Website for my new game

jasonrohrer · 2014-08-26 10:31:46

Here's what I've been working on...

http://cordialminuet.com/

thoth · 2014-08-26 10:58:41

sounds intriguing, where's the buy button?!?

cullman · 2014-08-26 11:31:58

Looks fascinating. When is the beta?

jasonrohrer · 2014-08-26 14:11:18

There will never be a buy button. :-)

Once the game is live, it will be available for everyone to download, but you have to deposit money into the game in order to play against someone (you enter your CC directly into the game interface, securely).

So, you can only play the game for real money (no pretend money games), but the smallest possible game buy-in is 1 cent, so you can learn to play with low risk.

The game is fully playable and functioning now, albeit with placeholder graphics. I've been testing it with a few local friends (and winning/losing money in the process).

I'm not sure about the ETA for more broad testing.... a few months at most.

AMWhy · 2014-08-26 17:15:46

Demonic Ritual...

voxel · 2014-08-30 22:00:20

jasonrohrer wrote:

you have to deposit money into the game in order to play against someone (you enter your CC directly into the game interface, securely).

It think that will definitely scare people off (I can't name any other games that do that); is there a big enough advantage (convenience?) over depositing through a webpage to justify it?

42dustman · 2014-09-01 11:50:32

voxel wrote:

jasonrohrer wrote:
you have to deposit money into the game in order to play against someone (you enter your CC directly into the game interface, securely).
It think that will definitely scare people off (I can't name any other games that do that); is there a big enough advantage (convenience?) over depositing through a webpage to justify it?

It did scare me off already!

Blip · 2014-09-01 14:02:59

I agree with dustman here. No matter how much I want to play this game and how much I trust Jason, I'm not going to enter my credit card number. If there was some way to hook the game up to a PayPal account that would be better; that way I could put a limited amount of funds into the account so that even if there was a security issue, I couldn't lose more than that amount of money.

cullman · 2014-09-01 15:27:19

The other option is just buy a prepay Visa in the amount you want to spend on this game.

mala · 2014-09-10 18:45:07

i play online poker everyday, and beside the fact i trust the platform i'm playing on, i always use a prepaid visa to make transaction that stays empty most of the time... just in case.

@ Jason: online poker rooms use to send money automatically on the last card that has been used to deposit money, and if you want to withdraw on a different one you have to pass trough the user support. That grants another layer of security in a really simple way. Maybe you can implement something like that to make things safer.

iceman · 2014-09-23 20:40:29

Just found this article on Jason's new game:
http://kotaku.com/a-mildly-satanic-new- … 1638083439

mala · 2014-09-24 05:17:33

as a former poker player, i can't wait to try it.
The game seems really challenging.

@Jason: i want to address you something : Known Nickname vs Random ID

keeping the players identities a secret will have a huge impact on the strategical aspect of the game.
Since i'm always playing against "Random villain" i will have to find a well balanced strategy to apply to each and every game i play, regardless of my opponents tendencies.

Given that someone finds the right "game theory optimal" strategy, that will result in an endless series of EV=0 games, but since there's rake it actually became EV-.

On the other side, by knowing my opponents identities i can start "unbalancing" my game to exploit his weakness, thus rendering the game again EV+ for those who adapt first and better.

For those of you that are unfamiliar with the game theory concept, here's a brief explanation based on the "Rock, Paper, Scissor"

http://gametheory101.com/Rock_Paper_Scissors.html

Last edited by mala (2014-09-24 05:50:06)

Passive · 2014-09-24 08:50:17

Blip wrote:

I agree with dustman here. No matter how much I want to play this game and how much I trust Jason, I'm not going to enter my credit card number. If there was some way to hook the game up to a PayPal account that would be better; that way I could put a limited amount of funds into the account so that even if there was a security issue, I couldn't lose more than that amount of money.

Thired(?) I would like to limit how much i had at stake at a time, like other online gambling. Also how are you going to make money from it? Taking a rake?

cullman · 2014-09-24 09:15:16

mala wrote:

as a former poker player, i can't wait to try it.
The game seems really challenging.
@Jason: i want to address you something : Known Nickname vs Random ID
keeping the players identities a secret will have a huge impact on the strategical aspect of the game.
Since i'm always playing against "Random villain" i will have to find a well balanced strategy to apply to each and every game i play, regardless of my opponents tendencies.
Given that someone finds the right "game theory optimal" strategy, that will result in an endless series of EV=0 games, but since there's rake it actually became EV-.
On the other side, by knowing my opponents identities i can start "unbalancing" my game to exploit his weakness, thus rendering the game again EV+ for those who adapt first and better.
For those of you that are unfamiliar with the game theory concept, here's a brief explanation based on the "Rock, Paper, Scissor"
http://gametheory101.com/Rock_Paper_Scissors.html

