After almost a week, it looks like people have found the bot. It now plays roughly 300–400 games a day, with a peak of 513 games on Saturday.
Whether players like the bot or find it more human is hard to say, I haven’t received feedback, but several regulars return daily to try beating it, which is a good sign.
The bot currently sits around 30–37k, though that rank isn’t very meaningful. Most people play it unranked first and only try a ranked game once they feel confident. A more telling metric is the winrate: about 70%. Strong enough to be interesting while leaving room for a weaker companion bot…
A second bot, Tiger, is now live. 
Tiger runs on the same engine as Oroton but is tuned for contact-play and aggression. Two rules are removed:
- OpenArea (play in open spaces)
- InfluenceDelta (play where influence is largest)
With these gone, Tiger has no sense of “big” moves. It only sees “urgent.”
What remains:
- liberty rules (increase yours, reduce theirs)
- connection and cutting
- eye-making
- anti-blunder rules (ladders, sealing territory)
The result is a fighting-heavy style. It loves capturing groups, so while it is overwhelming at first, beginners who learn to tenuki will be able to beat it, especially on larger boards. In my testing it’s stronger than Oroton on 5×5, roughly even on 9×9, and weaker as the board grows.
Despite its flaws (same unnecessary eye-making as Oroton), it’s fun to play. Give it a handicap and try out-fighting it.
Early statistics, after one day, show a winrate around 60%.
Great suggestion, I added size*2 boardmoves as a requirement for passing to both bots.
Passing when the player passes is a design choice, so that players learn how/when to pass, without a bot always taking care of it. But letting incomplete games go to scoring is an unintended consequence. I have fixing that on my to-do list, but this is already a nice patch.