Life In 19x19

Hi,

I have just put my new Go tutorial online at https://www.crazy-sensei.com/rules/. It runs a small neural network in the browser to teach beginners to play on 5x5 and 9x9. The score is the number of stones on the board, to make it simpler to understand.

I still have very little experience of observing newcomers using the tutorial. If you try it, I would be happy to hear your opinion.

In the future, I'd like this tutorial to become completely interactive. But it will take more work.

Any feedback is appreciated.

Rémi

I like the idea of using computer go for teaching, and I think that the page you put together looks nice. One thing I might consider would be to do something about the computer's continued play after it's obviously lost the game.

One of the most confusing things for beginners, in my experience, is understanding when the game is finished. When Crazy Sensei continues to play moves when it's already lost the game, this could contribute to that confusion. Maybe this is why you decided to score based on the number of stones left on the board, but it might still be confusing.

That being said, it's a very nice page. Thanks for making it.

Kirby wrote:I like the idea of using computer go for teaching, and I think that the page you put together looks nice. One thing I might consider would be to do something about the computer's continued play after it's obviously lost the game.

One of the most confusing things for beginners, in my experience, is understanding when the game is finished. When Crazy Sensei continues to play moves when it's already lost the game, this could contribute to that confusion.

That being said, it's a very nice page. Thanks for making it.

I decided to use the number of stones on the board as the score. I believe it helps beginners in the very first games. Explaining life and death, as well as territory, seems extremely complicated to do in a clear and rigorous way. So both players have to fill the board completely.

In the future, I might extend the tutorial to use Chinese counting on 9x9, and make it pass early.

Rémi wrote:I believe it helps beginners in the very first games.

Okay. Maybe I found it confusing since I hadn't learned that way, but I admit, I've never tried teaching a beginner using this method.

Good luck.

I am in fact following the recommendation of Dieter Verhofstadt, who seem to have a lot of experience teaching beginners this way:

http://senseis.xmp.net/?DieterVerhofsta ... troduction

I will make my own experience whenever I can get the opportunity to actually watch a beginner use the site.

Hi Rémi,

Thank you very much, and Congrats.
The stones, the shadows, and the UL highlights are pretty.

Testing:

• Firefox, Windows 8: OK.
• Safari, iOS 7, iPhone 4S: "Rotating cursors" after 300+ secs (hang?).
• non-Safari, same as above: ~22+ secs to load; runs OK.

The vertical wood grains look nice;
the wood may be just a bit too much green component.

EdLee wrote:Hi Rémi,

Thank you very much, and Congrats.
The stones, the shadows, and the UL highlights are pretty.

Testing:

• Firefox, Windows 8: OK.
• Safari, iOS 7, iPhone 4S: "Rotating cursors" after 300+ secs (hang?).
• non-Safari, same as above: ~22+ secs to load; runs OK.

The vertical wood grains look nice;
the wood may be just a bit too much green component.

Thanks for your feedback. The iPhone 4S might not be powerful enough. I have an iPhone 4, and it does not work at all, either in Safari or Chrome. Which non-Safari browser did you try?

I have reports of bugs on recent iPhones, too. But don't have one for testing. I tried in the emulator on my Mac, but could not replicate the problem. If any iPhone owner here has bugs and wants to help, please contact me.

Work perfectly on my nexus 5 that it almost look like remi test specifically for this phone.

Hi Rémi, something called Mercury, from the App Store.

Thanks for working on this! More high quality beginner materials are really valuable!

Edit: Everything worked fine for me in Chrome.

Worked OK for me on FIrefox and Windows 10. A suggestion to improve the discussion. Go is explained at first in terms of alternating play. Some of the problems about capturing ask the player to capture (with minimum number of moves) a group of stones. To me it wasn't clear at first that the player in these problems is to make consecutive moves to fill in all the liberties.

This is a great tool to help beginners learn how to play. I wonder whether you might want to say something about the scoring part being slightly different from what would usually be used online or at a go club. I agree that knowing when the game is over and how to decide which player wins are big difficulties for many beginners and I like that this stone counting scoring (old Chinese) method solves two big problems for beginners. For me a possible difficulty with this scoring method would be when the beginner learning this way goes to a go club. Here in the USA probably most go club members use casual Japanese territory scoring. Especially weaker players, would not know how this stone scoring method works and I can imagine an experienced (say 10k) player playing our beginner correcting her when she starts playing inside her own territory towards the end of the game. I wouldn't want to start people out with the most used rules but maybe have a version of your player software that could score by territory as an option for when the learner is ready to play online or at a club.

At first, I couldn't figure out what you guys were talking about...scoring.

Its not obvious how to get to the scoring -- yes, there is a "pass" button, but when does one press it, and nowhere mentions that its tied to ending the game or scoring. As far as I can tell, CrazySensei will never pass first?

I remember when I was first learning the game, I never knew how to end a game, and always hoped my opponent would pass first. Technically, I believe a "pass" as an ending move (or a move in general) is a modern/western rule invention, and as such, has to be explained.

The computer passes first when there are no other possible moves to play, as far as I can tell. Like I said earlier, I found this confusing, but maybe it's since I am so used to the normal way.

It sounds like Remi's method has been used to teach beginners successfully in the past, so it'll be a good test case for that method.

In general, ending the game is a tricky topic, so there must be a tradeoff: (1) postpone the complexity until the learner has more experience; or (2) try to introduce this complexity from the beginning.

Remi's app aims for the formal, which seems good. Though, in my case, the new way of scoring seemed to add to complexity rather than reducing it

But, for someone just learning the rules, maybe they'd get it. I think it's good to just try out Remi's method. If it works, great. If not, modify the approach.

xed_over wrote:Technically, I believe a "pass" as an ending move (or a move in general) is a modern/western rule invention, and as such, has to be explained.

Traditional/eastern rules do not need to be explained?

I've forwarded this site to a bunch of newbies. Hoping for useful feedback!