CORDIAL MINUET Forums

It would be nice if the tournament pages listed profit in coin instead of/in addition to dollars. Especially since the pages don't list any information like buy-in or per-game stakes.

Is it time to ask yet whether these regular tournaments are working? Perhaps they're too frequent to attract people. Also I've heard several people say they preferred longer ones, myself too.

Wow, it's a happy coincidence that 52 = 6*6 + 4*4. I'm impressed at this squeezing of poker into CM, awesome prototype!

I haven't managed to get into a game with anyone, but I did play a couple rounds against myself.

I'm surprised at the way that  community cards work. So you've replaced CM's reveal-a-column step with revealing the full set of 4 community cards. Before and after the reveal there are respectively 24 and 6 possible hands for each player, compared to CM's 6 and 2. That's quite innovative, but I 'm not sure I agree with hiding two of the community cards. Yes, I think having more than 2 possible hands on the final betting round is an improvement. But I think this worsens things in earlier rounds.  It adds so much uncertainty because there are so many possible hands to consider, in addition to more choices at each move. I feel like I won't manage to do that in my head, which is going to make it hard to figure out what hands the opponent is comparing and hence what they will do. Looks like this is a game about working out hand odds and betting instead of reading minds.

Also, there's a second potential problem, which is that if you want to pick two shared cards with no preference to their order, then having each player pick two rows/columns would be bad, because each player has no preference to the ordering of their row/column, resulting in 50/50 chances of each two different pairs getting picked even after both players pick their columns.
Now, that's not the situation here because there is a preference: you get to see one card and not the other, so you would ideally order columns so that that result of one column is important to you or both players and the other one isn't, so that there's information asymmetry. Still, that seems difficult to consider. A lot of players thought CM's experimental mode was too difficult to reason about, and that's miles simpler than this.
Of course, if you only picked one community cards column per move then there would be nothing to reveal in the reveal step, it would be a big change.

When you finish the round it breaks. Firstly, the community cards grid doesn't update to show the two cards the other person picked after they are revealed (in the Community Cards: line). Secondly, I see this, after going all-in on the second last betting round, with 10 coins initially in the pot and each player having 90 coins. (It doesn't seem to screw up without an all-in).

You won

Your stack: 0
Their stack: 95

Pot: 10

Which doesn't really make any sense. After going to the next board, I get

Your stack: 8
Their stack: 93

Pot: 4

Cool! So each tournament occurs every 14 hours, so there would be around one per day per stakes during busy hours.

Yeah, yikes, I see that the server is only refunding in the case where 1 person entered.

Even refunding when no games are played in the tournament isn't ideal either. What if I enter and sit around for the whole contest with no one to play with, meanwhile 2 (or 4 or 6 or 8...) other players join and play games against each other. Seems like entering a tournament in off peak hours is actually very risky with the current rules, it certainly puts me off entering them. The obvious solution is to refund your buyin if you don't get to play any games. Sure, for a "serious" tournament it would be your own fault if you buyin and then don't play any games, but if it's out of your control then it's not fair and a big deterrent.

Also, consider the case where 2 people enter the tournament and they play games back and forth for hours, and end up nearly even, but both with negative profit. No one wins a prize either.

I think these cases are quite likely to happen all the time since there are tournaments all day and most of them will have very few contestants.

Is it possible to run two tournaments at once? Of course only makes sense if the stakes are vastly different.

Is the time listed on the past tournaments list the start or the end time? (Column header maybe?) Would be nice if it gave the duration too.
It would be nice if the li

We're just trying to explore the space of possibilities a bit. Certainly none of these ideas are quite as simple as CM.

Since that's the case anyway, I don't see the use of the "restriction that the index of a chosen row can't match the index of a chosen column".

(EDIT: OK, I worked out the reason: it's so that you learn something about the column the opponent picked for themselves, in the same way as CM. That makes this variant is very similar to CM.)

I don't think the possibility for such a tie is a problem. CM has ties anyway. We can fairly confidently assume there is no pure strategy which is optimal in any of these variants (which I think Jason or Josh showed for CM?), which makes the possibility of the players making identical moves unimportant.
But those sorts of ties aren't unavoidable in a perfectly symmetric game. Consider rock-paper-scissors. For example if CM was played on a Latin square instead of a magic square then it would be unbiased to either player. But it would also be much more boring.

