New GoGoD database update issued

[John Fairbairn]

Yes, I know it's hard to believe but it's true.

The new version of the legendary GoGoD database, encompassing just short of 125,000 games, has several highlights, most of which I've forgotten because it's been so long since the last update. But my favourite one is that I found a new Shuei game. There are several more by Shuho and Shuwa (and others) but Shuei is a bit special.

At the other end of the scale, Sumire is moving towards 500 games, but if AI Superheroes are your thing, Sin Chin-seo is around 800 games. Cho Chikun, around 2,500 games, is still our leader.

On the technical level, the database is apparently too big for several major zip programs to unzip (though free 7-zip works a treat, I am told). But I gather we now get round this by creating several smaller download files.

Note that all the games are, as usual, built into the SmartGo package, but the downloadable version, if you need that, is available at https://gogodonline.co.uk/.

That site also has details of all my available books. You may wish to note that Go Consultants is currently being proofread by an outsider, and Volume 2 of the Encyclopaedia of Classical Go Problems has now entered my own proofreading stage, but that's quite a long process (it involves all the various indexes and page references), and of course that too needs the eye of an outside proofreader eventually.

[S0nge]

Hi John,

That, together with the re-issue of The Go Consultants, is amazing news! Thank you for the hard work!

[xela]

I've just purchased the download. I appreciate this still being available for those of us who aren't mac users and don't have smartgo available.

[Ferran]

Got mine this morning. I do have a couple of questions...

First, one on... "procedure". This is my... 4th? purchase of the database. My first was in Sept 2012, I believe. Then I got some more. Apparently, I hadn't set up a user until today. Is there any way to grandfather previous purchases? Sometimes I'd like to compare things. Not much of an issue, but I thought I'd ask.

That said, I was glad to find it has some more games on some interesting characters. Yokozeki Iho, Hayashi Sano... Nothing new for Kim Ok Kyun, I'm afraid (not that I really expected one). I'm frankly not drawn to most current games, on several levels. And, after all, you can find most with a web search. Not so with a Hattori game (always sneaky, those ones), or a Tokugawa... You get the point.

And yet, there are games by one modern player I regret to find missing (it was two, but I must have typed something wrong previously). Demis Hassabis. I'll grant you he does not rate a professional grade, but, in his way, he's changed Go as much as some of the players in the previous paragraph. Sadly, I haven't been able to find ANY of his SGFs, anywhere. I think it would be a pity to lose them if they are somewhere available. His or any other players in DeepMind. Thomas Hubert, for example, should have some games. There are two on gobase that I can't download, and his appearance at the World Youth Go Champ might have been documented.

In any case, thank you for your tasks. Take care.

[John Fairbairn]

I don't maintain an archive of past editions in the form you mean, partly because I make corrections over time.

I share your view of preferring old games but (as for go history in general) we are not just exceptions but extreme exceptions - outliers. Proudly, I hope

I haven't seen any Hassabis games but Charles Matthews (who taught Demis go at Cambridge University) may know of some. My memory of what Charles told me is that Demis was actually rather a weak player, but he was in ay event interested in too many other games to concentrate on go. I think he was a strong chess player, but in my experience it's quite rare for chess players to make even a moderately significant crossover to go, and vice versa. The same in shogi/go and xiangqi/weiqi. In the latter case, Nyu Eiko is an interesting instance in that she is the grand-daughter of a xiangqi grandmaster, Zhao Guorong. Why did she choose go?

As to other players in the AI teams, one of the Facebook Elf programmers told me that the strongest players on the programming side were only about 4-dan AGA but they were all super-strong on linear algebra (pause for huge shudder). That was some time ago. More recent developments may have attracted stronger players.

[mumps]

A game by Demis Hassabis was published in the British Go Journal a long time ago (maybe 30 years+). I forget why. He was definitely not very strong then.

[Ferran]

AlphaGo special, number 175, pg 24. Commented by Charles Matthews. As soon as I copy the ones on gobase I'll have a game by each (Hassabis, Huang and Hubert). I wish we had more, but it's a start.

Take care

[Ferran]

Links to the games I've found, here.

Take care.

[Charles Matthews]

I haven't seen any Hassabis games but Charles Matthews (who taught Demis go at Cambridge University) may know of some. My memory of what Charles told me is that Demis was actually rather a weak player, but he was in ay event interested in too many other games to concentrate on go. I think he was a strong chess player, but in my experience it's quite rare for chess players to make even a moderately significant crossover to go, and vice versa.

While I was coaching Demis in the 1990s, he improved from about 12 kyu to about 4 kyu. When younger, he'd been near to International Master level in chess.

In Teach Yourself Go/Be a Master at Go the five-stone game in Chapter 6 is one I played against Demis. I have other game records.

[Ferran]

While I was coaching Demis in the 1990s, he improved from about 12 kyu to about 4 kyu. When younger, he'd been near to International Master level in chess.

I know next to nothing about chess, but wiki says he is a Candidate Master, had a top store of 2300 in 1990, and still kept 2220 in 2019. It reads worthy to mention.

In Teach Yourself Go/Be a Master at Go the five-stone game in Chapter 6 is one I played against Demis. I have other game records.

I have Master somewhere. Since I'm moving my Go library to a different apartment sometime next November, I won't look at it right now.

That said... I'm developing a strong opinion regarding the games of the Deepmind people. I think they're as worthy of preservation as those of, say, baron Okura, Kim Ok-kyum and others. Will they ever win an even game against a Honinbo? No. But they made history. So, if for some reason you don't want to publish them, please be sure they're preserved and backed up.

Take care

[Charles Matthews]

That said... I'm developing a strong opinion regarding the games of the Deepmind people. I think they're as worthy of preservation as those of, say, baron Okura, Kim Ok-kyum and others. Will they ever win an even game against a Honinbo? No. But they made history. So, if for some reason you don't want to publish them, please be sure they're preserved and backed up.

They're in an exercise book (squared paper).

[Javaness2]

There is a nice interview here, https://www.sciencefocus.com/future-tec ... part-of-me
You might well imagine that Demis could have probably have gotten to IM level, but he doesn't seem the single minded type. I seem to remember that JH told me that they went and played pool instead of staying on at the Go club after the opening night lecture, something along those lines.