I’ve trained a convolutional neural network that beats our long-standing equity engine (HastyBot’s static evaluator) in direct play - 51.7 % win rate over 250K self-play games - making it, as far as we know, the strongest pure static evaluator in Scrabble today. In this post I walk through the feature set (board image, rack, bag, tempo, power-tile counts, more), the training pipeline, results, and why a value-only CNN is already changing how we build BestBot’s Monte Carlo search. Code, weights, and data scripts are all open-sourced for anyone who wants to tinker.

I just released a new web app for learning words called WordVault. Read more about it here.

Arch Linux is the main system on my desktop computer. I love it but sometimes it screws up. Here are some tricks for unscrewing it up.

Scrabble endgames are computationally difficult to solve exhaustively. I used some chess programming techniques to greatly speed them up.

When I was 11 or so my friends and I snuck into Sesevenen (Se7en). I remember being excited that this was an R-rated movie and that we were going to see some cool violence or whatever. However, my main memory of the movie was that it was a disappointing slog. I think at the time I wanted something like an R-rated version of Independence Day, a movie I enjoyed very much. Or stylized violence ala Tarantino, not some film noir I didn’t really understand.

Weezer’s Falling For You is an amazing song

Linux provides the best and most cost-effective desktop experience.

A few months ago, I noticed my app was having CPU spikes. Read on to see how I diagnosed the issue on a running production server.

I have found that in the literature about games and AI (artificial intelligence) that Scrabble is invariably referred to as something of a “solved” game, in the same way that Chess, Go, etc are “solved” by AI —in these latter games, there exist AIs that have superhuman performance and can basically beat any person.