Giving players control of risk

[Adam Dray]

In David Bapst's Difficulty/Success Needed as it relates to GM Fiat, people posted a variety of ways to remove GM Fiat from setting difficulty levels. In my Verge game (in progress), I am using an alternative that I think is very obvious: let the players choose the difficulty.

The key to making it work is tying the level of risk to the reward system. No risk, no reward. High risk, high potential for reward.

In Verge, the GM gets no dice that the players (collectively) do not give him. If a player wants, his character automatically wins a conflict but he will not be rewarded for that. Players have two opportunities during a scene to give more dice to the GM to roll. There's still an element of luck due to reroll mechanics. Players are encouraged by the reward system to give the GM as many dice as they dare while managing still to win the contest. They must win to get a chance to roll Experience dice, which are the same number of dice the GM got to roll in the scene.

I've added on top of that a mechanic called Drive that gives players free dice in every conflict but also subtracts from the Experience dice. Whenever a player rolls 4 successes on Experience dice, their Drive increases a point.

Thus, the more a player takes risks, the more Experience he earns (and he increases the traits that gain him rerolls that win conflicts), but his Drive also increases slowly, forcing him to take greater and greater risks to earn Experience.

Have any other games given risk-control to the players? I'd like to see what other games have done in this area.

[Mike Holmes]

In version 2.0 of Synthesis (a game of mine), I've done precisely what you suggest. In that game, the result of any conflict is a trait equal to the level of successes rolled. But the total level is capped by the difficulty of the task. So if the player takes on a level 1 difficulty, they will likely gain a level 1 trait. Interestingly, the higher they go, the less likely they are to achieve the cap, and the more likely they become to have a trait assigned against them.

In effect, however, there's a mechanically determinable sweet spot to my system. That is, for every level of ability that you're rolling, there is a TN that you can choose that gives you the greatest expected value. I've never noticed anyone do the calculation, in part because A) it's not simple to perform, and B) becuase maybe they just don't care. But it would be a simple thing to produce a table with the expected value, and let the player select from off of it. If that were to happen, I'm not quite sure what it  would do to people's play.

I think your system has the same "problem." If you changed your system, however, to where the reward is not based on needing to succeed, then you'd eliminate the problem. Of course, you've probably capped it with success because you've noted that if you don't then a player can go up against One million difficulty, and get a lot of rewards - basically there's a point at which they'll just assume that they're going to fail, and then pick an arbitrarily high difficulty just for he rewards.

But that assumes that one failure is the same as the next. If you make worse failures something that happens with higher difficulties more often, then the player has an incentive not to lose by too much. In which case these things might balance.

This gives me an idea for Synthesis 2.1 :-)

Mike

[Adam Dray]

I've also not given the primary stakeholder sole control of the risk. If the other people at the table think the stakeholder needs more risk, they can give the GM dice. There's no currency involved. Just a limitless pile of dice that any player can hand to the GM.

[Adam Dray]

The other mechanic that may impact the type of metagaming you describe is my Life mechanic. In Verge, Life represents ability and will to go on. When you lose your last point of Life, you retire your character and narrate his or her end in your next Development scene.

Whenever a conflict puts your character's Drive at stake, you lose Life when you lose. One's Drive could be Reputation, Heroin Addiction, Psychological Control over Others, Need for Speed, Family, Thirst for Knowledge, Hot Dates, and so on -- anything that can be interpreted as an "addiction" (The more of you get, the more of it you want.). Verge is a game about obsessions and unquenchable desire.

I think the player who takes unnecessary risks for the chance of a big payoff will, more likely than not, burn out fast (lose all their Life).

[Lxndr]

Fastlane puts much of the risk into the hands of the player, by taking advantage of the Roulette game. Players control their bids, and Roulette has a flat risk/reward odds system going up the line, so it's entirely up to the player how much risk is being taken, with an entirely predicatble odds ratio.

On the other hand, the GM also has chips, a limited supply that he takes from to set the difficulties (and which is only replenished from chips a player bids).  So the player isn't the only controller of risk, the player just decides how much risk each of his chips will be taking.  The GM decides how many chips he'll need at the end.

Oddly enough, in Fastlane there is also a Life mechanic, which represents the ability and will to go on. When you lose your last point of Life, you have one last scene, after which you retire your character (this is calling Burning Out). Life is connected to particular things that you like - which pretty much include stuff like your list.  Family is Life, Drugs are Life, Speed is Life, etc.  Whenever a conflict puts a Life on the line, you lose a point of Life when you lose.  On the other hand, when it is threatened, you can get strength out of it.

