Is there any plan to implement tsumego battles on OGS?

As the title says, chess.com has a feature where you have to solve problems faster than an opponent. Fox Go also has a similar feature. It’s quite fun and I would certainly use this feature if it existed.

5 Likes

Great idea.
As far as I know this is not available right now.

1 Like

That sounds awesome! I don’t think anyone has tried coding something like that for ogs, but i’d love to see that as an option ^^

3 Likes

Not only L&D. Fuseki, joseki, tesuji, middle game, endgame (Regrouped by theme)

I played endgame battles irl once, it was fun.

We had cards with a local position on the front side, and the value of the best endgame move on the back side.

We place a number of cards on the table, front face up, equal to the number of players.

Then every player grabs one card. Then we check the values. Whoever grabbed the highest-valued card wins.

2 Likes

I had with 9.9 to finish and if you do it well, black wins with 1 point usually. In a summercamp, we had series of problems to do when eating,ike 2 A4 pages. That was cool, had to do it quickly. Then got a mark each, and some surprise (like a middle kyu finishing in the top about fuseki)

I don’t think there’s any short term plan for it, unless someone takes it on.

Probably a good first step in that direction would be big puzzle database, that you can solve one at a time and either randomly get a new one, or a rating system to keep showing you puzzles at your level.

Once the database is there it makes sense to extend that functionality into a 1v1 or bigger multiplayer puzzle race I suppose.

Then the question is how to collect or generate and possible rate a lot of puzzles.

  • Ideally come up with a good system for generating puzzles from games
  • Automatically classify those puzzles
  • Probably have users rate them (thumbs up down) but also rank them (based on a user puzzle rating say)

I imagine it might be a lot of work to manually curate puzzles, and I’m not sure if there’s any big puzzle databases that one could just get access to for this purpose (OGS has user created puzzles, but I’m not sure they’re necessarily curated enough necessarily, though one could include some in sure). So generating from games seems reasonable.

So supposing you have a katago review of a game, how could you generate good puzzles from it? :slight_smile:

2 Likes

Maybe not puzzles for everyone, but puzzles for yourself to review your game would be nice.

Lichess has a functionality like this. After playing a game, the AI shows you the situations in which you made the biggest mistakes, and asks you what you would play instead, and keeps telling you “wrong” until you find a move that the AI thinks is good enough.

3 Likes

Really good feature

AI-Sensei does something like that too.

I find the tsumego content on OGS a bit lacking. There are good and well-documented problem collections, but they are often small and feel disjointed. I would love to have an English version of 101weiqi, but sadly I don’t think the current playerbase is big enough to crowdsource such thing. Maybe someone will at some point create an easy-to-use and robust AI tool that automatically solves and collects all the interesting variations for a given problem.

2 Likes

Hey, I have a great idea right now. Think about some kind of “Race Go”, the score is not area, it’s area/time. In this way, the player who domains more area and takes less time wins the game!

1 Like

With ing rules tournaments, you can exchange points for more time.