Sources of Fortune: In defense of the unopposed roll

[JMendes]

Yello, :)

I would like to add some points in favor of the opposed/unopposed dichotomy (sp?):

Myth #1: Opposed die rolls have no more information than unopposed ones
Take an average able-bodied dude out of the general population and have him run the 100m dash 30 times in 30 different occasions. Check out the distribution on the times. Now, take a world class runner and do the same. Now, put the two together in a two-dude relay. Do you really expect to have a similar distribution? My point: whether in opposition or in conjunction, multiple sources of randomness should have distributions differing from that of a single source.

Myth #2: Opposed rolls should be less predictable than unopposed ones
Take two dudes and pit them against each other in the aforementioned 100m dash. Possible outcomes: win, loose or draw. Actually nowadays, it's win or loose, as photo-finish usually takes care of the draw. Now, take just one of them. What are the posisble results? Any one from 9 secs to unfinished. However, let's say we are interested in the actual times obtained by both dudes. We now have a two-dimensional distribution, which is inherently less predictable than a one-dimensional one. My point: if all you care about is who wins, opposed contests should be more predictable.

Myth #3: If you must have opposed rolls, everything should be abstracted to an opposed roll
Let's take that 100m dash again. Can we call the track the opposer and roll for it? Sure. We can. If we rate simplicity several orders of magnitude higher than realism. The truth is, the track is just there. It is not a source of randomness. My point: other things being equal, the number of rolls should exactly equal the number of active participants, i.e., the number of sources of fortune.

Of course, the above is not an end-all-be-all of... well, of anything. I have nothing against people who pursue consistency for consistency's sake (and in fact I myself am undecided on that point). However, I felt that the opposed/unopposed distinction was getting the short end of the stick and I decided to speak up. :)

Comments, however thermal, welcome, as usual. :)

Cheers,

J.

[Jonathan Walton]

This has me wondering why virtually all opposed rolls in RPGs are represented as dialectical.  Surely there are cases when more than one thing is acting on another (as when things are interacting on each other or more than two things are involved), yet those kind of situations are most often solved with complexities of modifiers of various types, all of which are static in nature.

Some games, I imagine (none come to mind at the moment), probably use modifying dice to represent how much a factor influences the event, but this is still basically a modifier to the main roll and not a roll for the inflence itself.

Shouldn't there be a way to roll for a three or four-way conflict all at once, without having to break everything down into dualistic battles?  Can't we roll for the runner, the track, the weather conditions, the will of the gods, and the runner's old injuries (which might act up), all in a single system of resolution?  Couldn't the weather be the overall winner of such a conflict?  Wouldn't that be an interesting outcome to narrate?

Just some thoughts.

Later.
Jonathan

[Le Joueur]

May I add...

Myth #4: Role-playing games are best served by modelling reality at every turn.

The reasons should be obvious...

Dang Langford

p. s. And if the difference between reality and a game isn't obvious, send me a Private Message and I'd be happy to explain with links to relevant discussions.

[JMendes]

Hello, :)

Myth #4: Role-playing games are best served by modelling reality at every turn.

The reasons should be obvious...

Hmmm... Ok, I'll bite. Please endeavour to explain myth #4 to a simulationist without insulting him.

Myth #4: Role-playing games are best served by modelling reality <...>
p. s. And if the difference between reality and a game isn't obvious

Likewise, if the difference between reality and modelling reality is not obvious... well, you get the point. :)

Cheers,

J.

[Le Joueur]

Hey J.,

Myth #4: Role-playing games are best served by modeling reality at every turn.

The reasons should be obvious...

Hmmm... Ok, I'll bite. Please endeavor to explain myth #4 to a Simulationist without insulting him.

That may be hard.  First of all, what kind of Simulationist are you?  Second, how thin-skinned are you?

The biggest cannon brought to bear on this situation is the difference between realism and verisimilitude.  Are you going to have magic?  That's not the least bit realistic.  Superpowers?  Psionics?  Races other than human?  Same, same, and same.  It is one thing to give a game mechanic that 'feel' of internal consistency oft referred to as verisimilitude, it's wholly another to think an abstract mechanic can faithfully model reality.