Anyway, perhaps we should compare notes.  It sounds like Verge is taking a similar attitude to what Fastlane took (though obviously, you're not using a roulette mechanic, and I've got a limited pile of chips, instead of a limitless pile of dice).

So, er, yeah.

[Adam Dray]

Hi, Alexander.

In fact, I discovered Fastlane a few weeks into my own design and went through a bit of a personal crisis because of the similarities (control of risk and Life are two). I posted a fair bit about it in my LiveJournal. *sheepish grin*

I ended up buying your game to do an analysis of the similarities and differences (and for ideas, of course). I'm a big fan of Fastlane.

Check out Verge and tell me what you think. They're similar but I think Verge is different in a number of ways.

[Jasper]

Adam,

In a game I wrote a while ago, Loqi, I wrestled with this same basic idea. In Loqi, the player and GM both come up with an outcome to the conflict. By default, there's a 50/50 chance of winning/losing. But the more beneficial the player's outcome is, the more the odds get pushed to failure; and the same thing applies to the GM's outcome, so the more harmful it is, the more likely the player is to win.

If both player and GM propose wimply outcomes, it's 50/50. If both propose reallly huge, mega-damaging or mega-helpful outcomes, it's also 50/50. But when there's an imbalance in outcome, there's a reciprocal imbalance in the odds. They usually end up being in the same ball-park, with some degree of fluctuation.  

It's cool because if a player's character, frex, is fighting some fantasy monsters, he can say to himself "Do I want to be absolutely sure of killing a baddy or do I take a risk and slay tons of them -- with the possibility of losing and coming to harm myself?" In other words, the player can choose to play it safe, or be adventerous.  Some zany outcomes can get proposed, but it's okay, cause they're always balanced by how likely they'll be.

Hope that helps.  The key in Loqi is that the players get to choose the details of their risk-taking by describing the whole outcome too.

PS. Loqi is free, and available here: http://primevalpress.com/games/loqi

[Jasper]

Sorry, that should have been: http://primevalpress.com/games/loqi/index.html (no, it shouldn't matter, but it does)

[Callan S.]

Heya Adam,

What stops the players from assigning just enough dice to get through, with a pass and some level of reward?

Currently it would seem to reward turtling...just getting by with a slow income of rewards. Sure, I could ramp up the difficulty for a thrill, but how is that smart gaming, rather than an occasional exciting gamble? As soon as they get stung for that, players are penalised back into turtling.

[Adam Dray]

The question is, what is "enough dice just to get through?" There's still a roll involved. There's still an element of chance.

Those rolls where you just squeak by -- those are the exciting ones. You have 5 successes on the table versus the GM's 3 on his 8 dice. You give the GM 4 more dice after the initial roll. He lets the first die drop. A match; crap. He drops the second. No match -- whew. The third clatters to the table. No match. The last one could tie things up. It falls to the table, bounces off your Coke can and comes to a rest near your character sheet. No match! You've squeaked by this time, with a total of 12 dice towards your experience. Sweet.

When the player gives the GM too many dice or too few dice, that's the only time that the result is known (or at least probably known). It's when the player is trying to sneak by that things get exciting.

[Mike Holmes]

You're missing his point, Adam. He's saying that a likely result of this would be to have the player just give one or very few dice to the GM each time. Yes it's more dramatically interesting as you point out to "sneak" by your definition. But the reward mechanism will basically promote a certain behavior depending on the schedule that it gives out the rewards. If the player learns that by "sneaking" that he progresses more slowly than if he goes at a lower rate, then he'll go at a lower rate.

This is exactly the point that I was making earlier. It comes down to the "expected value" of a gamble. Basically there's a curve in the EV that peaks somewhere. For your game, it's hard to calculate. But it's true that if you have the same number of dice as your opponent that your chance to win is 50%. Well, let's say that you have 10 dice rolling. So selecting a Volume of 10 means that you'll only win half the time. The expected value of ten dice in terms of the experience die roll outcome is something like 3.5. So the overall expected value of this situation is 1.75.

Rolling only 5 dice gives only a reward of about 2.1, but you win the roll about 90% of the time. Meaning that the expected value overall is about 1.9. Better than 1.75. Note that none of this takes into account all of your various mechanisms for adjusting die rolls. They confound the stats a lot. But, unless players can use them for Experience rolls (and would) I think that the overall conclusion is true.

