“Decision: case of using computer assistance in League A”

[theoldway]

Now that I have a PC with gpu I let Leela analyse for 2.5m sims and q15 did rise from choice #18 to #5. So Leela did start to think about it more but was still far off local moves around the cutting stone (net and p7 peep top iirc).

2.5 m sims seems a huge amount of time to me. How many minutes it will take to calculate 2.5 m playouts? I guess that the player took much less time to play that move. didn't he?

It seems to me that q15 sequence worked only because of his opponent cooperation. Black should extend on M10 at some point in the process as suggested by Leela. The game would have been close this way.

[Gobang]

I have not read this entire thread, so forgive me if my words are a repeat.

Given what has clearly happened, a well founded suspicion of a player using computer aid in an online tournament, has a profound meaning for any future online tournament whether the suspicion is true or not. The implications are:

1. A player will not be able to earn a rank, money or prizes from playing in an online tournament, given that there is no sure way to determine if he is honest or not.

2. In the future any online tournament will be played for fun with a gentleman's agreement not to cheat.

3. Any results will be considered in a lighthearted way and they will not be assigned any deep meaning. As such they will be quickly forgotten.

4. Face to face tournaments will hold the status of being the only meaningful Go tournaments. (Cheating in such a tournament is still possible but it must be much harder and less likely)


Regarding players who are guilty of bad manners and poor sportsmanship:

metta [4k]: i'm sorry but i'm used to play go not this disgusting game, please when you'll learn to play this game inform me so i can take you off of my censor list

In my opinion such players should chose another game. They are not fit to play Go.

[RobertJasiek]

What is all this "the Go world is breaking apart" talk? That there are different opinions on one topic does not mean that the Go world would break apart.

[RobertJasiek]

It is also the account of a very frequent escaper. Not that this says much about the cheating case but if this is/was indeed Carlo Metta's account, the idea that such a player can be a referee during the EGC is disturbing/laughable...

We do not need rumours and speculation. Was it or was it not? What is the evidence of very frequent escaper? And do not repeat the cited talks, which show nothing but different, slightly heated opinions on whether undo is good.

[jlt]

What is the evidence of very frequent escaper? And do not repeat the cited talks, which show nothing but different, slightly heated opinions on whether undo is good.

See game archives here. There are many forfeited games (games are marked as forfeited when they are escaped by a frequent escaper).

Note however that

1. We are talking about old behaviors. People shouldn't be blamed for past behaviors if they have improved since then.

2. Being a sore loser (or having been one in the past) doesn't make you a cheater, whether in go or in other games. Arguments like that should not be taken into account when accusing someone of cheating.

[Javaness2]

So the new evidence needs new work before it can be treated seriously.
For point 1, I think it would be interested to plot rating performance per online event, and per offline event. Of course, you need several unbiased counter examples.
For point 2 lots more data needed to make a serious point. A huge work.
Point 3, I am satisfied with Carlo's explanation as to this point. In particular, I have to imagine that this game was already analyzed, so I don't see why it would be hidden.

Regarding #1 and #2, I suppose a lot of us have a morbid fascination with this case, but I think the effort is not worth it. Even if there was cheating here, I can't believe it was done in so stupid a way as to leave a clear trail behind. Each time new evidence is badly presented, it weakens the case.

[Gobang]

I believe the discussion of guilty or not guilty with respect to an individual player is not of much value. People can and will make up their own minds.

Computers have become much stronger than human beings. This is a fact. In my opinion online tournaments become casual affairs with no meaning beyond the pleasure of those who participate. I don't see any means to police the use of computer assistance in such tournaments.

[John Fairbairn]

There has since been a new case of a very lowly ranked Italian, with a previous history of alleged match-fixing (according to the BBC), causing a major upset over one of the world's strongest players.

Marco Cecchinato, ranked 72nd in the world, has just (June 2018) defeated tennis superstar Novak Djokovic in the French Open quarter-finals.

Cecchinato had never before won a main-draw match in a Grand Slam before Roland Garros.

With other superstrong players crying their eyes out in the locker room and crying foul, umpires are urgently trying to discover where Cecchinato is hiding his AI, but the skimpy shorts are making this difficult. Nevertheless, repeated cries of "New balls, please!" show where their suspicions lie.

[Uberdude]

Now that I have a PC with gpu I let Leela analyse for 2.5m sims and q15 did rise from choice #18 to #5. So Leela did start to think about it more but was still far off local moves around the cutting stone (net and p7 peep top iirc).

2.5 m sims seems a huge amount of time to me. How many minutes it will take to calculate 2.5 m playouts? I guess that the player took much less time to play that move. didn't he?

It seems to me that q15 sequence worked only because of his opponent cooperation. Black should extend on M10 at some point in the process as suggested by Leela. The game would have been close this way.