And on the point of simulation, Walt made a really good point about probability distributions, not just realistic ones, in a thread about how linear die modifiers are broken.  There, as here, the point needs to be made about how much of this the people playing the game will even notice.  This is not so much a point of 'is it realistic,' but 'how realistic is realistic enough.'  Take your three myths as an example; how many people will even notice this in a system that isn't broken (has glaring errs that stop play while someone figures out how to 'fake it')?

And then onto Simulationism.  What kind of Simulationism are we talking here?  Simulationism (despite its name) has little to do with actually simulating anything.  Simulationism (as listed in the GNS) is about 'Exploration' of the imaginative commitment in a relatively causally-consistent realm as a priority over everything else (like the goals of Narrativism or Gamism).  It's also primarily of use for diagnosing problem games.  Not only that, but it becomes complicated to speak of any Simulationism that isn't narrow to only a few of Ron's list of game 'parts' (Story, Situation, Color, Character, and System).

I mean, as Ron describes The Riddle of Steel, it has a great deal of Simulationism (Exploration of System) completely in service of driving the Narrativistic engine that is the Spiritual Attributes (if memory serves).  Is The Riddle of Steel realistic?  It was written by a practitioner of the arts clearly described within (not just a theorist or Society for Creative Anachronism member).  And yet Jake, himself, admits that certain liberties were taken for the sake of 'game.'  For example the late discussion of the real mobility issues involved in plate armor; you are not as impeded as the system suggests.  That was a concept introduced to make other armors more attractive in the game, in reality cost and other issues made other armors more attractive.

Right there is an example of a game leaving reality behind, because it's a game.  Sure you could put in extravagant mechanics to balance armors out fiscally, but that takes the game far away from the designer's intent.  For what reason?  To be realistic.

Now let's get back to your three myths.

Myth #1: Opposed die rolls have no more information than unopposed ones.[/list:u]That one is actually true.  

They're just dice, the numbers could mean anything.  You are inferring a certain system bias here that doesn't need to exist.  Now I'm not one to be able to describe all kinds of examples, but I think it clear that a system could be written where neither has "more information."  You may be talking about systems you have experience with, but that doesn't change what can be done.

Myth #2: Opposed rolls should be less predictable than unopposed ones.[/list:u]This depends largely on two things.  What you mean by predictable and how important it is to be able to predict the outcome of a game mechanic.  

Mike Holmes makes a really succinct statement about how opposed rolls tend to 'center out' more where unopposed are a bit more 'all over the map.'  Is that 'more predictable?'  Similarly, why are people playing the game estimating probabilities anyway?  Do you consider the probabilities of your times in a foot race?  What about when you use an ordinary skill?  In fact, in normal life, does one even think in percentages?  In predictability?  In repeatability?  Or do they 'just do it?'

Myth #3: If you must have opposed rolls, everything should be abstracted to an opposed roll.[/list:u]This gets at the heart of

Mike's Rant, and I think he has (and can) explain it much better than I.  I echo his commentary on sticking to one or the other.  The point demonstrated by your examples (though I am surprised to have to point it out) is that it will be quite hard to create a system that has both opposed and unopposed mechanics that appear the same.

It would be quite easy to stake out to completely different mechanics.  Say the opposed one is as you describe, "the number of rolls should exactly equal the number of active participants," and models that well.  The unopposed one would be built to create information how far, how fast, how well, and so on.  The problem comes when you put both into one game.

There's consistency and then there's consistency for its own sake.  Have two separate resolution mechanics and people will ask why.  Try to make one look like a mere feature of the other and you sacrifice that realism or verisimilitude or Simulationism or whatever else was at stake here.  (What was the problem, anyway?)  That clash is the problem that makes #3 not so much a myth as bias on your part.

But then there's the whole problem with bias.  Bias is the definition of opinion, these myths may be your opinion, and you have every right to these preferences.  But when you pose them as myths, you tickle the Devil's Advocate in me.  Contrasting your list isn't so much as a condemnation of your opinion, but condemning it as fact.

This posting is, in fact, my opinion.

Fang Langford

[Christoffer Lernö]

Maybe it's just me, but I think that at the heart of the matter lies not how you roll the dice, but how random it is. The fact is that most systems have one or two models of randomness and then violate almost any situation by imposing that model on everything.

Bell curves are very nice and well, but I rarely see anyone thinking about the effect of the standard deviation of the distribution.

