Saturday, December 26, 2009

More on Gambling and Auctions

Earlier I wrote about an idea for using gambling mechanics as a sub-system within the game. I still think this idea has merit; clearly gambling games hold our attention, as they are widely popular. I also like the idea of different subsystems to resolve different tasks.

Upon further thought I'm not exactly sure I'm implementing it in the right way. When you gamble in poker, in general you place your bet based on how strong you think your hand is relative to every other player's hand. Obviously there are strategies which deviate from this such as manipulating the pot odds which makes the game interesting, but that's the general rule.

What I am trying to do at the moment, however, is find a free market, interactive and exciting method for determining how "costly" a commodity (success at a task) should be. Bear in mind that "cost" could be a finite amount of resources (X Healing Surges, Y Spells, Z Gold Pieces, whatever) or a percentage chance of expending resources (X% chance to die, Y% chance to lose healing surges, etc).

One traditional method is an auction. Basically, in an auction, you take an item for which a general cost is known and allow the purchasers to decide its value. Just looking at Wikipedia, there are a wide variety of auction techniques out there.

One major problem I can see right away with using auctions is collusion. Unless you're playing a game like Agon that pits players against each other, in general, the players will want to collude to get the best possible "price" for their "item" (success at the skill check), because usually their goals overlap.

I see a few ways to beat collusion:
- Offer the players incentives to compete against each other. Agon does this rather well with its emphasis on individual glory over group goals. It doesn't matter if the whole team makes it from Athens to Marathon; what does matter is who makes it first. This gets closer to creating a true market.
- Allow the DM or NPCs to bid on the task as well (which denies the players success if they are unwilling to match the bid). This however gets us back to the issues of requiring the DM to arbitrarily come up with the "costs" for items.
- Use a Dutch Auction. Dutch Auctions are less vulnerable to collusion. However, I think you'd still have to include one of the former methods too; just using a dutch auction alone will not be enough to solve the issues.
- Use a silent auction. This reduces the information available to the players and thus they will need to wager more to ensure winning, especially if one of the first two methods is in play.

So, in short, I think a Silent Dutch auction combined with either NPC/DM bidding (easiest to implement, theoretically) and intra-party competition would be the ticket.

Here's an example of play using some 4E terminology in italics (but feel free to substitute resources from your favorite game:

DM: "Ok, you come across a chasm with boiling lava running through it. The first one across will win an Action Point for their daring and pluck. Bids will be in increments of Healing Surges, with a maximum bid of six surges and a minimum of zero.

Additionally, I'll roll 1d6-3; anyone who bids less than this secret reserve will lose their nerve and fail to make the crossing, and be stuck on the former side for a turn while they hem and haw and get their nerve back up. And of course, every turn in this dungeon requires a wandering monster check..."

All the players then think about what is at stake and how much they value winning that Action Point. They all notate how many surges they are willing to spend, perhaps by secretly setting up a six-sided die in front of them. When all the "bids" are uncovered the one who is willing to bid the most will be the "winner" who gets across the chasm first. The party may well be split if some players fail to meet the secret reserve price; to guarantee getting everyone across, all will have to bid at least 3 surges.

Great Success: The winner of the auction who bids the most. They pay their bid, cross the chasm, and win the action point.
Success: Anyone who bids more than the reserve but does not bid the most. They cross the chasm.
Failure: Anyone who fails to meet the reserve. They fail to cross the chasm and waste a turn to boot.

As a twist, you could work skills into it in several ways. Perhaps skilled characters automatically get a +1 or +2 bonus added to their bid if they can explain how their skill is relevant.

Saturday, December 19, 2009

Gambling for Skill Checks

One issue I've been thinking about recently is skill checks. One problem with them is that the difficulty class (to use a nasty 3.5 word -- others might refer to it as the "odds of success") is essentially arbitrary. In 1E, the DM must decide if a save is at a penalty or a bonus. In 3.5, they must set a DC. In White Wolf, they need to decide if there is a dice pool bonus or penalty, and how many successes need to be accrued for an extended task. In 4E, there are at least "suggested" DCs for Easy/Med/hard tasks, but the DM must still decide what level the task is and whether its E/M/H. In C&C, the DM must decide what "level" the task is.