2.5m sims of Leela 0.11 on my PC with a GeForce 1060 takes ~9 minutes. And yes Carlo generally plays pretty fast so doing that many sims live would be impossible (unless he had access to some monster computer in the cloud). A previous user (and myself) reported that Leela didn't find the q15 push, but wondered if it was due to insufficient resources. So I am answering the question that, even with millions of sims, it is only choice #5. Thus we have evidence that this move, which Lukan considered the most notable example of suspicious strength in the game (I didn't consider it so unlikely a good 4d could find it), is not a suggestion of Leela so is evidence Carlo wasn't cheating with Leela, at least at this point in the game. As you (and I previously) say if black had displayed more fighting spirit and extended at m10 then the game would have been more complicated and Carlo not been able to win so smoothly.

[Shenoute]

It is also the account of a very frequent escaper. Not that this says much about the cheating case but if this is/was indeed Carlo Metta's account, the idea that such a player can be a referee during the EGC is disturbing/laughable...

We do not need rumours and speculation. Was it or was it not? What is the evidence of very frequent escaper? And do not repeat the cited talks, which show nothing but different, slightly heated opinions on whether undo is good.

You have apparently not check metta's KGS account. I did, so please refrain from talking about "repeating cited talks".

August 2014
37 rated games
5 forfeited games won by metta's opponent
4 interrupted games in the last 10 games

It seems safe to conclude that metta escaped at least 5 games out of 37, and maybe up to 9. Is this enough to use the words "frequent escaper"?

I agree that this should not be of any use when discussing the cheating accusation but since Carlo Metta's behaviour has been used as an argument since day 1 (see the facebook post quoted in post#1 of this thread) I think it is only fair to point these things out.

[theoldway]

2.5m sims of Leela 0.11 on my PC with a GeForce 1060 takes ~9 minutes. And yes Carlo generally plays pretty fast so doing that many sims live would be impossible (unless he had access to some monster computer in the cloud). A previous user (and myself) reported that Leela didn't find the q15 push, but wondered if it was due to insufficient resources. So I am answering the question that, even with millions of sims, it is only choice #5. Thus we have evidence that this move, which Lukan considered the most notable example of suspicious strength in the game (I didn't consider it so unlikely a good 4d could find it), is not a suggestion of Leela so is evidence Carlo wasn't cheating with Leela, at least at this point in the game. As you (and I previously) say if black had displayed more fighting spirit and extended at m10 then the game would have been more complicated and Carlo not been able to win so smoothly.

What about other strong engine suggestion? At the time Leela Zero was not on the scene, but Zen 6 or 7 was already there. Anyone here can take a look to Zen suggestion? If I remember correctly Zen suggests only its first 5 moves

[bugsti]

So the new evidence needs new work before it can be treated seriously.
For point 1, I think it would be interested to plot rating performance per online event, and per offline event. Of course, you need several unbiased counter examples.
For point 2 lots more data needed to make a serious point. A huge work.
Point 3, I am satisfied with Carlo's explanation as to this point. In particular, I have to imagine that this game was already analyzed, so I don't see why it would be hidden.

Regarding #1 and #2, I suppose a lot of us have a morbid fascination with this case, but I think the effort is not worth it. Even if there was cheating here, I can't believe it was done in so stupid a way as to leave a clear trail behind. Each time new evidence is badly presented, it weakens the case.

I agree with you. Fisrt we need to establish solid method and anti-cheating technique otherwise any unclear attempt will weaken this case and also other future cases.

[bugsti]

A little off topic but still related to future regulation: they say that Leela Zero runs on Android now!

https://www.reddit.com/r/baduk/comments ... n_android/

I wonder how this will affect real tournament conduct. Referee will need extra work and training

[Uberdude]

<moderator hat on>
Regarding Lukan's screenshot message on Carlo's character.

The forum rules say

1. Personal Attacks
Remember the Golden Rule: do to others what you would like to be done to you. Trolling, flaming, bashing, or otherwise verbally attacking anyone is not allowed. When writing about a move that a member made, ensure all criticism is directed towards the member's move rather than the member. Also, please consider that not every member in the forums is a native English speaker as you write your post.

Does that post fall foul of this rule? I don't know, but I seem to recall we have removed milder personal criticisms. That post was reported to the mods, and also became pending approval (I'm not sure exactly how this works, sometimes (but not always) posts from new users are hidden until a mod approves them as an anti-spam measure, maybe being reported made it pending approval to show again). I messaged the author, saying that given the nature of the post I thought he/she should identify themselves so people can better judge if it is a credible witness or just some troll out for character assassination. They preferred to remain anonymous, so were okay with removing the message (or rather now, not approving it).

If this were a legal case, would this be "inadmissible evidence"? Should the moderators suppress it? I suppose lots of people have seen it now so the point is moot, but I hope this clarifies the situation.
<moderator hat off>

[Gobang]

I wonder how this will affect real tournament conduct. Referee will need extra work and training

No training needed. Anyone caught using a phone during a game is simply disqualified.