Observable Probability

[Marco]

A few thoughts on the topic:

1. I generally favor the player having a good idea of their capabilities and leave the uncertanty up to resistance/unknown factors (I see a man simply dressed with a sword--I know I'm good, how good is he?)

Although *I* don't know how far I can jump exactly, an olympic high jumper knows to the inch. My mechanic is really good at estimating how long it will take to fix a known problem, etc.  I.e. your odds of knowing how well you do something improve with your ability in it.

2. Dice Pools just make it harder for some players to make the odds calculation than other players.  And if a player demands time to figure out what the character is gonna do, it's hard to deny them that.

(I was told by a player--a pretty dedicated player: "I don't know what the hell to do yet--but my character's been in scrapes before and he already knows exactly what to do." It was true for the character--and I found it hard to argue with that logic. [argument about GM moving game along ignored for purposes of this example])

3. I think it's more important to have a strong success curve than to know exact probabilities. If I'm really good at climbing, an 80% on a D100 isn't good enough (1 x in 5 the character will fall if the rules say "make a climbing roll or fall ... if the rules don't truncate the fall based on how much/badly it was missed by, it might be a fall and die). On the other hand, 80% is a pretty good number to bet on in a cassino so it's a matter of perspective.

-Marco

[Bankuei]

Mike, I like the idea of not having an exact take on probability.  From a Sim point of view, most of us are very aware of only a few of our abilities, and only at using them in a routine fashion.  

Yes, an olympic jumper knows exactly how far they can jump, because they do it all day.  They know only after having had the experience and recorded the results, not before they've trained.  Second, they only know how well they jump in a gym, with the proper equipment, not say, jumping over a brick wall in the rain.  All of us know about how well we drive, not too many know how well we drive in car chases.

From a gamist point of view, part of the fun is trying to measure the odds but never knowing exactly how it's going to turn out.  For example, in Poker, I know what kind of hand I have, but I have no clue as to what the other folks have.  I have to watch them and see what they bet and use my knowledge of experience as to their habits to make a guess.  Gamism gains a great deal from doing the best you can with limited resources, including limited knowledge.

One of my biggest pet peeves with games is that there is very few games that call for a judgement check before you do something.  For example, if I'm about to climb a cliff face, I'd take a minute to check out how tough its going to be, and what the best route would be.  If I had bad judgement, I may overestimate or underestimate the challenge by a lot.  With good judgement, I'd know exactly how tough it would be.

As far as actually knowing the probabilities, I think the designer should know as much as possible, but if you've designed a good game, the guidelines you give the GM and players should be sufficient, even if they can't work out the actual statistics.

Also, I read this philosophy about a videogame that I think may apply"....Instead of the character getting better, we wanted the player to get better.  As you play, you see your skill grow, and you have a feeling of accomplishment.  You don't have to rely on levels or power ups to be good."  

Great question Mike, I'd be very interested to see how you'd handle a game with fuzzy odds.  

Chris

[Matt Wilson]

I think knowing the odds has a lot to do with the price of failure, based on the style of game in play. If failure is Old Skool death and dismemberment, then I'm a little more wary. If failure is Sorcerer and Sword-style complications, then who cares? Either result is as fun as hell.

[M. J. Young]

[Hopefully Relevant Story]

In college, I was the tech head for the college drama group. They did their plays in the chapel (it was a very small school--this was the first one I attended). Before the play, maintenance hung a bar from the ceiling, and gave me a tall stepladder. Up I went, and began running my wires and lights where I wanted them. Then I made a wrong move, the ladder fell, and I was left hanging from the bar.

Well, others who happened to be there started to panic, running around trying to figure out what to do. Move the ladder, I told them. I can drop from here, but I don't want to land on the ladder. Once they did that, it was simple to fall the distance.

[/Hopefully Relevant Story]

Now, I could not tell you then or now what that distance was; I can tell you that hanging from that bar, I had a lot of information about how far it appeared to be, how hard the carpeted floor was, how far I had fallen in the past. Near twenty years of life experience went into that assessment, and I concluded that this was a drop I could make.