If skills range from 1 to 10 and there is a 10% chance I'll either perform at +5 or more or -5 or less of my average level of success, then that's gonna be a whole lot different in feel from if the numbers would have been +2 and -2. The opposed/non-opposed thing is only really about introducing slightly different distributions, but in general those changes aren't very drastical.

Because we're all friggin weaned on having randomness being as important or even more important than skills and the like. The most simulationist games with millions of modifiers can suddenly in a massacre of irrationalism introduce a random factor which outweighs the impact of any of the modifiers mentioned.

In most rpg worlds it's surprising that people even survive at the rate they are supposed to fumble or make mistakes.

Why does armwrestling yield the same distribution as playing cards? Why is the time it takes for me to cook eggs for breakfast and the time it takes  to debug a program equally uncertain in most games?

I don't care what damn mechanic you use, just give me a world I can depend upon, not one that would break down in minutes if it followed the same laws that apply to the characters.

In real life, some things are more predictable than others. Is really having that in a role playing game so much to ask?

(Incidentally you can use the same randomness scale if you make sure that the effect of low and high results depend on the randomness of the situation. So a bad result on cooking eggs might add a minute, whereas a bad result for debugging might add a few weeks)

[Andrew Martin]

In fact, in normal life, does one even think in percentages?  In predictability?  In repeatability?  Or do they 'just do it?'

In real life, I've come across a number of people who are definitely not gamers, who measure their health status by saying they're 100% or even a lesser figure. I've also found that for skills like typing and computer programming and a number of other tasks (eg production lines) it's possible to get a percentage for success and failure, based on what works and what doesn't; this can directly translate to skill percentage in a game system.

Certainly for intelligent/sentient machines or androids (and logical aliens), there's lots of examples in the media of their stating chances and percentages, which usually become 100%. :)

And for those tasks where people "just do it", they usually have near 100% ability anyway. :)

[Andrew Martin]

In most rpg worlds it's surprising that people even survive at the rate they are supposed to fumble or make mistakes.

When playing/GM-ing Avalon Hill's Griffin Island for RuneQuest III, I came across a similar problem. A formation of several hundred archers were firing at the PCs. (These were very tough Dwarf PCs with lots of magical enchancements). Every combat round, around 5 per 100 archers broke their weapons, or injured/killed a comrade or themselves. The PCs strategy was to wait until the archers broke all their weapons then attack! The PCs were dwarves and this was quite a logical method according the rules of the game system. I let it work 'cause I didn't know any better and this was following the rules of the game. I grew greatly annoyed with this problem.

[Andrew Martin]

Shouldn't there be a way to roll for a three or four-way conflict all at once, without having to break everything down into dualistic battles?  Can't we roll for the runner, the track, the weather conditions, the will of the gods, and the runner's old injuries (which might act up), all in a single system of resolution?  Couldn't the weather be the overall winner of such a conflict?  Wouldn't that be an interesting outcome to narrate?

It's easily possible with a simple, powerful system. I've got at least two systems on my site which can do this. Accord does this all in one die roll if participants are working on the same task, or multiple dice rolls if parallel tasks. S and Swift do this with one die roll per participant.

[Christoffer Lernö]

Every combat round, around 5 per 100 archers broke their weapons, or injured/killed a comrade or themselves.

There is a notion among many players and GMs that "you should always have a chance" (to succeed or fail)

The usual motivation you hear for this is that: "It was so incredibly cool when my character Bob the one-legged Dwarf, running away from an Ogre Horde managed to jump a 11-meter cliff because I rolled 5 open-ended rolls in a row!"

I guess unlike them I don't see this as a good motivation for a game allowing  ludicrously unlikely things.

I see a game that is so flawed so that the players turn to random luck to provide any feeling of protagonism to the characters. That the same die also made Vokk the Highlander Barbarian, bearer of the Runes of Khator the great and destined to lead the people of Kor to freedom from the oppressive Southern Dawntribes, slip as he tried a jump of measly 3 meters and fall to his doom three a thousand feet down as he tried to hunt down a pickpocket who had stolen his favorite dagger.

Basically I'd like to see games: stop relying on randomness for drama and protagonism. Make things really random only when it makes sense that they should be. Like if they are gambling or whatever.

[Le Joueur]

In fact, in normal life, does one even think in percentages?  In predictability?  In repeatability?  Or do they 'just do it?'

