- A rewrite of the core program is under way (version 0.6).
- As of June 2006, the bot is nearly able to resume online play.
- Started in March 2007?
- A completely new bot. Designed from the ground up using an Agile programming approach, and written using d
- Multi-threaded. Supports pondering (thinking on opponents' time), multiple control interfaces, searches that use multiple cores, and multiple simultaneous search types
- Includes monte carlo searching option
- Based on a rewrite of the move value and move selection code to use object based values instead of a per-brain numerical value.
- Large emphasis on using a score estimator fed with stone stability information. Both pieces of that (score estimator and stone stability) will continue to be enhanced throughout this development cycle
- Still under development as of Nov 2006, but starting to make debuts on KGS
- Added benson style unconditional life detection and support for accurately marking alive/dead stones at the end of the game.
- Added handicap support (up to 9 stones)
- Brain: Decrease Liberties - Value moves the reduce enemy chains to some number of liberties. (0.4.0 only valued atari plays higher...)
- Brain: Protect Connection - Similar to protect_base but blocks enemy from passing between closely positioned friendly stones.
- Switched to many simple brains (and stopped using Bouzy)
- Brain: Gold Silver Grass - play in corners preferentially
- Brain: Play Light - Don't play near other friendly stones
- Brain: Increase Liberties - Extend/Connect to reach 4 liberties
- Brain: Protect Base (Skirt) - Descend to block invasions
- Beat its first human (so what if it was their first game after learning the rules several months prior!)
- Added chain & liberty tracking
- Understands stone captures and avoid suicide/ko
- Brain based computation
- One brain for legality (don't play on top of other stones)
- Another brain for score estimates - Bouzy 5/21 Score Estimator
- GTP operation
- played along diagonal then resigned.