If I'm playing a character, I lose all of that. I have no connection to his experience, and don't know what he's successfully done before or how well he can land. I can't see the distance, really; I can only know the numbers. I don't know how hard the ground is unless I've already asked that, and even then, my estimate of the impact is based on an extremely subjective concept of hard at that point.

I replace that with game numbers. I build an idea of what my character can and cannot do from the mechanics of the game.

It would be more realistic if I could play the character 24/7 and learn who he is, watch him grow up, experience falling out of trees and leaping off rooftops and all the things I did as a kid that taught me what I could manage and how to do it. But I can't. Even as a dedicated role player, I would always have this distance, this failure of connection to the character, that I can't completely cross. In a pick-up game, that's amplified abundantly.

That said, in Multiverser we make the point that the character never knows his exact chance of success. The system is simple enough--percentile based for skill checks, which are a major resolution mechanic in the game. The referee knows (if he cares to check--the point is made that you can speed play by approximating, and only getting an exact number if the roll is "close" to what you expect), and knows in some detail how his modifiers impact the chance of success and the relative outcome; but there are, as someone observed, always factors unknown to the player. Even in the case of that fall, where the player knows the distance and the nature of the landing surface, there are factors he can only estimate (World Bias is always involved and never precisely revealed). Thus even with a mechanic that easily provides an exact probability of success, the player doesn't have that.

I don't much care for dice pools as a player because they aren't particularly intuitive in terms of estimating my chance of success; in a sense, you need experience with them to begin to know to what degree you are likely to succeed or fail (just like I needed experience to know whether it was a reasonable idea to drop from that bar). As a referee, I like them less, because it's a lot tougher for me to get my head around how much change I'm making in the difficulty of something--too many dials. Is it easier on five dice to roll three eights or two nines? Do the odds change appreciably if it's on seven dice, and which way? No matter what I'm running, I'm always going to come to that moment when I need to have the player roll against something for which the rules are not clear, and I'm going to have to create a roll that "fits the system" and at the same time creates a likelihood of success that accords with "reasonable expectations".

Now, if the system models our reality with great precision and the setting is described in such detail that the players can "almost see it", players don't need to know probability at all; they can use their experience. But my need to know the probabilities always kicks into high gear when
1) I can't really see the situation well enough to assess it (how far is that fall, really?)
2) My character is doing something with which I have insufficient experience (how good is that armor in blocking my attacks?)
3) I'm still feeling my way through the quirks of the world and the system and don't know what really works and doesn't work here.

So I favor a system in which probabilities are clear, but not giving all the factors to the players.

--M. J. Young

[Jeremy Cole]

Marco & Bankuei,

Its all good and well for the player, but what about the GM?

The GM has a man with a sword, and the players knows he spent half his character creation skills on swordplay.  The dice sets are very complex, so have difficult to determine probabilities.  I know how good I am, how good is he?  Unfortunately, the GM couldn't tell either and he hacks you to pieces, or impales himself before you reach him.

The same mystery can be achieved with a simple probability system, and just not telling the players the probability.  Which now becomes my question, why have hard probabilities?  This makes it harder for the GM and adds nothing to the play experience of players (they can just be told he looks tough, not he has level 4 swordplay).

Jeremy

[Bankuei]

I'm not arguing for complex systems, nor against them.  The designer is responsible for giving the players and GM sufficient information to run their game correctly.  What I am saying is that some systems lend themselves to giving the players too much information from a Sim or Gamist view.  

You are quite correct that what the players don't know can only be learned through experience, but the other factor to bring in is that since we don't live in our character's body, as M.J. brought up,  we need some means of communicating information that our character would know or recognize.  

Going back to my example, each person would have a different level of judgement and experience in rock climbing, and would see the cliff face, and its difficulty differently.  In your example, different characters would think the swordsman is a ruthless murderer, a pushover with a skinny blade, or a left handed duelist who favors thrusting and quick strikes depending on their own understanding and awareness.  

In all cases, we come back to Mike's query: How do we adjucate character self assessment, judgement, discernment, or common sense?  If you want to compare what a difference judgement makes, look at the difference between Mr. Wolf from Pulp Fiction, and many characters out of a Guy Richie movie.  How much accurate information is available to the character?  To the player?  How does that affect in game decisions?  Is it a benefit or detriment to the style of play you're trying to create?

