’s say a mod tells me that Eugene, that I reported for cheating, isn’t cheating and that’s that.
Is it OK for me to come to the forums and say “Eugene cheated and beat me in a game, I reported it and I didn’t like the answer?”
I would guess it’s not OK.


My issue is everyone is having such a knee-jerk response to the slightest mention of cheating, they don’t care about being right, just to be punitive.

And this isn’t helping or moving things forward; this is just people looking for someone to put their frustrations on. Which, understandable, but let’s not name it “caring for what’s right”, because it isn’t.

3 Likes

I find it very hard to understand the decision of the tournament organizers. Not the part about whether there was conclusive evidence of cheating, but the part where they completely forfeited the whole Montreal team from the tournament because Rémi publicised the issue.

Is that really a proportional response?
Imagine they did, for the sake of argument, conclude in their investigation that Tong had cheated, which I see as worse than throwing public accusations, I would expect that they should only disqualify the cheater, not the whole Cincinnati team.
It’s not a calm, inconsequential measure. I’m sure many participants are now asking themselves if they want to play in this tournament again in the future.

I looked at the game to see how Rémi came to be so frustrated.
It is not unthinkable for a 1k to beat a 6d on a good day, so we cannot conclude cheating purely based on ranks.
AI’s recommendation will also vary quite a bit depending on software version, network version and performance settings. So it’s hard to say whether a person did cheat based on one fixed review run from e.g. AI Sensei or OGS AI review.

In my personal opinion, there is nothing in the first 70 moves that a real human 1k could not also play as black. White attempted to start a fight on the lower side and black refuted the attempt with simple, straightforward moves, putting white at a disadvantage.

Soon after, black puts down some really excellent stones. Not just once or twice - they just keep coming.

73 and 77 are just the most beautiful combination to play around white’s shape. The cut 77 had some Korean pros pause in admiration. I can feel that.

Sincere question to 1 kyus: would you see that 93 is the best point to counter-attack the white’s last move ogeima? Won’t there be a problem when white draws out the cutting stone above? Remember that black used almost no time to think.

There is a lot of talk about black’s mistake 105 (-80%).
This must be what Rémi means by being “toyed with”. He is allowed to capture the cutting stones at A and save his group. He even ends up ahead in the eyes of the AI!

The correct black move instead is A, the knock-out blow to capture the white stones and win the game easily. Instead, black will have to catch up the loss.

I can understand why black would defend. It’s too risky to let white play tsumego inside. I, for one, can’t read that out.

Speaking of tsumego, this is a really cool invasion. It lives with ease and elegance.


For some more comments, you should check out Montreal team captain Yoonyoung Kim’s facebook post.

Robert Tirak also made some comments. Taken at face value, it sounds like he figured out the exact method and settings of AI cheating that black allegedly did to find these moves in this game and several others.

I wonder how this compares with the AGA’s own analysis, and if the AGA will even consider additional evidence at this point.

11 Likes

Read the rules more carefully:

See Pandanet: Pandanet-AGA City League - Pandanet-AGA City League Rules Year 12

3 Likes

A simple google search might save you some time

If anyone is caught cheating the following will occur: The player will be banned from any Pandanet AGA City League tournament in the future. The team will be disqualified immediately. If the team was eligible for a prize, they will not be for this season.

https://pandanet-igs.com/communities/agacityleague/287

__

@jlt was faster :slight_smile:

5 Likes

Good point about the rules @jlt and @Gia. I honestly did not expect that because it is tough to be held responsible as a team for a single troublemaker. :sweat_smile:

1 Like

Publicly accusing someone without proof, after a judgement was made? I would say it is.

The public accusation shows disregard for the tournament and the organizers, disrespects the opponent and defames them, and also disrupts the whole tournament.

1 Like

I usually see it when it’s a team thing.

Captain of a team is responsible for the team, so also for how the members conduct themselves.
Other members are supposed to support and check their team-members, so it’s responsibility in their part.
It prevents a simple troublemaker ruining the thing for everyone and the team “crickets”.

I get your point.

In the real world, especially in the Go community, it is impractical to keep players with bad behavior in check. Every day there is a new gullible bleeding heart who just cannot keep themselves from giving them yet another chance. They can be exploited in rotation, because the disappointment and grudges only hold for a few weeks before they are once again open to the idea that the troublemaker has reformed.

This may sound cynical, but I really appreciate the sincere goodwill of my fellow Go players while also recognizing the weakness to narcissistic social engineering.

To counter-argue and try to do something about it is just such a constant uphill battle.
When I was captain of our PGETC team, I held to my standards of responsibility as you say.
There was definitely pressure and I totally understand just simply giving in.

So I still think punishing the whole team is draconian and I can’t agree that it “prevents a troublemaker” from ruining anything.

1 Like

I agree with everything about collecting evidence and not notifying the accused of being accused, but I must say, that if I were the moderator who dealt with such report, I would’ve considered move 105 as possibly a deliberate mistake, and imagined the hypothetical game where that one mistake was not made:

This looks like a graph of a 6d playing a 1k. Except that I’d say the 6d is playing with Black. Strong enough reason for me to ask mark or xhu to have a look at the game.

10 Likes

I don’t consider “let’s not all jump on the hate-train just because Yooyoung said so” a gullible bleeding heart. :stuck_out_tongue:

Thing is, everyone is projecting on this their own issues with cheating, while in my opinion the better response is the one that is fair, regardless of who we favor.

If indeed the 1k won fair and square, but their career is destroyed because of all the internet bullying, I’m afraid people feel it’s “collateral damage in the war against cheaters :fire:”.
But if the 1k actually cheated, the fact that an 8p took part in a city league and that the Montreal team let the webcam thing slide until it was about them (basically, they couldn’t care less if they ended up with the prize anyway), will go on unquestioned.

And I’ll stop now because Discourse says I talk too much :stuck_out_tongue:

1 Like

105 is one evidence the 1kyu possibly did not cheat. How in the world you folks rationalize the move against him?! :sweat_smile:

It’s neither evidence that the 1k cheated, nor evidence that he did not cheat. One move is not enough to explain anything. The whole game is barely enough for any conclusion.

What I’m doing instead, is considering 105 as a singular outlier and instead looking at the overall course of the graph: generally it’s a graph that very monotonically moves the game to Black’s favour between the start and move 190. That’s far more damning than the mistake of 105 is evidence for innocent: to play better than your opponent for almost the complete game is far harder than to make a mistake that loses you 30 points; anybody could do the latter, but only a high ranked player could do the former.

Of course it works similarly in the other direction: anybody could play a “great” move (some do it by knowing, others by accident), yet it’s very hard to consistently do so (since one cannot consistently play good moves by accident). This 1k was consistently playing better moves than their 6d opponent.

7 Likes

I think one way that would be a great detector of AI cheaters, is to let the suspected player explain how the game progressed, and what they were considering during the game. Ask them what other moves they considered, why they played tesuji X instead of sneaky mistake Y, etc.

When someone recites a textbook, I could not know if the person knows what they’re talking about. If instead I ask them to summarise the textbook in their own words, it becomes painfully apparent when they don’t know what they’re talking about.

10 Likes

Thats actually a good idea I wonder how that would work though on a good improviser.

1 Like

I think if he did cheat. It was not AI, but another high strength human player. That would explain the move 105 and 20 minutes.

Very likely Team Canwa Vancouver 1 in A league cause they didn’t want Montreal rise to A league, especially dislike Montreal’s recruitment of a 8P. LOL

1 Like