I agree, it should't be anonymous usernames or random usernames. I think while it was a cool idea for TCD, it was a failure in the sense it seemed anyone could figure out their name. Jason has stated that he wants it to be random and anonymous to prevent money transfer, but the reality is if I ask to play a game for $2173, wouldn't it be obvious to who I want to transfer money with who is me? Unless you can only play in units $10 or $100s when you get above the single $1 level.

voxel · 2014-10-02 04:13:13

This is a very interesting game. Jason's clearly hoping that the use of a magic square as a payout matrix means that each players have equal possibility to win or draw; that is, that a zero-sum Nash equilibrium exists for every possible board position. The most obvious candidate is both players choosing columns uniformly at random, which is I believe the assumption you're making. However, I'm not convinced, and might run this past some mathematicians here. Unfortunately the strict "subject to chance" test means that if this assumption fails, you're in trouble. (I've never before seen anything where mathematics could be applied so directly to a legal decision, pretty interesting!)

Even more importantly, how the hell are you going to keep bots out of this game? And the complete source code for the client is available, so there's nothing stopping this from becoming bot infested.

cullman · 2014-10-02 08:42:20

Perhaps bots aren't necessarily a negative. In California card rooms (which differ than casinos because the house does not gamble it only collects the rake), you will see these guys that play either poker or California blackjack for companies. They are literally paid employees that sit there with a big rack of chips in front of them, they make $20 an hour and they play the game with a very explicit set of rules. They are essentially human bots. Now granted, they don't have perfect memories or the ability to do millions of calculations a second, but they do actually in some ways improve the experience at the "casino", as there is always someone there to play. Also, if you can figure out the rules, or the fact that they play a certain way all the time, it was often easy to take money from these guys. These guys were essentially designed to sit there and grind away money from the people who don't play the game well at all. Anyway, I don't know the rules of CM that well, and I haven't thought that much about whether or not someone could create a bot that is better than playing ANY human, I do, however, think there is a much greater chance of people creating flawed bots vs unflawed bots.

Another way to think of it is this, if I was playing against a poker bot and knew I was playing against a poker bot online, I could quickly take advantage of the fact that bot is going to play certain cards a certain way every time. Bluffing, is a particularly hard thing to implement in a bot.

I don't know if the game is going to allow you to chat with your opponent, but perhaps that is a way that would help someone determine whether or not they are playing a bot. Now someone has to create a bot, that also has a chat engine that can pass the turing test

Osmium_Planet · 2014-10-02 11:45:22

Ok, I'm on this Website, and I can't find this game. Call me a noob or what, but I can't find it. The only Link on the page is to Jasons page. Is it a code or something I have to solve?

jere · 2014-10-02 12:19:19

Yes. Jason posted a riddle that you must solve.

I'm not sure about the ETA for more broad testing.... a few months at most.

I'm super pumped about this. If it's anything like the real money contest, it'll be quite intense.

My main concern is the possibility, as others have mentioned, that the optimal strategy will be random. Essentially: what's if it's just a more complicated version of rock, paper, scissors? On that note, would a real money rock, paper, scissors game be considered a game of skill? Would it skirt gambling laws? Regardless, I want to try it out. It reminds me a lot the game Yomi.

voxel · 2014-10-02 21:30:36

Anyone can download the source code and create an account. There is also a webpage hosting binaries for Windows and OSX, but I'm not going to post the link because if Jason wanted a flood of users he would have posted it himself.

I created an account. I'm user 26. Considering that there were only 7 users before the hacking challenge was put up, it appears that most of the game's users are hackers It's a good thing that unlike TCD, CM can be made cheat-proof.

cullman wrote:

Another way to think of it is this, if I was playing against a poker bot and knew I was playing against a poker bot online, I could quickly take advantage of the fact that bot is going to play certain cards a certain way every time. Bluffing, is a particularly hard thing to implement in a bot.

Sorry, no you couldn't. If you were playing against a good poker bot then it would have no significantly exploitable weakness (they aren't perfect, but they can get very close to it). I have actually taken a course on game theory, read poker AI papers and implemented a simple 2 player Texas Hold'em poker bot. However I will agree that bluffing isn't easy. Good bluffing requires having a model of your (human) opponent. But as someone pointed out up the thread, in CM all opponents are anonymous, so you apparently can't learn your opponent's behaviour anyway. Anyway, here is where you misunderstand: you don't need an opponent model; you can just play the Nash equilibrium and you won't lose money.