Not quite going as far as a myth, but a standard is that players know everything that their character is capable of doing.  One of the uses of questioning it, is that few people have done so, and said, "What can we do otherwise? What other doors open?"  

My only point is that none of us really know how far we can push ourselves until we do it.  Knowing the odds changes gameplay decisions a lot, and that information is freely handed out, often without questioning its effect on the style of game people are trying to produce.

Chris

[Jeremy Cole]

So nice I posted twice, sorry.

Jeremy

[Jeremy Cole]

The thread is questioning the assumption that a system should produce obvious probabilities, following the logic that it is not always good for players to know the exact chance of success.  I'm saying is no.  

Stating that players not knowing the exact probability is a good thing is not the same thing as hard to determine probabilities.  

A hard to determine probability makes it a bugger for the GM to set the right level challenge.  Further, any exact probabilities can be held back from the players, just don't tell them the exact tgt number.  If you have clear probabilities, and hold information back from players, you have the best of both worlds.

So, is it good for a system to have clear probabilities?  I can't see why not.

Jeremy

[Valamir]

Marco & Bankuei,

Its all good and well for the player, but what about the GM?

The GM has a man with a sword, and the players knows he spent half his character creation skills on swordplay.  The dice sets are very complex, so have difficult to determine probabilities.  I know how good I am, how good is he?  Unfortunately, the GM couldn't tell either and he hacks you to pieces, or impales himself before you reach him.
Jeremy

Well, I guess that all depends on whether you make a distinction between Pregame manipulation of events and during game manipulation of events.

With a clear probability curve (or D&D style Challenge Ratings) you can as the GM set a challenge to an appropriate level of difficulty for the party.  If you want them to win easily, you set it up so they should win easily.  If you want them to have a challenge you set it up so its a challenge.  If you want them to get their butt kicked...etc.

You seem to be saying (and I agree) that this is a good thing.

What I would suggest is that the following two things.

1) Not knowing the probabilities is not the same thing as having no idea of the magnitude.  I may not be able to calculate the precise odds as the GM, but I do know that More Dice and a Lower Target number is better than fewer and higher.  I can simply do a comparison and get a good guage of challenge levels without needing precise "He has a 4 in 6 chance of hitting) type of calculations.

2) The same kind of manipulation that you'd be doing before a game can be done during it as well.  Misjudge the enemy strength?  Have planned for reinforcements fail to show up, or have additional reinforcements rush in.

[Le Joueur]

...the other factor to bring in is that since we don't live in our character's body, as M.J. brought up,  we need some means of communicating information that our character would know or recognize.  

Going back to my example, each person would have a different level of judgment and experience in rock climbing, and would see the cliff face, and its difficulty differently.  In your example, different characters would think the swordsman is a ruthless murderer, a pushover with a skinny blade, or a left-handed duelist who favors thrusting and quick strikes depending on their own understanding and awareness.  

In all cases, we come back to Mike's query: How do we adjucate character self-assessment, judgment, discernment, or common sense?

I've always felt that a certain amount of this arises from the players' personal expectations for a game, not something a game designer can really manipulate.  I felt strongly that mechanisms that allow a player to override the results of a game's fortune mechanic were probably the best we to keep from disappointing them.

I tried to tie the two together and get these 'unspoken expectations' out onto the table with Scattershot's Genre Expectations Technique.

This brings up a whole 'nother idea.  Since many systems allow these kinds of overrides, they are permanently impossible to perfectly predict.  But those same overrides keep the game from failing the players' expectations for the fortune mechanic.  So the question could become, "which is better, calculable results or meeting participant expectations?"  I believe the above conversation has quietly implied that these are manifestly the same; id est, the player will be disappointed (their expectations thwarted) by incalculable results.  I think there may be more than one way for 'probability' (which seems to be spoken of interchangeably with 'performance expectations') to be 'observable' (which seems to be spoken of interchangeably with 'dependable'); player meta-game affecting mechanics have been overlooked.

Or I might be off my bean.

Fang Langford