In real life, I've come across a number of people who are definitely not gamers, who measure their health status by saying they're 100% or even a lesser figure.

The point was that percentages don't come up (except in...well, exceptional circumstances) when preparing or deciding to do something.  As in, "I've got a 50% chance of finishing this before lunch time; let's do it!"  Rating a "status" with a percentage is not related to the original article about opposed rolls, I think.

I've also found that for skills like typing and computer programming and a number of other tasks (eg production lines) it's possible to get a percentage for success and failure, based on what works and what doesn't; this can directly translate to skill percentage in a game system.

Virtually all of these examples are calculated after the fact supporting my contention that few, if any, 'percentagize' a situation before making a decision.  (I own that people often weigh their chances, it's just the absurd idea that they might with the precision of determining a role-playing game's resolution probability, that I have a problem with.)

Certainly for intelligent/sentient machines or androids (and logical aliens), there's lots of examples in the media of their stating chances and percentages, which usually become 100%. :)

That's fiction.

And for those tasks where people "just do it", they usually have near 100% ability anyway. :)

Actually, in informal survey, this occurs because they consider the 'time coefficient' as flexible.  I'm going to make this 'thing,' no matter how long it takes; that's not 100%, that's 'repeat until successful.'  When people don't have time, they 'take their chances' but don't estimate them.  I believe.

Fang Langford

p. s. Still trying to establish whether the exceptions prove the myth that people estimate their probabilities before acting the way gamers seem to require of die resolution systems.

[Ron Edwards]

Hello,

This thread is wandering badly, I think. Let's please focus on the stated issue in the first post.

I was afraid that people would mis-read Mike's rant about opposed vs. unopposed rolls, and now I think it did happen.

The point is not about whether one is better than the other. The point is about combining the two methods in one game system. The classic example is to say, "When climbing a cliff, roll against your skill target number," and to say, "When fighting, roll against your skill target number, your foe rolls against his, and then compare the degrees of success."

Whether you call it "active vs. passive," "opposed vs. unopposed," or any of the other terms found in game texts, it's always the same thing.

Here's the problem with it. What happens is that the skill target numbers and the fighting/conflict target numbers are operating, in terms of success, on two totally different probability scales. That's in and of itself inelegant, at the very least ... and then, the real trouble shows up when the "pips" of the target numbers are "bought" using points of equal value, which is the most widespread method of character creation in games that do this in the first place. This leads to breakpoint problems.

The actual recommendation is this: use only target numbers, with what's called unopposed rolls, for everything (The Whispering Vault); or use only opposed rolls, without target numbers, for everything (Sorcerer).

[I am talking about developing a single resolution system, not about having two systems for non-overlapping elements of play - in the latter case, then having one be opposed and one be unopposed is cool. It's the combination, in-play or across-points, that causes trouble.]

I hope this clears up the issue slightly. There's no point in discussing the differences between opposed vs. unopposed in terms of unique virtues.

Best,
Ron

[damion]

Fang,
       I think people DO estimate their chances before doing something, if they mattter.  It's just that as you and Chris pointed out, most tasks in reality will almost always 'succede', the distribution is one of quality of the output, rather than a probabilty densify function of success or  failure. Quality can refer to how long the task takes, the quality of the output, or both. Very simple tasks only vary in time(running a mile), while cooking varies in both quality and time, and our old friend, jumping the chasm varies only in quality(how far you jump-which the chasm makes a binary value). What failure chances there are come from the task being impossible(no way I can do this without an allen wrench). Most systems conflate the impossiblity of the task with the quality of the trial.
       Back to the original point though, I think people DO estimate their chances in a heuristic manner, if  failure is an option.  A crook may estimate their chances of commiting a crime(we figured we could get away with it). If given a task, a worker will pick an amount of time for that they think they have a good chance of being done in (assuming the penalties for not being done in that time are fairly bad).  If I throw something away I estimate my chances of getting in the wastebasket before deciding weather to get up or not, ect.
In conclusion: I think people DO estimate their chances, but there are two other factors : 1)most tests aren't pass/fail in RL, usually it's the output quality measured against something to determinet success/failure.  2)The probablities aren't as obvious  as they are in a RPG.

There's no point in discussing the differences between opposed vs. unopposed in terms of unique virtues.

