I copied this and kept it because I had a feeling this was going to happen (after this post)

My 2 Cents, "Probably" Won't Have an Impact

· WhipLash775

· | Jan 30, 2016 at 12:10 AM

· | Posted in: WhipLash775’s Blog

· | 8 reads

· | 0 comments

I retired 7 December 2014. One of the first things I did was turn in my work phone and buy a new cell phone. It had all these apps and gadgets and stuff. Since I now have plenty of time, I decided that I was going to get back into chess seriously (again); instead of just endlessly passing time playing blitz games. So I looked for chess apps on my phone. Low and behold, I found Chess.com. Over time, I started to get more and more into it, and even became a paying customer.

I've been on a few chess sites over the years and feel that I am qualified to critique a little bit. Besides I am a paying customer; and this is America (at least where I am at), and I can say what I need to within reason.

I've observed some very good things about Chess.com. It's a huge community with lots of titled players. Great articles and news updates. Games of all kinds available anytime, day or night. All kinds of cool features like game explorer and what not. Really, too many good features to mention. I thank Chess.com and their staff; and I applaud them. The only detractors that cross my mind are that I think the live interface could be much better and I don't care for the tactics trainer much. I've seen better and faster live interfaces at other sites; but then they do not even come close to all of the other features that Chess.com has. The reason that I don't like the tactics trainer much is that on a lot of the puzzles, there is certainly more than one right answer, yet you only get credit for the one answer they have determined is correct (no partial credit either). On top of that, if you don't solve it in time it's as if you never solved it all; it's simply a fail. I don't like it, I can think of at least one or two sites right off the top of my head that are better in that regard. But overall, Chess.com is awesome. Version 3 has great potential.

There is one thing about Chess.com I have observed, though, that I absolutely do not like. I have never been on a site where cheating is such a hot topic. I have never been on a site where so many players get banned (presumably for cheating). (A couple of recent ones were "friends" of mine) Part of this may be due to the shear size of the community. However, Chess.com and the associated community have all aspects of cheating represented; and represented fully.

You have your cheaters and yes they are out there. You have your cheat detection. You have your players being banned; left and right. (mostly based on "probability", not hard evidence) You have your appeals (well sort of). You have your slueths and snipers, who spend most of their time investigating and reporting on other players versus playing chess. You have your guys that cry wolf every time they lose a game. You have your guys that witchhunt or scapegoat for personal reasons; using cheating accusations to discredit; because it's an easy, convenient method of attack. You even have your titled players that refuse to take a loss under any circumstance; that know they will be catered to by staff. Chess.com has it all. To me it is all very disheartening. Can't we all just play chess? I'm about at the point where I don't even want to play rated games here any more, just unrated and maybe not even then. So far, I'm probably not telling you anything you don't already know; but some recent events got me thinking; flashing back really. Back to 2006.

The entire discussion of cheating and computer assistance (the modern version of it anyway) began in 2006, during the world championship match between Kramnik and Topalov. Just before game 5, Topalov's team lodged a protest, insisting that Kramnik was using computer assistance (from Stockfish); and recieving information when he went to use a private bathroom during the match. It became widely known in the chess world as bathroom gate. Specifically, Topalov's team was trying to assert that Kramnik's moves at the board during the games matched 100 % with the computer chess engine output and that he was cheating. Nothing definitive was ever determined. Kramnik went on to win the match and once again unify the World Chess Championship Title; and as they say the rest is history. Or was it?

Then followed a massive amount of games being analyzed all over the world by computers; first in an attempt to find the best lines and improve play, and second to determine how often the moves of a player matched with the engines output. The big question was and still is; if the players moves match with the computers; is he cheating and using computer assistance? Would that particular player have naturally found the move on his own?

In an attempt to answer this question, the very esteemed Dr. Ken Regan went to work crunching the numbers as to just what the "probabilities" are/were. I have absolutely nothing against Dr. Regan. I have the utmost respect for him and his work. However, from a common sense perspective, I do have a problem with "probabilities" in this instance. Simply put, just because something is mathematically probable, does not make it certain. If probabilities were that accurate and controlable, we could all go to Las Vegas or Monte Carlo and make huge sums of money. We certainly know that not to be true. Simply put, probabilities are not hard science or direct eveidence. If we used this type of methodology in our criminal justice system (if it walks like a duck, and quacks like a duck, it must BE a duck) we would be falsely imprisoning and executing people all the time.

Yet, these fancy algorithms are being sold to the world at large as gospel. Algorithms that assume we have perfect knowledge of chess (which we do not and we may never have, considering the near infinite nature of the game). Algorithms that are based on the assumption of having perfect knowledge (down to every legal move) and how a human player will "probably" assess and act on the situation. Never taking into account quantum theory of learning or anything else possibly explainable. Dr. Regan asks "What is gods rating"? Then admits we don't really yet know, but we are getting close. We only need to fine tune the algorithm some more and we will have it for sure. I posit that what we are actually looking at is very fancy mathematical assumption and not fact. Why does the algorithm constantly need fine tuning, doesn't the model work? Yet based on this type of fancy assumption, decisions are being made, peoples accounts are being closed, reputations ruined, discredit is being levied and various shenanigans as mentioned above.

