Wanderer_c is a bot, playing a decent game of Breaktthrough.
It is developed by Richard Lorentz and some of his students.

Now it has reached good strength (almost 2000 rating points
here on LG), and has entered the top-10 of the rating list.

My questions:
\ Is there still a big gap in comparison with the best
humans?
\ Does wanderer_c have a playing sstyle different from that
of human players?
(Would it pass a Turing test?)
\* Has the community learned something about the subtleties
of BT from wanderer_c?

Ingo.

Hi, Ingo. I'm still by no means an expert in this game but I can give my opinions / partial answers to your questions. I hope some BT players will followup with more authoritative responses!

1. Big gap? My impression is that the answer is “no”. Of course this answer is based on scant data, especially since the current version has been running on LG for only about 3 weeks. But looking at the moves Wanderer is making (and its recent results) it seems to me to be playing at least at a level to stay in the top 10 and maybe even the top 5 or better. I guess only time will tell.

2. Style different from humans? This is harder for me to judge, but I think the answer is “no”. Only because of the incredible openness and helpfulness of the players here has Wanderer been able to make the strides he has. All players, including many of the strongest, have been willing to answer my questions about the games and offer suggestions for where Wanderer is strong and where he is weak. Because of this: (a) Wanderer has improved very quickly and (b) Wanderer's “style” is based on these suggestions which almost surely means it plays like a human.

I find your implied question intriguing: Can a human easily distinguish games that are played by Wanderer from those with two humans. Again, I think the answer is “no”.

Related to this style question, I have tried some experiments to have Wanderer play in a different style. I am curious if humans have really found the best strategy for BT plus it would be very cool if we could find some new and improved approaches to the game. So far I have been unsuccessful in finding anything that approaches the strength of the current version.

3. Has Wanderer taught the community anything? This is the hardest question for me to answer, but since Wanderer seems to pretty much just use human ideas for finding its moves I fear the answer to this question is also “no”.

-Richard (for Wanderer)

Hi Richard,
thanks for the answer and your judgements.

Concerning the LG ranking, it might indeed be possible fro Wanderer_c
to reach rank 5 soon. Currently it is at rank 7 with 2004 points,
and rank 5 has 2035 points. But there is a big gap to rank 4 (with
2083) and rank 3 (with 2089).
So, rank 5 might also be the maximum within reach.

It might be interesting, how Wanderer_c (and humans) would perform at
another board size.
8x8 has some special symmetries and some simple strategic elements
(keep bc-fg on the bottom rank) which would not hold for instance on
7x7-board.

Ingo.