@hicetnunc - Yes, it's a good thought experiment. But what makes it convincing is the 1600 has an OTB rating and OTB games to examine. This is already more evidence, and more kinds of evidence than just the algorithm applied to some online moves. I would find the OTB games especially persuasive, as this is an apples to apples comparison.

@Confused_by_Theory - "Borderline cases will exist". Yes, absolutely, this is an outstanding point. No matter where the border is placed, there will be borderline cases.

@MNb - Relating to DNA evidence, of course it's not 100% reliable and of course we don't reject it for being less than 100%. But is it literally the only piece of evidence against the accused? Like: we found some DNA on a murder victim, we matched it to you based on a database, you are going to jail. I know many people would say, "DNA, you did it", but I hope that in a court of law there would be additional evidence.

@everyone - Please don't take my attitude to mean that online players can't be banned. I actually don't have any problem with players being banned for "strong suspicion of cheating", which is all that I think an algorithm can provide -- as long as the ban is narrow. Meaning, we don't ban a player from OTB events for suspicion of online cheating. Meaning, if we can meet a rigorous legal proof of cheating, then a broad ban is justified.

So my ideal policy would be, in the "terms of competition", it's specified if the algorithm says you are using a computer, you are penalized in this event, or on this server, or by this organizer, or whatever the terms say. National federations and FIDE should not be extending these bans, only in the case of the national federation or FIDE is the organizer should they be applying a ban. And ban only online play.

But if you are brain-dead enough to, for instance, leave your webcam on and thus we have proof beyond a reasonable doubt, then the sanctions can be broad in scope, e.g. banned from OTB events as well.

hicetnunc wrote on 08/27/20 at 08:00:59:

Wrong question; I refer again to the example of DNA and fingerprints.
The correct question is what margin of error we think acceptable. For that question your extreme example isn't helpful.

Hi.

If you in theory have one algorithm checking moves against the computers chosen moves, another checking time management for patterns and say one checking how long before games deviate from previous games. Then surely each can be considered evidence.

If all these three algorithms were meshed into one and all given output was some numerical representation that got dubbed probability of cheater, then I get how that might be troublesome to trust blindly in a legal sense. Some understanding of the algorithm used is probably needed or maybe some kind of output that is detailed and not a single numeric value.

Let's save that for the big questions.

Jupp53 wrote on 08/26/20 at 17:21:42:

Then I suppose you reject DNA comparisons and fingerprints too. They only provide such probabilities, even if much smaller than your 1 : 1 000 000.

Hi.

I guess working with probabilities is generally not complexity free. Is there not some legal principles that if there is uncertainty don't give the maximum penalty though?

Have a nice day.

Thanks!

Another thing that may have already been covered (but I missed) would be:

After a valid tournament, I have just made my norms for Candidate Master. Hooray! But now, it turns out when I finished at 11 points @ 10-0-2 that the 2 draws were against a player who has since been D/Q'd for cheating. So am I now only 10-0-0? Could this mean I don't qualify for CM anymore?  Or if I won the tournament but now I only have 10points, but some other player has 10 1/2 hence, I lose the tourney as well! So now, I not only don't make my norms but I have to hand over the prize (if any) as well? Very goofy premise, yes. But if cheating is indeed becoming more common than this will become relevant sooner or later. 

Do any here know how such things are handled? Or is this still in unknown territory too?

Just curious,

Andrew

P.S. The FIDE regs give this:

Quote:


But I assume those are for the offender, not their victims.

Bobby Fischer famously accused the Russians of collusion against him, an accusation that is often reported as later admitted to having been accurate. Former Director Fidomanes has had similar accusations made against him regarding tournament shenanigans.

What's the point re: computer-assisted cheating? Because I am less concerned with how many decimal places the "statistical estimate" is than the possibilities of abuse of said tools when a human Arbiter has the final say. For example; If the tool gives a 99% probability, but cannot be enforced except by a FIDE official who can arbitrarily decide, "Naw! that guy's OK," then just decide to ignore the results.( No matter he's a Fav of FIDE and not some protesting, trouble-making fellow like Kasparov!) Well, that worries me more than the tool itself. IOW, not only is there an old joke about statistics ("garbage in, garbage out") but also in any chain of authority it is man that is always the weakest link. Remember this story:

https://www.youtube.com/watch?v=S2JedouTRZo

Just sayin"!

Andrew