On top of all that it is essential that everything be held in secret by administration for national security purposes; so that the actual real life cheats (and everyone else for that matter) can't figure out what is going on. When an account does get closed, you have no idea whether there was actual eveidence to support it or it was arbitrary and capricious. Just administrationsarrogant assurance that it is right. You have no idea whether the person was in a live interface, running something in the background (which is easily detectable), with consistent computer like move times and a move list 100 % matching the computers (i.e. actual eveidence) or they were simply banned based on "probability" in say a series of correspondence games. Just because we say so. Zero transparency; and quite frankly a bit of arrogance.

Lack of transparency also lends itself to the possibility of being very arbitrary. Remember those Titled (entitled) players I mentioned earlier? Whether a titled player is involved or not, I believe some arbitrary solutions have "probably" (there's that word again) been enacted in the past. I've never seen a business operate in this manner; especially one that involves paying customers. Strange. Chaotic. (Over a game no less.) (A game that I do love and respect.)

With all of this in mind, I decided fine, I'll commision my own study. I figured if the the sleuths and the snipers can do it (on a daily basis), why can't I ? My study is not based on probability, however. My study is based on pure observation of the naturally occurring and actually existing. It is observable, recordable and repeatable. In other words, its science. I am not offering any conclusions as a result of this study, I am simply presenting it. It is what it is, it stands on it's own. You can repeat this experiment yourself if you so desire. I leave it to you to draw your own conclusions at the end of the presentation.

The method of observation I employed involved using a Stockfish chess engine similar, but more modern than the one used by Topalov's team back in 2006. I used that engine to examine five different sets of games to determine what exactly was the match rate between the players/groups with the computer in each case; and to compare those rates against each other.

Note that I am only concerned with whether a move matches a book/computer move or not. I did not exclude book moves. Book moves represent "perfect" or known play, by both the computer and the player. The computer has access to the book; the player either has access to or has memorized the book. The point is whether the two agree; is the player doing what the computer would do or something different ? I also discarded the computers percieved quality of the nonmatching moves (whether the computer percieves the move as an inaccuracy, mistake or blunder). There is no extrapolation of the percieved probability of the player finding that move either. The players move either is a match to the book/computer move (in other words what Stockfish would do) or it is not; and what is that match rate? I am also not concerned with "forced" moves. Forced moves are brought about by introducing certain lines of play beforehand. Both the player and the computer have choices before engaging in that particular line which introduces forced play. Again, did the computer and the player agree on that line of play (move) or not ?

The five groups are as follows:

Group # 1 (11 games from the 2006 Wchp Kramnik and 11 games from 2006 Wchp Topalov)

I did not include the 4 tie break games which were played at rapid 25 minutes + 10 seconds.

The quality was not as high; that was a hell of a way to determine a world champion.

What was their individual and collective match rate with Stockfish?

Group # 2 (20 games from 1972 Wchp Fischer and 20 games from 1972 Wchp Spassky)

What was their individual and collective match rate with Stockfish?

The reason I included this group is that it is noted for it's high level of play and we can reasonably assume that no meaningful computer assistance was available at the time. Unless the CIA or the KGB had something no one knew about. It's unlikely.

Group # 3 (The last 10 games each from the top 10 correspondence players on Chess.com)

100 games total, the games have to be meaningful, rated and of sufficient length, each of the 10 players must contribute a minimum of 200 total moves to the sample.

Group # 4 (The last 10 games each from 10 known, banned cheats from Chess.com)

100 games total, the games have to be meaningful, rated and of sufficient length, each of the 10 players must contribute a minimum of 200 total moves to the sample.

Group # 5 (The last 10 games each from the top 10 players that are most active/vocal in the Chess.coms Anti-Cheating Forum)

100 games total, the games have to be meaningful, rated and of sufficient length, each of the 10 players must contribute a minimum of 200 total moves to the sample.

I included this group for a couple of reasons, 1. Just for fun so don't get upset. 2. These are the guys that are generally above reproach, and rightfully so 3. It makes a good comparison, might illustrate what is normal. No names of any players from any of the groups active or otherwise will be shared, just numbers.

Group # 1 (2006 World Chess Championship)

Kramnik made 563 moves, 545 matches for 96.8 %

Topalov made 565 moves, 545 matches for 96.46 %

Collectively they made 1128 moves, 1090 matches for 96.63 %

Group # 2 (1972 World Chess Championship)

Fischer made 906 moves, 866 matches for 95.58 %

Spassky made 907 moves, 864 matches for 95.25 %

Collectively they made 1813 moves, 1730 matches for 95.42 %

Group # 3 (The top 10 correspondence players)

Collectively the group made 3,666 moves, 3,543 matches for 96.64 %

Group # 4 (The Ten Little Indians; AKA Banned Cheaters)

Collectively the group made 3031 moves, 2881 matches for 95.05 %

Group # 5 (The Anti-Cheaters)

Collectively the group made 2984 moves, 2619 matches for 87.76 %

Admins of "The Not Wanteds"

Super Admins

Admins