The player will learn this over the long haul, even if he doesn't have the figures in front of him. That he advances faster over time by taking less risk. Especially since victory is generall more fun than failure. You're double incentivizing rolling less dice.

Now, thinking about it, it is probably possible to find a way to make the expected value higher as you go higher. Such that the trade-off is less chance for success, for an actual output that's higher in terms of experience. One way is what I suggested. If you allow them to get experience when they fail, then you really incentivize pushing the dice up. Again, perhaps too much, however. But there might be a way to make the rewards trail off as you get beyond that dramatic 50% level.

That would seem to be the optimal curve to shoot for. Increasing but diminishing returns as you increase the Volume.

Mike

[Adam Dray]

After a couple readings, I understand what you mean. I'm trying to think through how the Drive buffer works with this.

Essentially, at low levels of Drive, the numbers you gave hold. Drive is not much of a factor against EV.

As Drive increases, it starts to diminish Experience rewards. Your Volume 4 minus Drive 3 nets you 1 die to earn Experience (guaranteed 1 Experience). Volume 8 minus Drive 3 nets you 5 dice for Experience (EV 2.44) but you win less often.

I don't have the statistical knowledge or tools at hand to calculate this out (I can compute EV from my probability tables via sum of products, but I don't know how the EV of N dice vs. the EV of Volume V turns into a probability of winning).

It looks like I have some work to do! Thanks for the insights!

[Mike Holmes]

(I can compute EV from my probability tables via sum of products, but I don't know how the EV of N dice vs. the EV of Volume V turns into a probability of winning).

You can't. You can say that the higher has a greater than 50% chance of success, but you already knew that with the higher number of dice.

And, frankly, I don't have the mental wattage today to figure out the formula (have you seen the formula threads, however)?

As for drive, It'll have two effects. It decreases the EV, of course, which pushes the curve up. But it also increases the chances of success all along that curve. Since the largest increases are found nearest the 50% mark (one die superiority is the largest marginal increase in EV that you can get - the next die gets you less of an increase), that means that the overall EVs on the low side will be less impacted than those on the high side. Basically the curve will "bulge" towards the higher volume.

This is probably good news, but not a lot of it. Given that characters can and will be at zero Drive at times, they might experience a "wallowing" effect here, where they're incentivized to go slow until they get some drive back.

OTOH, this does seem to be the overall design goal of the mechanic. What you might want to do, however, is simply to make something like a 3 Drive the minumum.

Again, a lot of this is horribly complicated by all of the other mechanics that modify these curves all over the place in terms of success rates. I sense that, in fact, they tend overall to make one more successful. As such that could throw this analysis all off. It also means that, fortunately, a table of basic odds does the player little good. So there should be no actual probablility analysis going on. What'll happen instead is that you'll get analysis by trial and error. Which is a good thing, IMO.

So what I'm saying at this point is that you need to test the system. Do some dry runs just rolling with sample characters, and try it on a test subject and see what happens.

Mike

[Adam Dray]

Thanks for your analysis, Mike.

I've done some simple test runs myself, but not with actual players yet. It's definitely time to get some real people involved.

In the car on the way home from work, I considered an alternative experience method.

Opposing dice. You roll (Volume) dice and subtract the result of (pool size) dice for Experience. If you rolled 10 dice and the Volume was 6, you'd roll 6 dice for Experience, then roll 10 more and subtract the result. Maybe you roll even if you fail but you can go down in Experience (bad habits), win or lose. For example, you roll 10 Volume dice and get 3 successes. You roll 6 pool dice and get 4 successes. 3 minus 4 means a loss of 1 Experience. Certainly, you'll want to maximize the Volume to avoid losing Experience.

Losing is usually its own penalty (in many cases, Stakes include loss of Life, Gear, or other traits). You're right that I don't need an additional incentive not to lose.

Drive never goes down, though. There's no mechanic that reduces it.

[Callan S.]

Thinking on this, what you might want to try is that the higher the difficulty is above the stat, the more every other player gets some reward (or the higher a chance all of them get a reward), not just that player.

This way you apply social pressure, as everyone wants an exciting game and wants a good reward, so they will bug and pressure the player to take on more than he otherwise would. He doesn't have to, but he will be more inclined.

Which is sort of what a GM used to do (or was supposed to), but really this is more how it works, but now with mechanics.

Still, there's something I don't like about my pushing the meta game aspect of 'lets make an exciting game' here, but I don't have time today today to get into it.