Magic squares don't always use the numbers 1 ... n*n. I suspect that if every number appeared twice (1,1,3,3,...35,35) then it would be possible for the transpose of the square to be a permutation of the rows and columns of the original. This would require that every four cells forming a square contain both duplicates of two numbers, like:

. . . . . .
. . . . . .
. . . . . .
. 9 . . 3 .
. . . . . .
. 3 . . 9 .

And maybe it would then be possible to deduplicate the numbers (change one 1 into a 2, etc) and get a magic square with zero or very low bias?

This is really mind-bending, so I'm not sure I'm getting this right...

Firstly, there are different ways of comparing the player's hands. There seem to be two obvious ways of doing it.
1. If one player has a pair (of the same rank), the other sees them having a pair of the same suit, which is normally worth nothing. So one solution is to make the rank-suit swap of a hand equal in value to the original (e.g. a flush would rank equal to a triple). However then there's no point doing the rank-suit swap at all, as it would be merely cosmetic.
2. What I am pretty sure you actually meant: each player gets the hand THEY see, so you can give the other player what looks like a pair, knowing it's actually worthless to them. (Assuming hand ranking is done in an asymmetrical way similar to Poker). This seems like it could add a bit more depth to the game.

Such a mirroring of rank and suits isn't a trick specific to the suited game. In general for any game variant defined by a "hand" ordering rule (e.g. sum of values in CM) you could label each square with two "values": what player 1 receives on getting the square, and what player 2 receives on getting it. In your rank-suit swap there is a simple mapping between the two values which makes it easy to represent both of them at once. But I could just as easily combine a normal 6x6 magic square and its transpose into a single, so there is no relation between the two numbers ("us" and "them") in a cell. As long as your transpose-mirror condition holds, this game is completely fair with no bias to either player, unlike CM.

Hmm, I would take part, but Google Codejam round 2 starts at 2am my time and this is at 6am after a 90 min break. Makes it pretty hard to do both.

I noticed that you could receive a prize even if you've played no games. Was that intentional? I had thought it was discussed before. It's not unfair though (you won't be winning back your buy-in).

Sadly these short contests are still awfully susceptible to collusion  - with the 2 cent profit limit you can play 3 games and win $0.0297 profit, which seems like it's doomed to always be high up the scoreboard and make collusion profitable. Is the contest this weekend going to run for 3 hours? I think all the amulet contests we had showed that longer contests can work

Hah, I wasn't paying attention, but luckily I happened to finish off my last opponent a few seconds before the end of the contest, so managed to win back some of by buy-in. Contest ended and prizes paid between when I won and when I left.

You didn't fix the 10 cent typo.

Awesome.

Hah, if both players record their game with sparse commentary, speaking less than half the time, conceivably you could show their two boards side by side and splice together the audio tracks. That would be fantastic but surely also a heap of work. Maybe try that later. (Actually, looking at the two boards at once is quite confusing. The colours are opposite, rotated, an in different places, even the score graphs don't match up, so they actually look very different. If you used the recordedGames file I guess you could mod the client to swap the colours or even rotate the board, and then playback the recorded game to record a new video, so that they're easy to compare.)

Sure you could run a $20 buy-in tournament for the 3.2% of the playerbase who can afford it. Running tournaments should also attract some bigger investments. But I think it would be insane not to also run plenty of lower stakes tournaments for the enjoyment of everyone else. (I wouldn't be entering a $20 one)

Having the chance to win far more than you're risking is surely the main appeal of gambling. Hence the excitement generated by CM tournaments. Erm, we haven't had one in a while. Weren't weekly tournaments floated as an idea? Think quiz night - a developed habit. (Run a poll? "What night of the week...")

It would also be possible to emulate the chance to win big from a single CM game. Such as bounties for defeating a player (like in some poker tournaments, I read). Basically the same thing as winning amulet points. Unfortunately, collusion....

Speaking in fragments. Four AM.

I don't see why the same couldn't work for CM (or online poker or two-player skill games): a few fish losing a lot of money, so that the pros randomly get paired up with other pros most of the time (breaking nearly even on average) and some of the time win off the fish, on the whole profiting.

That Japan Jackpot thing is pretty nuts. All kinds of roulette wheels and minigames, and half of them are physical!

Hmm, it does indeed create the tables with those fields; you're right that there's nothing wrong. I probably accidentally created the databases with an earlier hg revision before switching to later version.