How effective are the 4NCL Online’s Fair-Play Guidelines at preventing cheating? Sceptical? You’re not the only one!
The 4NCL Online’s FP guidelines certainly recognizes and its rules prohibits a number of distinct types of cheating. This is a quotation from “#2 General Fair Play Guidelines”  listing examples of prohibited behaviour:
- Help from any other person, player, or coach;
- Impersonation/use of other people’s accounts;
- Use of chess engines, bots, plugins, analysis tools, tablebases, blunder checks or other best move tools;
- Artificially inflating or deflating ratings by intentionally losing, or arranging with an opponent to win;
- Interference with other members’ games;
- Reference to physical opening books and static (i.e. non-engine based) tablebases.
No question about it: the intention here is good. The 4NCL are 100% right in calling time on any and all of the above. But there is a distinction between proscribing an activity, and taking active measures to outlaw it. So, how good is the 4NCL in monitoring, detecting and thereby deterring these type of cheats?
Undoubtedly, the 4NCL puts in a serious effort into preventing engine-assistance. (We’ll come onto that in Part 2.) But how about the other types of cheating? Let’s take a couple of examples.
Allow me to tell you a little about my home environment in which I participate in 4NCL Online matches. I always play on an iPad Air 10.5”. The touch-screen sensitivity is just perfect for LiChess: so important in the nerve-racking end-of-game time-scrambles to which I almost always choose to subject myself. Next to it is a XPS13 laptop with a CB15 instance containing MB2020 (obviously), all my games since the 1990s (mainly annotated) and all of my opening repertoire (mostly engine checked). However, like most chess-players, I also possess several hundred books on the game (although I’ve – almost – now managed to give up buying these). All of these are on book-shelves surrounding the desk on which I play.
Do you think it might be useful to access this intellectual collateral during a game? I’d even put a figure on it. I’d say that the advantage of swiftly and accurately churning out machine-checked or GM-author-checked repertoire in the accelerated 4NCL online time-control is worth about 100-150 ELO points. Chances of detection under 4NCL FP procedures? Negligible. How could one possibly tell the difference between this type of cheating and thorough (or just plain lucky) preparation?
[There is an excellent demonstration of this point using the OTB game Arnott-Thomas 4NCL 2020. See Jon Arnott “The Numbers Game” Chess September 2020]
Here’s another example. FM Martin Walker has been our board 1 for much of Season 1, and I am sure that everyone in the squad will be pleased to hear that he has re-enlisted for Season 2. Martin and myself have almost purely complimentary opening repertoires. Martin plays only 1 e4 as White. I have almost never played 1 e4 for the past 3 decades. As Black, Martin plays the Grunfeld against 1 d4 and Open Games and the occasional Alekhine against 1 e4. I play exclusively King’s Indians & Sicilians.
Let’s suppose that Martin & myself are playing in the same team in a 4NCL online match. How about if Martin were to play using the LiChess account designated as mine, while I play using Martin’s account? (This would mean that Martin’s opponent would actually be playing me, but think he was playing Martin: whereas my opponent would be playing Martin, but think he was playing me.) It’s difficult to quantify the advantage gained by this switcheroo. But it would certainly side-step the majority of our opponents’ preparation. Add in the surprise factor: surely it’s got to be worth something?
I have had the pleasure of knowing Martin since 1983. If we felt that impersonating each other in a chess match was an acceptable thing to do (which, for the record, we don’t) it would be exceptionally easy for us to arrange a simple exchange of LiChess sign-in details. No one else would ever know. Martin & myself have approximately the same rating, so its unlikely any anomalies would show up on Professor Regan’s celebrated “Z-scores”. (We’ll come onto these in Part 2.) Chances of detection: virtually nil.
I could labour the point with further examples, but you get my drift. There are just so many different types of cheating which while problematic, not to say downright risky, to execute in OTB games, are ludicrously easy in On-line Chess. While prohibited, these are evidently undetectable by either the 4NCL’s original or “improved” FP Guidelines & policies.
Does the 4NCL understand the size of the hole in their anti-cheating procedures? Take a look at the lessons learnt on anti-cheating online following Season 1 by 4NCL Chief Arbiter, Alex Holowczak . It’s a nicely written piece, and (as far as I can tell) gives an honest account of his own perspective on the anti-cheating controversy. It’s just a little strange that what purports to be a rounded view on cheating in the competition only ever talks about one type of cheating.
I am pretty sure that the 4NCL are well-aware of the potential problem, since one approach to handling myriad types of cheating is even mentioned in #3 “Anti-Cheating Measures” of the revised FP guidelines:
4NCL Online has considered additional measures for detecting cheating such as requiring webcams (as FIDE have done with a number of their events). There are no plans to implement such measures because the gains are felt to be marginal and would be outweighed by the practical difficulties of implementing such a system. 
If one is seriously concerned about cheating, video monitoring seems to be something that is worth looking into: perhaps by doing a “proof of concept” using 1st Division teams, or the critical Championship play-off matches. Martin & myself don’t look remotely like each other: so it would put an immediate end to the switcheroo wheeze. Even the most brass-nosed cheat would be reluctant to hazard look at opening manuals intra-game: knowing that every eyeball movement is potentially under scrutiny.
To be frank, I don’t follow the reasoning here for dismissing this idea out of hand. The claim that “the gains are …. marginal” is certainly under-evidenced, especially given that (aside from engine-assistance) there does not appear to be any effective monitoring of other types of cheating. The reference to “practical difficulties” sounds to a nation grown accustomed to Zoom calls during the Covid-19 lock-down like nothing more than a lame, technophobe excuse for inaction.
Where does this leave us? The 4NCL Online provides active, systematic and indeed (as we shall see in Part 2) aggressive monitoring, detection and sanctions against one type of cheating: but operates a laissez-faire approach to almost all others.
Is there anyone outside the 4NCL to which this policy makes sense?
[to be continued]
(c) AP Lewis 2020