Honestly, as a long time Voronoi enthusiast, the idea of turning a regular grid into a Voronoi diagram is kind of obvious and I admit that this is definitely something that I would like to have come up with myself, but the idea never occurred to me.
The implementation is obviously pretty complex though. And I remember trying to get into regular Go several times along the years and I've always found it very difficult to grasp it (the computer thoroughly beat me every single time!) and I lost the motivation.
I read your tutorial pages, but I learn best by doing so I guess only playing for some significant amount of time will actually help me have a more than skin deep understanding of the game.
I do wish there was a Windows desktop version as the web version is eating 90% of my laptop's CPU when playing against the bot.
Thanks! I think that the high CPU utilization is probably the bot itself. It's a full-blown neural net running in-browser, so it's definitely a resource hog. If you ever play against a human performance should be much better (though please let me know if that's not the case).
Very well done. I had to enable desktop mode on a phone because the help dialog hides the board and the Next button is some meter below the bottom edge of the screen. That's the only issue and the site warns that it's best played on a desktop. Everything else is very well done and it's really a game of go, with eyes, invasions, connections, cuts.
Thanks for playing! Yeah, the mobile experience is pretty wonky still. I mostly use it to respond to correspondence games when I'm afk but otherwise I think most people prefer to play on desktop.
Gonna be honest, this got a click from me as a golang developer having just come from the terminal multiplexer thread and I fully expected this to be about developing voronoi cells in golang.
I don't know Go well enough to even begin to conceptualize how deeply this changes the rules and interactions of the game, so I'm not going to be able to comment on the concept itself, but I bounced off the tutorial the first time and by the end of it had no desire to play a full game against a bot.
Maybe if the tutorial had walked me through the beginning and middle of the match with the bot and engineered the match so that the tutorial scenarios appeared within that game, it would have felt like less of a committment to actually start a full game.
Thanks for the feedback - the tutorial is definitely dense. It existed a long time before the bot did, so it had to cover everything in one sitting. I think now though that you're probably right about an improved tutorial flow being mostly playing + some explanation.
My takeaway from the past couple months is that the game is just kind of hard to learn. I actually felt the same way about trying to learn traditional Go - it was very daunting to start a game. Basically all of the players who have stuck around / joined the discord were already pretty familiar with Go, so I think it helped bridge that gap a bit. I've spent a good amount of time trying to figure out how to make the game more approachable, but it kind of seems like the core audience for now might just be people who already like Go. There also seems to be a lot of overlap with people who are interested in designing a bot to play, so I've open sourced some code to allow for that: https://github.com/csun/voronoi-go-rs
Looks really cool. Saved for later because I'm on my phone right now. Ignored the warning and it was fine but the site refreshed sometimes which restarted the tutorial. The third time I gave up.
5 kyu in normal go here: I played a couple of games and got crushed by the AI. Groups are much harder to kill in this game, so having a large group of connected stones with zero or one eyes (what we call “dragons” in normal go) is a much better strategy (I think) then to create potentials in the corners.
This is counter to the best strategy in normal go. We actually have a saying that “big dragons never die”, however it is actually an anti-proverb, because most dragons can be harassed while your opponent builds territory. So even if your dragons lives, you wasted a lot of stones keeping your dragon alive, while your opponent was calmly building territory, and gaining points.
I'm similar strength (4k ogs) and I had the same experience. I think the essential difference is the cross-cut in this game can produce 3 groups (a connection on the shorter axis remains), whereas it produces 4 in Go. That makes it possible to rampage across the board, cutting a weaker player to ribbons while staying in one big group. Fun game.
Thanks for playing! I don't have a ranking in traditional Go, but after tens of hours playing this game I'd generally agree with your assessment.
One issue is that I think I made the small board too small. This makes it very easy for a monolithic group to dominate. The bot prefers to open right in the center, and I've generally found that to be strong.
I _think_ that playing on the big board gives some more variety to the strategy, but I haven't finished enough games yet to be sure. Playing really connected though seems to still be hard to beat. If players continue to get stronger and play big board games it will be interesting to see how the meta evolves.
Love it, such a novel idea!