Thanks for the information about California card rooms though, very interesting.

The ONLY part of Cordial Mineut which is non-trivial for a bot is bluffing during the betting/raising part, based on each player's partial knowledge of the other. CM definitely looks like poker distilled to its essence, with all that randomness removed. Pretty brilliant. Since the game gives you that partial knowledge at the side of the screen I guess that the aim is to reduce the advantage that a bot has over a human in computing what their odds are. However those bars don't convey everything that can be learnt (eg. exact odds), so bots are still at an advantage.

Computing the optimal strategy for this game is trivial: the game tree (ignoring bets) has only 518400 leaves. For comparison, that's only twice the size of tic-tac-toe! If I had time, I would write a CM bot which you could play against locally (I wouldn't release it on the official server) in order to investigate game strategies.

Last edited by voxel (2014-10-02 22:43:23)

cullman · 2014-10-03 00:23:46

voxel wrote:

cullman wrote:
Another way to think of it is this, if I was playing against a poker bot and knew I was playing against a poker bot online, I could quickly take advantage of the fact that bot is going to play certain cards a certain way every time. Bluffing, is a particularly hard thing to implement in a bot.
Sorry, no you couldn't. If you were playing against a good poker bot then it would have no significantly exploitable weakness (they aren't perfect, but they can get very close to it). I have actually taken a course on game theory, read poker AI papers and implemented a simple 2 player Texas Hold'em poker bot. However I will agree that bluffing isn't easy. Good bluffing requires having a model of your (human) opponent. But as someone pointed out up the thread, in CM all opponents are anonymous, so you apparently can't learn your opponent's behaviour anyway. Anyway, here is where you misunderstand: you don't need an opponent model; you can just play the Nash equilibrium and you won't lose money.

This message board is a constant source of humility for me, no place am I told more how little I understand about things I do appreciate you did it in a respectful way, though, voxel. And I am the first admit that I have much to learn about many topics (especially Cordial Minuet, I have not played, and barely read anything about it), but I do know a little bit about AI, and a decent amount about Poker and poker bots (with the caveat I haven't been in the the online poker world since "Black Friday"). I assume you are familiar with Hold 'em Inspector? I will take my chances against a Poker Bot with Hold em inspector on my side any day of the week. I have a friend that wrote a very very successful poker bot. Your BASIC poker bots make money in two primary ways :

- Grinding perfect game play against fish, with the added advantage of perfectly adjusting the odds on the fly based on every card seen (just like Hold em inspector can do). In fact, anyone can be as smart odds wise as a bot with hold em inspector.

- And the main way poker bots (used to) make money....sitting at multiple seats at the same table. While it is a big advantage for even humans collaborating, it is enormous advantage for a well written co-ordinating poker bot. If you hear about Poker Bots making money, this is the main way they do it.

Now let's talk about advanced poker bots. Advanced poker bots also can play perfect odds poker, with the added advantage of not only having a perfect math when it comes to card probability/pot odds, etc, but they can also keep perfect records on how individual players have reacted and played in certain situations. Does cullman tend slow play with pocket pairs? How often has cullman been caught bluffing, and so on. This is where a poker bot can have a huge advantage. This huge database of opponent behavior in many situations combined with perfect odds play. Now put 2 bots back channeling together on the same table - and completely forget about even getting lucky for an hour at that table.

I stated that "If I knew I was playing a bot", which is basically impossible, I was sort of speaking rhetorically without saying so and thusly I suppose I deserve to be accused of not understanding. Because I don't know how you would know you were playing a bot in Cordial Minuet. I guess my point is this (and I am certainly open to the idea my point of view is wrong - often is), is that in Cordial Minuet you cannot have 2 bots at the same table against one human. And more importantly in CM the bot doesn't know who they are playing, so it has no history of how cullman may play in certain situations, am I aggressive? am I slow player? It just doesn't know. So maybe best case the bot plays perfect strategy (if there is perfect (odds based) strategy for this game, again I am so ignorant about CM, I don't even know if that is possible). Anyway, it seems to me it's no worse than playing against another player that plays perfect strategy. I am sure we all have had experiences at the poker table where someone much better gets furious because you just beat them by "playing stupidly, or wrong". Because they were playing the odds, and playing like their opponent should be playing the odds. It seems that bots will be susceptible to that, just like any other theoretical perfect strategy player. Don't get me wrong, I think there is definitely some edges to be found through computation in this game or in any game, all I am saying is I don't think bots will necessarily ruin the game, or come take it over. Plus, I think if there is a rush of people writing bots, I think they might be surprised just how hard it is to write a bot that can beat a human, especially a dumb unpredictable human, or even more especially a human that is picking and choosing strategy it may be getting from it's own bot ala Hold Em Inspector - the best of both worlds a machine's analysis and perfect memory (though that memory across games matters not in CM), combined with the intuition, psychology and randomness that a human can provide.

