I've been interested in this idea of games with modifiable rules on and off, but something about the voting aspect doesn't appeal to me that much. Not entirely sure why. I like the self-referentiality of the rules being modifiable, but it just being a vote, "what rule change do you want?", feels unsatisfying.
I'm not sure what to replace it by, admittedly. I did one experiment where instead of saying what the rules were, your actions were taken as an example of what you thought the new rules should be: if you did something not previously legal, it was inferred to now be legal [1]. But that just led fairly quickly to chaos, unless it were carefully restricted. So maybe not the right approach either. But I still like the idea that the game-mechanics changes themselves are related to the game mechanics in a cybernetic-feedback-loop kind of way, rather than a completely "outside the system" free vote. Now as to how to implement that in a way that makes for a playable/interesting game...
My friends and I used to occasionally play a card game called Mao. The winner of each round (the first to get rid of the cards in their hand) could make up a new rule governing game play for following rounds. But they didn't announce their rule, instead punishing players who broke it by giving them an extra card. By watching the punishments, you had to try to work out their rule so you could avoid breaking it.
With the right group, it was quite fun, if rather tiring. You need to not take winning too seriously, because it's quite easy to make up a complex rule no-one will ever get - one of my friends had 'you may not play a prime number card on another prime number card'.
Mao belongs to a class of games called inductive reasoning games. They're mostly crap, but there are a few gems. IMO the best one of the bunch is Zendo where one player makes up a rule that determines whether a grouping of playing pieces is valid or not (usually played with Treehouse pieces, but e.g. a limited selection of lego pieces can also work). The other players then basically need to use an experimentation / induction loop to determine the rule. This can be really fun in a hackerly group. And once the players get experience, it's surprising how complicated rules are solveable.
Zendo is great - I played it again with my nine year old godson last weekend when he was here having a sleep over.
Tangentially: this kid is an awesome proto-nerd, and I fully expect that he'll be posting at HN in another 6 to 8 years - for the last two years he's wanted to program a computer game, and I told him that he and I would learn HTML5 and JS together when he turns 10. We've already got the domain name picked out.
If you haven't engaged with an intellectually precocious kid, you can't imagine how fun and rewarding it is.
Voting games can be painful. I remember Mafia, which is very crowd-dependent. People who slow-play/minimax games make Mafia intolerable. At 3:00 in the morning with 3 players left, the 30-minute discussions are painful. Just kill someone and lose or win already.
The variant I built (to give something useful to start the discussion, rather than blind suspicion) had a combat mechanic. You actually had to "buy" arms in order to do your killings. This meant that Mafia actually left a trace, because it was public information who purchased. (There was no in-game currency. The cost of buying arms was visibility.) If you were Mafia, you'd probably want to buy arms. On the other hand, the good guys also could benefit by arming up. The result of this was that there are two different strategies. You can not buy arms often (or ever) and evade suspicion during the day, but be weak against brawls and Mafia... or you can buy them at every opportunity and be strong in combat, but arouse suspicion and be more likely to die in the day phase.
There was also a brawling phase. If you thought someone was Mafia and wanted to go vigilante, you could. However, they could defend themselves. in addition to dying, you could end up wounded, which would make it obvious the next morning that you were in some kind of fight.
- Actual data to base your votes on (Mafia can be fun, but the games of it I play never have 30 minute discussions because there's just nothing to discuss -- we just pick someone semi-randomly to lynch.)
The commercial release is very nice, but it can be played with regular cards too.
I used to play board games with people like that, and the problem is easily fixed with a kitchen timer :)
Timer goes off? Turn over, too bad. Given the average number of turns in a game and how long you want to play for, you can work out what to set it at (maybe with a 25-50% leeway)
An interesting variant is to put the timer in a box so that you don't know how long you have left.
I'm not sure what to replace it by, admittedly. I did one experiment where instead of saying what the rules were, your actions were taken as an example of what you thought the new rules should be: if you did something not previously legal, it was inferred to now be legal [1]. But that just led fairly quickly to chaos, unless it were carefully restricted. So maybe not the right approach either. But I still like the idea that the game-mechanics changes themselves are related to the game mechanics in a cybernetic-feedback-loop kind of way, rather than a completely "outside the system" free vote. Now as to how to implement that in a way that makes for a playable/interesting game...
[1] http://www.kmjn.org/notes/reflexive_rules.html