All y'all motherfuggers better listen up!

[benlehman]

It has come to my attention that most people in RPG theory have little or no knowledge of probability, and thus tend to get into long arguments about dice vs. dicelessness, with Erick Wujcik on one side saying that any randomizer means that the RPG is shit, and dicelessness-with-hidden information is the way to go, and Ron Edwards on the other side saying that role-playing games without chance cannot properly be called role-playing games at all.



Both hidden-information games and random games are the same, probabilistically speaking.

Let's pretend that we're playing a game -- I roll a six sided dice behind my palm, and you try to guess the number it sits on. (this is a boring game, yeah, but it illustrates a point.)

Before you guess, you can associate a probability with any face being up (this probability will be 1-in-6). The point is, even though I've rolled the number and have seen it, it is still random *to you*

Let's play a different game: I set a six-sided die to a particular value, and you guess it without looking.

Before you guess, you can associate a probability with any face being up (this probability may not be the same for every face.) In other words, despite the fact that no die was rolled (I made a decision about the die), the hidden information means that it is still random *to you*

Philosophically, you can argue that there are two different things going on here, but mathematically they are identical.

So, for one, when you play Amber, you are using random numbers all the god-damn time. So stuff it.

So, for two, there is no tangible difference between a diceless-but-hidden-info game and the roll-a-die game. So claiming that they are fundamentally different at a mathematical level is wrong wrong wrong.

In terms of the ephemera and toy quality, of course, they are very different. They *feel* very different. But they really *aren't* very different.

And I hope that shuts you fuckers up.

(P.S. As far as I know, there are no well-played diceless RPG systems that do not include randomness in the form of hidden information, possibly outside GM fiat. Cradle could do it with a few nips and tucks and, I think, still be a fun RPG. So I even disagree with Ron at that level.)

Comments

[apollinax.livejournal.com]

We had this debate last night, neh? I agree, that many of the people on the theory board don't grok probability. But, to be fair, it's really not simple stuff. Complex enough that I'll say you're WRONG. But I think we actually agreed on this last night. I also think, though, it should be made clear.

From the perspective of a single event, you are correct: they are indistinguishable, in that their probability distributions are identical. However, you're making a big assumption, that there is no prior knowledge and no knowledge contained in the event. Let's take D20 as an example, as it is fairly simple probabilistically (flat distribution).

Let's assume that you have no prior knowledge on the difficulty of the task, no indications from the GM such as "this looks hard." It is a complete black box. If the GM rolls hidden, then you gain one bit of information: success or failure. With enough such events (under controlled conditions, no changing modifiers, etc.) you can estimate your success probability to reasonable accuracy. This is a simple Bernoullli trial.

Take, in contrast, when you roll openly: you can gain a *lot* more information. If you roll a 10, for example, depending on success or failure you have defined the probability -- with certainty -- as greater than or less than fifty percent.

This is important when you make series of rolls; if the underlying mechanisms are changing so quickly that each action is functionally independent of the others, then sure. But this is rarely the case.

To be specific, much more important than raw probabilities is conditional probabilities: p(A) | B, etc.

[benlehman.livejournal.com]

Agreed in full. The amount of information revealed has a *huge* effect on the random quality.

The point is simply that a random quality is present even in apparently randomless situations.

yrs--
--Ben

[apollinax.livejournal.com]

Nonono.

Uncertainty is present in deterministic situations.

[benlehman.livejournal.com]

Uncertainty, fine.

As long as you admit that what you are talking about is mathematically indistinguishable from randomness, you can call it "Shirley" for all I care.