All of these methods are at best arbitrary and at worst highly judgment dependent. They can also cause consistency problems; if the DM makes an off-the-cuff ruling on DC every time, then eventually some inconsistent results will occur ("Last time we tried to make that leap, it was DC 15, why is it DC 20 this time?"). If the game includes a long table of modifiers or preset DCs then it becomes impossible to memorize and quite hurky.

I was brainstorming some possible solutions and an idea jumped into my nugget while driving to work. Why not create an "economy" around skill checks? Instead of a command-and-control centralized planning system where the DM sets all "costs" (DCs), why not a "free market" model where the players can barter on what the true cost of an action is? One model for that is a gambling model. For example, in a round of poker, each player estimates how much their hand is worth in relation to the pot.

One problem with this model is that the players may generally have an incentive to gang up against the DM; all members of the party probably want to succeed and the adversarial DM wants them to fail. So the system must either strengthen the hand/influence of the DM (i.e., different amounts of pull or influence), impose transaction costs, or both.

Here's an example of how such a system could work.

* A player wants to leap across a chasm.
* The DM decides whether its worth opposing or not. If the DM decides there should be a risk of failure or an expenditure of resources, then he will oppose it. If he decides there is not, then he just allows the action to succeed, no check required. This step should reduce frivolous checks.
* If the DM decides its worth opposing, he "antes up." He puts some resource into the pot that everyone at the table values. For example, this could be action points (usable by NPCs if in the hands of the DM or by the PCs in the hands of the players), "plot points" (again, usable by both sides), healing surges, gold coins, whatever.
* The player then decides if they are willing to ante up to match the DM's wager. If not, then they do not succeed (nothing ventured, nothing gained). If they meet the ante, then the gods rake the pot and then the mini-game begins.

Raking the Pot: The "Gods" or "Fate" should take a cut from the pot. These chips should effectively leave the game (although groups could allow them to re-enter the game in certain circumstances, such as the common house rule in monopoly allowing people who land on "Free Parking" to take the "kitty" of previously paid fines). This incurs a transaction cost and discourages frivolous skill competitions which waste everyone's time.

One could use a variety of models for the minigame -- poker, dice games, etc. Let's go with a simple variation on "klondike," a poker like dice game.

*Throw a total of five dice. In traditional klondike, the order of dice is 1 (highest), 6, 5, 4, 3, 2 (worst).
*Score "hands" of dice to determine the winner. From best to worst:
* Five-of-a-kind
* Four-of-a-kind
* Full House (Three-of-a-kind and a pair)
* Three-of-a-kind
* Two pairs
* One pair

* For a quick and simple skill check, you could just play a quick version of Klondike as above. The player and the DM roll 5 dice and see who the winner is. As a wrinkle, highly skilled characters could roll more dice and take their favorite 5, or reroll some dice.

The winner gets to take the pot, but either way the action succeeds. Note that its up to the DM to set the risk of the action by anteing up at the start.

* For an important skill check, the process should be more involved. Either stud poker (such as 7-card stuck) or community card (Texas Hold 'Em style) would make good models. These include multiple rounds of betting and allows for more strategy. Here's a procedure using a variant on Omaha Hold 'Em (similar to Texas Hold 'em) rules.

- Ante up
- Each player rolls 2 dice as their "hole" cards. If skilled, roll 3. If highly skilled, roll 4. These should be kept secret if possible.
- Bet
- Set up community pot by rolling three dice ("the flop")... Anyone can use these dice; they are shared.
- Bet
- Add one more die to the community pot ("the turn")
- Bet
- Add a final fifth die to the community pot ("the river")
- Bet
- Rake the pot
- Score hands and determine winner. Players must use exactly TWO dice from their private "hole" dice and exactly THREE dice from the public community dice. The winner gets the pot. The action always succeeds; we are just determining how much the cost is.

Here's a wrinkle: Say additional players want to help with the skill check. They should be allowed to ante in and play the mini-game as normal. However, for each additional player, the DM gets to add one die to their private hole. This, plus the rake of the pot, should limit unlikely to succeed "aid another" attempts. This also creates competitive motivations within the party, as its a winner-take-all system.

I'm sure this needs some refining, but I think it could be quite interesting.