Go rating in BGG

Written by alejo on April 29th, 2012
2 Comments

While updating my collection and my ranks at Boardgamegeek (BGG), I realised that I still place Go at the top of my rankings, giving it 10 out of 10 points. Though you may consider, as I do, that the perfect game doesn’t exist, you may agree with me in the awesomeness of this game. Go starts with very simplistic rules and its strategy it develops surpasses that of any other game *, it’s visually appealing and challenging at the same time. It’s been there for millennia, withstanding centuries with only scoring modifications. Well, most of the games at the top of BGG stand on our shelves in a couple of generations. Certainly, if there is a game to be placed a 10 out of 10 points, it’s Go.

Fanatism? You may call it. Wandering around the BGG I found out there is a rating graph for every single game. Let’s check it:

Realising I wasn’t the only one giving it a 10 was a shock at the very beginning. After recovering, I realised the 10 points was the most frequent score (call it statistically mode) and that Go rating would follow a sort of normal distribution (Wikipedia)  if it wasn’t for the upper tower, with the 9-points tower being significantly lower than the 8 and 10. High as it is, the 10 points tower scores the 25.60% of its final average result, which is 7.551.

One would think this is normal for a game placed among the top games of the BGG. However, you should know that Go is placed in the 43rd position**. Right after Pandemic:

With a final average of 7.556, Pandemic shows a more traditional normal distribution, with only 5.58% of the votes on the 10 points tower. Though differences were to be suspected, I didn’t expect them to be this big.

How does Go perform when compared to the top games? Here you have all three graphics from the top 2 games:

 

Twilight Struggle
Twilight Struggle

Agricola

 

As you can see above, both Twilight struggle and Agricola get a decent amount of the maximum score, but they both show a sort of normal distribution with 8 as the statistical mode. Actually, Twilight Struggle, the best rated game from BGG gets a final score of  8.219, with 24.53% of 10 points. Agricola’s average score is 8.168 with 21.9% of the maximum rating. Compared to Go’s 25.60%, the difference is little but impressive if we realise that Go is placed 43rd.

Failing to find more games with a higher 10-points proportion within the first 10 games led me to wonder about other traditional games. This time with better results. Although Chess fails to reach the 20% of 10-points, it shows a similar graph:

Other games with well-known fanatism such as Poker or Magic: the Gathering show similar numbers: 9% and 14,2%, respectively. However, these two games score approximately the same in both 9 and 10 points, without a significant increase of the 10 points tower.

In the end, we should conclude that there is a remarkable amount of Go lovers within the BGG community. Not just people who have played go once in a while, not just regular players, but people who like it as much as me. So if you smiled in agreement during the first paragraph, don’t feel ashamed to give it 10 out of 10 points, 1600 users from BGG agree with you.

*: Most of the reasoning in favour of Go can also be made with Chess, though this last has more complex rules to start with.

**: The Geek rating includes a Bayesian regression, so that  games with few users can’t climb up too easily. If we only considered the actual user ratings, Go would be placed 52nd.

Tags: ,
Permalink

Zen computer Go program beats Takemiya Masaki (9p) with just 4 stones

Written by alejo on March 19th, 2012
Comments Off on Zen computer Go program beats Takemiya Masaki (9p) with just 4 stones

One would believe that computer strength was years below pro level. This week we’ve been proved otherwise. Zen19 managed to beat a top pro, Takemiya Masaki (9 dan pro), with 4 and 5 handicap stones. Since pro players can’t really give that much handicap to other pro players, some may argue there is still a big gap for artificial intelligence. We must not forget that the CPU power wasn’t that of a supercomputer and its strength has risen a lot lately.

The text below is extracted from a blog post from the Go Game Guru blog. Sincerely, the author made a great job with the article ant the actual piece of news are awesome. It’s a must read.

As part of the ’6th E&C Symposium’ in Japan, Japanese pro, Takemiya Masaki(9p) played two games against the computer Go program Zen (aka Zen19). Much anticipated by both Go players and AI experts, it was an opportunity for Zen to flex its muscles against a world class professional, though many still expected Takemiya to win. Both games were played on March 17, 2012, on a 19×19 board. Regular readers might recall that Zen played John Tromp in the ‘Man vs Machine match‘ earlier this year

Just in case you are interested into AI and computer go, I’d suggest you visit this interview.

While you are visiting that blog, you may like to take a glance at some other posts, such as “5 tips for dealing with unexpected moves “ and the “Learn Go the Easy Way series“.

 

Tags: , , ,
Permalink