One quite sensible restriction would be that each individual promise should only depend on the sequence of moves in the game (given a promise and a sequence of moves, it should be easy for a computer program to verify whether the promise has been broken or not).
While pretty restrictive, I still think there would be lots of possibilities with this. Some weird ideas just as examples:
I, Red, promise that the Blue player will never play A1.
I promise that at least one Yellow stone will be captured before move 100.
I promise that if I ever have the option of capturing a Yellow chain, I will do it.
Yes I believe it is possible to fix the “problem” in a rigorous way. I would formulate a promise as a mapping from the set of all possible game states, where a state is mapped to the set of moves that are promised not to be played.
The problem arises if we consider promises to be a part of what defines the possible game states. Then the definition of game states and promises reference each other and can’t be constructed.
So, if Red does not or chooses not to prevent Blue from playing A1, and Blue goes ahead and plays there, would Red be eliminated for “breaking” a promise?
I guess it could be a weird way to try to entice Blue to play at A1, or compel someone else to help block, but offering a suicidal incentive?
Yes, my point was that it is perfectly fine to make a “promise” even if you are not fully in control of the outcome! It’s not so much that Red “broke” the promise, but that Red failed to keep the promise (or to use more neutral language, the promise turned out to be false).
Right, it sounds weird in isolation, but I’m sure such opportunities to affect other players incentives could often be used effectively, especially towards the very end of the game
There are three topics on Lifein19x19 where they played “Mafia go”. I was very intrigued and tried to imagine how that would work, but I haven’t spent enough time reading the threads to understand the rules.
I think the werewolf-go setup with immortal wolf stones has potential, here are some ideas for special characters that should pose no severe balancing issues:
witch: Once per game remove a wolf stone, once per game place a wolf stone
mayor: may play up to two moves per round (if the role is revealed)
medic: each night protect one intersection from the wolves (not twice in a row the same)
cupid: chooses two lovers
I believe the information gathering roles will be difficult to balance, because I imagine that the pace of the game will be much different. The wolves can’t kill one player per night, instead they slowly creep upon their victims one immortal stone at a time.
I promise that for each turn n>2 if the difference between the 10^n th prime number and the 10^(n+1) st prime number is divisible by 6 then I will play at point n modulo 81 ( reading points from top left to bottom right across rows and then down A9, B9,…J9, A8, B8, … etc) assuming a 9x9 board.
Another idea. Combine Go with “The Resistance”, a commercial game pretty similar to Werewolf, but without player elimination.
In “The Resistance”, there are two teams: the majority of “innocents” who don’t know who’s who, and the minority of “spies” who know their fellow spies. Players take turn being “team captain”. The team captain selects a number of players to be part of this round’s team. Then everyone votes to either accept or reject the selected team.
If the team is rejected, the next player becomes team captain and gets to select another team;
If the team is accepted, then the team members take part in a secret vote; every team member secretly votes “Play” or “Sabotage”; if there is at least one sabotage, then the Spies score one point; if there are no sabotages, then the innocents score one point. The game ends once one side has scored a total of three points.
Team members vote secretly, so if we see for instance “1 sabotage and 2 plays” then we know that one of the three team members must be a spy, but we don’t know who.
We could use a similar system, but instead of simply voting to immediately score one point, we’d vote on the next move to play. Or on whether White or Black plays the next move. Or perhaps the team captain suggests a move; the innocent team members vote “Yes”; the spy team members can choose to either vote “Yes” or choose another move.
Thanks to anoek whitelisting multicapturego.antontobi.com for embeds, we have some new options for how to set up games like these!
Simply by making an iframe with the URL to the current game, we get an embed like this:
This by itself doesn’t do much - it’s still a static webpage so if you make moves they are not saved anywhere. Some different ways to play would be:
After making your move on the embedded board from the last post, it generates the code for a new embed, that you can paste into your next post to submit your move. This would get rid of the separate steps of having to copy both an image and an URL. I’m not sure if having hundreds of embeds in the same thread would be a performance problem (presumably Discourse only loads the ones on screen).
The top post of the thread could be a wiki post that anyone can edit. Making a move would then consist of clicking “edit”, adding the coordinates for your move to the end of the link (I would update the tool so that it uses human readable coordinates), and clicking save. One drawback is that the game discussion would be a bit separated from the game (but jumping from top to bottom of a thread is not too hard).
I was going to say add an undo button because it doesn’t look like I can go backward? Does the url store the move order or just the stones on the board at the start of the turn?
Maybe there isn’t a way to do undo without refreshing
Yes, the current tool is not made for this situation so it doesn’t make a lot of sense For the second solution I suggested above, it might be best to not allow inputting new moves by clicking at all (to reduce confusion - else someone might think they can just click their move and have it saved), but instead just have backward and forwards buttons to look through the game history.
I’m just asking for any other ideas for how to use a static embedded webpage to conveniently play a game, because there might be something I haven’t thought about. If we find a good solution, I will update the tool accordingly.
I guess it’s hard to come up with more ideas when we’re limited to the threads abilities.
One could also think of making a single locked thread for the game itself, there’s only one post that could keep being edited with the new board state, and the discussion can be done in a separate thread?
I’m trying to think of another way to avoid clutter for the game and discussion - just for options.
An entirely unrelated idea I had was that we could play a live Battle Royale with a “game master” streaming the board, with all the commentary and move submissions happening in Twitch chat (me and @martin3141 once played a game of Simultaneous Go like this, with @Vsotvep streaming and taking our move submissions as private messages).
If there are enough interested people and we can find a time that works, I would very much enjoy hosting such a game!
Here’s another game variant that I think sounds really good:
The host generates 2 random moves per player, similarly to how we did in the current game. But these stones are hidden, only known to the player who owns them (and the host of the game).
The host keeps track of all the stones on a secret board, and announces when a move doesn’t go through because of a collision (in this case the colliding player gets to try again until they find a legal move) or when a capture happens involving some hidden stone.
As soon as a hidden stone affects the game in either of the above ways, it is revealed to all players.
You probably wouldn’t want your hidden stones to become discovered too quickly, as you might want to do some bluffing (and they might also be safer while hidden). But this makes communication and collaboration tricky!
I’m currently reading Iain Bank’s The Player of Games, and it features a game not unlike go, but with up to 10 players on the same board, and a few hidden moves exactly like that!
For the second solution I suggested above, it might be best to not allow inputting new moves by clicking at all (to reduce confusion - else someone might think they can just click their move and have it saved), but instead just have backward and forwards buttons to look through the game history.