After the 10 game match between my ChessGenius Pro 2024(MCGP2024) and the Chessnut Evo Maia bot (level 9 and 6ply search) at active chess where the Maia bot achieved a rating of 2269, I decided to see how well Maia would perform against TheKing using it’s normal personality and HIARCS book at the same time control. Since my match with the MCGP2024 the Evo has been updated and now has an opening book. This means the Maia bots will no longer always open with P-K4 when playing White. However, I’m not sure how extensive the opening repertoire is and there is no control over this book, so it cannot be turned on and off for example. That being said, it now means the Maia bots provide a good opponent for casual to advance players with varied openings.

I've been trying to find an ELO rating for TheKing at Active chess but the Wiki ELO list only has ratings for tournaments with a time control of 2 hours for 40 moves. So I will calculate a ballpark rating for TheKing with the Normal personality and using the Master (HIARCS) book from the match that I'm currently writing up and will post as soon as possible.

Of course the result is one sided as would be expected considering TheKing is a very strong opponent. However, I think you will find the results to be fairly interesting.

Before the match started I wondered whether it was fair to only give TheKing 30 seconds per move considering that the Maia bots typically respond immediately or within a couple of seconds so depriving TheKing pondering time. This of course was the case with my match with the MCP2024. Considering the Maia bot has data from 12 million games to draw on whilst TheKing has to use programmed knowledge and look-ahead to work out a move it's difficult to know exactly what is a fair time control. Maybe to make TheKing move as rapidly as the Maia? Something else to try maybe. Either way, despite TheKing not always making sound moves, the Maia bot made some blunders throwing games away.

I'm using HIARCS ChessExplorer and DeepHIARCS for the game analysis. The games are complete just need to write up the results so watch this space.

I have just posted my comments on the last game of my 10 game match between TheKing’s Normal personality using the Master (HIARCS) book at average 30seconds per move. I knew the result would be one-sided but this match was about seeing Maia’s playing ability against a very strong opponent.

The final result of 8 ½ - 1 ½ to TheKing puts the latter at around 300 ELO points stronger in Active Chess for this match. In my match with Maia L9 against MCGPro 2024, Maia achieved an approximate ELO of 2269 so this make’s TheKing’s rating to be roughly around 2560 in this match.

I think the games show that Maia on Level 9 with 6 ply search (the current maximum setting) is a good opponent for many. The occasional blunder could be seen as an advantage making it a less robotic opponent and giving us more of a fighting chance!

I have said before if we use Chessnut Evo’s Maia with anymore than 0 ply search then it should be called Hybrid Maia as this is moving away from the concept behind this Neuronal engine which is to not always make the best move – which is actually still the case – but the more human move (this is debatable in my opinion when it makes too many severe blunders) with zero search. In my testing, Maia with zero search is very poor and whereas it may sometimes make the more human-move, it’s multiple severe blunders ultimately make it a play like a complete beginner. Now that it can search to 6 ply it at least avoids many blunders but this limited search still somewhat hampers it especially in endgame scenarios. Overall though, I think the updated Maia bot on the Evo is an interesting opponent for human play.

Regards
Ray