I like playing Go, but of the games I play at all regularly it's probably the one where there are the most people who are better than me: a lot of people play go, and many of them play it a lot.

My dad and I are a similar level (~11kyu) and last week a family friend who's substantially better (~6kyu) was looking for a game. Instead of either of us playing him 1:1 with ~5 handicap stones, we decided to play 2:1 even. As a casual game, we played untimed.

It was a really interesting game: my dad any I both compensated for each others weaknesses, were more thoughtful and careful, didn't make any major mistakes, and ended up winning by a lot. It felt a lot like pair programming.

I'm very curious now how much better playing as a pair tends to make people. While I expected us to do better, +5 ranks (+500 ELO) was a much larger change than I would have guessed. It also may not be linear: I would expect better players to generally benefit less, since worse players are more likely to have uneven-but-complementary skills?

Comment via: facebook

New to LessWrong?

New Comment
3 comments, sorted by Click to highlight new comments since: Today at 4:58 PM

Did you talk out loud when doing this? If so, was your family friend able to use your communication?

I wonder what the ELO difference would be for doing this with chess.

Did you talk out loud when doing this?

Yup. We didn't try to hide things. We figured he was enough better than us that we weren't going to be able to lay traps, and just focused on playing a solid game.

I wonder what the ELO difference would be for doing this with chess.

I'm curious too! People should collect more data!

In this case, I think it was probably larger than +500, since we won by a lot. Though it's just one sample.

There is a lot to be said about this for avoiding errors, common biases, or 'blunders.' This is probably why 'pair programming' works the way it does.