Realgoproblems.com

Hi all,

Tired of tsumego, finding them samey and tedious, I’ve made a website with whole board Go problems generated from pro games by AI. The user is challenged to find urgent moves, make good shape, tenuki when possible, etc. I thought it might be a fun way to study the game.

https://realgoproblems.com/

Let me know what you think!

13 Likes

I’d like to take a look without registering.

6 Likes

That’s being worked on. But you can use a bogus email if you want, there is no verification

I tried it out, and I like the idea! The problems seemed approachable even though they are from pro games. Also, I didn’t know Waltheri’s go engine is open source - cool!

+1 to Uberdude’s comment that registration is a barrier to entry. Would be great if I could have tried a few problems without making an account.

I’ll add that even after registration, I still needed to login. I recommend logging the user in as soon as they register (this is what OGS does)

3 Likes

My guess is that registration is a minimalist filter for spams?

On opera/android the number pop up at the top of the board is not working (open and instantly close the keyboard)

The database is made of very recent games (2024) and that’s nice.

I didn’t find a way to reset the problem (keeping instead the display of AI or pro move all time after analysis). Refreshing the page produces a new problem.

1 Like

This is now possible, click “try it out” on the login page. It will start you at 1700 elo and will remember your rating by cookie, but your actions won’t affect the problems’ ratings.

3 Likes

This last thing shouldn’t happen, but I suppose it’s one of those things I only tested once :wink: Will try to reproduce, thanks!

@Groin

My guess is that registration is a minimalist filter for spams?

It’s partly that, and also that registered users have a more accurate rating, which means the problems will settle at their correct ratings more quickly.

On opera/android the number pop up at the top of the board is not working (open and instantly close the keyboard)

Can you clarify what you mean here?

I didn’t find a way to reset the problem (keeping instead the display of AI or pro move all time after analysis). Refreshing the page produces a new problem.

There is an as of yet undocumented way to view a specific problem, go to …/problems/{ID}, it will tell you the id after you attempt one.
I will soon add the ability to play follow up moves post-solution, and the solution will then disappear.

3 Likes

So, I registered but cannot log in.

Maybe I’m just silly and typed in the wrong password, but it would be great if I could reset/change it.

Tried it and liked it, so probably will register soon to get more puzzles and a better and more reliable ranking.

Why did you choose for a one-move solution?
Why not the possibility to play it out with some more moves?

Why did you choose for a one-move solution?

At first I wanted to have problems with multiple solutions as well, but it turned out to be hard to generate the ones I was imagining. I was hoping I’d get problems in the opening or early middle game where there are a few big moves to choose from, but these turned out very rare and I was spending too much time sifting through many bad quality problems. Usually when the AI thinks multiple moves work, it thinks many moves work which makes for bad problems.

Why not the possibility to play it out with some more moves?

This will be added soon.

@Leira Password reset is low on the priority list but will eventually be added. For now you can just make another account with a bogus email and I can switch the username and email around if you want.

2 Likes

Nice!

I’d suggest that problem attempts by unregistered users should not be counted at all for calculating the problem ratings.

They’re not, iiuc

Yeah I meant they’re not counted at all and therefore it takes longer for the ratings to settle. I phrased that weirdly.

1 Like

It’s now possible to replay the rest of the game after attempting the problem

1 Like

The login dialog needs a keyboard interface.

I tied it out. I made a move, but nothing happened. It just showed a bunch of unexplained statistics. No clear indication of how good or bad my move was.

I guess my suggestion is to provide an introduction page after someone presses the Try It button.

++++

I play 9x9 games, and find that there are many good and bad moves determined by local configuration, but no way to learn them. As an example, playing on the 3-3 point at the start of a game is guaranteed to live, and there are not very many kinds of follow-up moves that occur in real games. Yet there is no set of puzzles that explore what to do in each such configuration starting from the initial 3-3 move. In practice, my 3-3 stones die almost as often as they live, and I have no way to learn why without spending hundreds of dollars for a skilled teacher, and even then they may not be able to show me everything I need due to their unfamiliarity with 9x9 play.

You can check the joseki database here, with sequences starting by the 3-3. It’s limited choices to approach a 3-3 because other approach moves will not be as good as the one repertoried, like leading to some bad shapes or depriving from options in the future.

Most of these sequences are calm leading to a clear cut between territory vs central influence or territory vs side devellopment.

Great idea!

There’s an about page, accessible from the login page or from the hamburger menu.

I play 9x9 games, and find that there are many good and bad moves determined by local configuration, but no way to learn them. As an example, playing on the 3-3 point at the start of a game is guaranteed to live, and there are not very many kinds of follow-up moves that occur in real games. Yet there is no set of puzzles that explore what to do in each such configuration starting from the initial 3-3 move. In practice, my 3-3 stones die almost as often as they live, and I have no way to learn why without spending hundreds of dollars for a skilled teacher, and even then they may not be able to show me everything I need due to their unfamiliarity with 9x9 play.

It sounds like you’re more in the market for regular tsumego (life & death puzzles). This project is trying to challenge players to find the correct moves in all kinds of situations that might occur in 19x19 games.

1 Like