The title is based on “Elo World, a framework for benchmarking weak chess engines”, and the talk (slides) was timely amid the Carlsen-Niemann saga. The two main ways of modelling performance of human chess players are discussed, with modifications for detection collusion or cheating.
From the analysis of Ken Regan, it was inconclusive there was cheating in the recent saga. It is quite hard to detect cheating in chess anyway as
- the form of cheating has to be known beforehand to those doing the analysis, and stayed the same throughout, and
- cheating only needs to take place at certain critical moves to give the player an advantage, but a correct / good critical move could be attributed to how good the player is.
From 2, detecting cheating of an average player then possible comes with less certainty.