Honestly, as a long time Voronoi enthusiast, the idea of turning a regular grid into a Voronoi diagram is kind of obvious and I admit that this is definitely something that I would like to have come up with myself, but the idea never occurred to me.
The implementation is obviously pretty complex though. And I remember trying to get into regular Go several times along the years and I've always found it very difficult to grasp it (the computer thoroughly beat me every single time!) and I lost the motivation.
I read your tutorial pages, but I learn best by doing so I guess only playing for some significant amount of time will actually help me have a more than skin deep understanding of the game.
I do wish there was a Windows desktop version as the web version is eating 90% of my laptop's CPU when playing against the bot.
Thanks! I think that the high CPU utilization is probably the bot itself. It's a full-blown neural net running in-browser, so it's definitely a resource hog. If you ever play against a human performance should be much better (though please let me know if that's not the case).
Very well done. I had to enable desktop mode on a phone because the help dialog hides the board and the Next button is some meter below the bottom edge of the screen. That's the only issue and the site warns that it's best played on a desktop. Everything else is very well done and it's really a game of go, with eyes, invasions, connections, cuts.
Thanks for playing! Yeah, the mobile experience is pretty wonky still. I mostly use it to respond to correspondence games when I'm afk but otherwise I think most people prefer to play on desktop.
Gonna be honest, this got a click from me as a golang developer having just come from the terminal multiplexer thread and I fully expected this to be about developing voronoi cells in golang.
I don't know Go well enough to even begin to conceptualize how deeply this changes the rules and interactions of the game, so I'm not going to be able to comment on the concept itself, but I bounced off the tutorial the first time and by the end of it had no desire to play a full game against a bot.
Maybe if the tutorial had walked me through the beginning and middle of the match with the bot and engineered the match so that the tutorial scenarios appeared within that game, it would have felt like less of a committment to actually start a full game.
Thanks for the feedback - the tutorial is definitely dense. It existed a long time before the bot did, so it had to cover everything in one sitting. I think now though that you're probably right about an improved tutorial flow being mostly playing + some explanation.
My takeaway from the past couple months is that the game is just kind of hard to learn. I actually felt the same way about trying to learn traditional Go - it was very daunting to start a game. Basically all of the players who have stuck around / joined the discord were already pretty familiar with Go, so I think it helped bridge that gap a bit. I've spent a good amount of time trying to figure out how to make the game more approachable, but it kind of seems like the core audience for now might just be people who already like Go. There also seems to be a lot of overlap with people who are interested in designing a bot to play, so I've open sourced some code to allow for that: https://github.com/csun/voronoi-go-rs
Anyway, thanks for taking a look!
To be fair, a game of this only takes a few minutes. Probably best as a "learn by doing" type of thing.
Looks really cool. Saved for later because I'm on my phone right now. Ignored the warning and it was fine but the site refreshed sometimes which restarted the tutorial. The third time I gave up.
5 kyu in normal go here: I played a couple of games and got crushed by the AI. Groups are much harder to kill in this game, so having a large group of connected stones with zero or one eyes (what we call “dragons” in normal go) is a much better strategy (I think) then to create potentials in the corners.
This is counter to the best strategy in normal go. We actually have a saying that “big dragons never die”, however it is actually an anti-proverb, because most dragons can be harassed while your opponent builds territory. So even if your dragons lives, you wasted a lot of stones keeping your dragon alive, while your opponent was calmly building territory, and gaining points.
I'm similar strength (4k ogs) and I had the same experience. I think the essential difference is the cross-cut in this game can produce 3 groups (a connection on the shorter axis remains), whereas it produces 4 in Go. That makes it possible to rampage across the board, cutting a weaker player to ribbons while staying in one big group. Fun game.
Thanks for playing! I don't have a ranking in traditional Go, but after tens of hours playing this game I'd generally agree with your assessment.
One issue is that I think I made the small board too small. This makes it very easy for a monolithic group to dominate. The bot prefers to open right in the center, and I've generally found that to be strong.
I _think_ that playing on the big board gives some more variety to the strategy, but I haven't finished enough games yet to be sure. Playing really connected though seems to still be hard to beat. If players continue to get stronger and play big board games it will be interesting to see how the meta evolves.