I am happy to report that my and Attila Egri-Nagy’s research ‘Derived metrics for the game of Go – intrinsic network strength assessment and cheat-detection’, which I linked on nordicgodojo.eu roughly a month ago, just passed peer review and will be presented in the CANDAR’20 symposium.
Given that we got promising early results and that the subject is potentially critically important for online go, I will do my best to continue researching the subject; most likely the next step will be to see how accurate my current analysis method is when faced with a larger number of game samples that randomly involve cheaters and non-cheaters.
Parallel to this research, me and Mr Egri-Nagy are also researching efficient ways to use the ai in go studies, which largely prompted me to write yesterday’s article. Mr Egri-Nagy has recently started a Youtube channel where he has uploaded ‘perfect’ self-played games by KataGo; ‘perfect’ in the sense that, with the huge number of playouts involved, KataGo cannot find a way for either player to get a better result at any point, and finally the game ends in a tie (with integer komi). Therefore, the ‘perfection’ is with relation to the playing entity (KataGo) at hand. Arguably, if we are to study and memorise games by strong players, these kinds of ‘perfect’ games should be the best possible material – that is, if the games involve ideas reproducible by human players.
All times are in Helsinki time (eet with summer time).
League standings and pairings updated at the start of each month.
Public lectures on Twitch every 2nd and 4th Saturday of the month at 1 pm.
Jeff and Mikko stream on Twitch on Fridays at 6 pm.