Hope my comments are taken with friendly tone they were intended to be delivered with.

Osmium_Planet · 2014-10-03 01:45:26

So, uhhh...I've solved the anagram. 3 times. Does it give you a URL solution or something? I'm missing something.

(I don't want the answer, just an idea of what I'm looking for. I always look too deep on these kind of puzzles, so what's the final kind of format I'm looking for?)

voxel · 2014-10-03 03:47:53

Well, I haven't actually played CM either; there aren't exactly many players. But the Kotaku
article I think covered pretty much everything there is to know, barring aspects
yet to be added.

I admit, when I talked about not being able to win money off a good poker bot,
I meant a poker bot designed not to lose money, which probably means one designed for
competitions against other bots. Real bots playing on gambling websites are going to instead
be designed to win money off humans, and to do that it's far more important to exploit your
opponent's weaknesses than to ensure you don't have any weaknesses yourself (the two are mutually exclusive).
So it's probably true that these bots are predictable and beatable by better humans.

I don't actually play poker. Writing a bot and doing all that reading was my introduction to it, and I haven't gone back since that time. :)
However I'm really interested in CM, it seems like Poker Improved.

I think you're right about everything you said about CM and poker.
Jason was surely expecting bots on CM all along. Anyone can use
a tool to compute odds, and already has decent information at the side of the screen by default
(maybe this should be improved. Someone can always fork the client to improve it).

Does it give you a URL solution or something?

Well it's that old trick for finding interesting webpages like admin panels and embargoed press releases: you just add obvious page names.

mala · 2014-10-03 17:52:56

voxel wrote:

Sorry, no you couldn't. If you were playing against a good poker bot then it would have no significantly exploitable weakness (they aren't perfect, but they can get very close to it). I have actually taken a course on game theory, read poker AI papers and implemented a simple 2 player Texas Hold'em poker bot. However I will agree that bluffing isn't easy. Good bluffing requires having a model of your (human) opponent. But as someone pointed out up the thread, in CM all opponents are anonymous, so you apparently can't learn your opponent's behaviour anyway. Anyway, here is where you misunderstand: you don't need an opponent model; you can just play the Nash equilibrium and you won't lose money.

by playing the Nash equilibrium you won't lose money, but you won't gain money either.
subtract the rake => every GTO player will be a looser.

If you do not "unbalance" your strategy based on your opponents tendencies, you cannot win money in the long run.

Game theory chapter 1.

cullman · 2014-10-03 18:14:23

mala wrote:

If you do not "unbalance" your strategy based on your opponents tendencies, you cannot win money in the long run.
Game theory chapter 1.

I thought Game Theory Chapter 1 was, "The only winning move is not to play". Or is that from movie Wargames? While I'm not certain about that, I am certain that Chapter 2 of Game Theory is, "You don't talk about Game Theory".

The Castle Doctrine Forums

#1 2014-08-26 10:31:46

Website for my new game

#2 2014-08-26 10:58:41

Re: Website for my new game

#3 2014-08-26 11:31:58

Re: Website for my new game

#4 2014-08-26 14:11:18

Re: Website for my new game

#5 2014-08-26 17:15:46

Re: Website for my new game

#6 2014-08-30 22:00:20

Re: Website for my new game

#7 2014-09-01 11:50:32

Re: Website for my new game

#8 2014-09-01 14:02:59

Re: Website for my new game

#9 2014-09-01 15:27:19

Re: Website for my new game

#10 2014-09-10 18:45:07

Re: Website for my new game

#11 2014-09-23 20:40:29

Re: Website for my new game

#12 2014-09-24 05:17:33

Re: Website for my new game

#13 2014-09-24 08:50:17

Re: Website for my new game

#14 2014-09-24 09:15:16

Re: Website for my new game

#15 2014-10-02 04:13:13

Re: Website for my new game

#16 2014-10-02 08:42:20

Re: Website for my new game

#17 2014-10-02 11:45:22

Re: Website for my new game

#18 2014-10-02 12:19:19

Re: Website for my new game

#19 2014-10-02 21:30:36

Re: Website for my new game

#20 2014-10-03 00:23:46

Re: Website for my new game

#21 2014-10-03 01:45:26

Re: Website for my new game

#22 2014-10-03 03:47:53

Re: Website for my new game

#23 2014-10-03 17:52:56

Re: Website for my new game

#24 2014-10-03 18:14:23

Re: Website for my new game

Board footer