Well, there's a slight theoretical advantage to unopposed in terms of handling time, as it's easier to parallelize. In my experiance it doesn't playout in practice though.

[Mike Holmes]

Thanks Ron.

Interestingly, I say nothing in my rant (to which I refer people, and to which I, perhaps arrogantly, assume this post is aimed) that contradicts anything that the original poster has said. I have my own opinions on those subjects brought up as myth, and we could debate those separately. My objection is that people put separate mechanics for these situations just because other systems do. Without any consideration of potential effects.

The one argument that I hear that contradicts this, potentially, is that separating these mechanics "feels" right. This is essentially what Mr. Mendes must be saying when he says that:

Can we call the track the opposer and roll for it? Sure. We can. If we rate simplicity several orders of magnitude higher than realism. The truth is, the track is just there. It is not a source of randomness.

This would be a valid reason to put such mechanics in if this were true.

But this is just an opinion not supported by facts. The track may feel non-random, but I assure you that it is in fact completely random. I can calculate a probability that the track will, in fact suddenly jump up ten feet in the air in the middle of the race, sending the runners all flying (1/i^e, I can explain in private mail if anyone want's to know what this means). See Hiesenberg. By this rule, you'd have to roll a near infinite number of dice as everything in the universe actually affects everything else randomly through gravity.

What he must be saying is that one should only roll for those things that "seem" random, or are random enough that they would have a discernable effect. Such that players will feel that the dice being rolled are having an effect on the game that somehow feels like how daily reality affects them.

I can buy this argument to an extent. My point regarding this is that players should ignore the mechanics, and just focus on the outcome. The mechanics are just a model in most cases (see TROS for a counter example of where the dice really go beyond simple modeling). If the players trust that the results will be realistic (and they will using a well designed opposed system), then such mechanics feel fine.

At least that's the testimony of many, many players who have played with such systems. That is, nobody mourns the loss of the other system in play that I've seen. I've never heard of a player playing Sorcerer say, "You know, it seems weird that you're rolling dice for the lock". In fact, they don't think that this is happening at all. They understand intuitively that the dice that the GM is rolling are just part of the randomization of the situation (I refer the reader to the rant wherin the reality of the statistics being the same is explained). And since Sorcerer is well designed to produce results that are "realistic" insofar as the results produced seem suitable for the sort of action that everyone expects from it, the system does exactly what it needs to in all situations. Players are satisfied by it.

Can anyone who has actually played Sorcerer, or any other game with a single resolution system (does not have the opposed/unopposed), actually say they felt uncomfortable with how resolution works? When the GM "rolls for the lock" (really rolling for the difficulty, different)? I doubt it, though there may be some.

In any case, if you are designing a game and your specific goal is to appeal to people who you feel would prefer this sort of dichotomy, and prefer it despite the greater complexity, then great. You have given it at least due consideration. Which is all I ask for. In fact, even better reasons have been given why multiple resolution systems have been included in some games. And to that extent, I support such decisions.

But I will still submit that the majority of times where such mechanics are used, they are employed because the person doing so is unaware that there are functional ways to make both situations work using only one mechanic. As such I feel that it's my duty to inform people of the facts of the situation, and will continue to do so. So that designers can make an informed decision. I've seen nothing here that would change my mind on that point.

Mike

[Le Joueur]

Hey Damion,

In conclusion: I think people DO estimate their chances, but there are two other factors : 1)most tests aren't pass/fail in RL, usually it's the output quality measured against something to determine success/failure.  2)The probabilities aren't as obvious as they are in a RPG.

I wasn't arguing specifically whether people estimate (and we're using that term rather loosely) their chances.  The point was how different and potentially unrelated this was to the phenomena of role-playing game resolution systems.

You look at a situation, you make some intuitive judgments, you decide; what I'm talking about is the stress some people put on being able to calculate exact percentages of chance in role-playing game resolution.  I suggest that the correlation between these two is actually defensive rationalization on the part of the calculators; I don't see a direct relationship at all.

The corollary would be looking at your skill rating, looking at the potential difficulty, and guessing not calculating.  People do estimate; they only vanishingly rarely calculate.  I was saying that to make that calculation easier in role-playing game design has little relativity to 'what people really do' and is mostly a style aesthetic on the part of the designer (regardless of their rationalizations).

In other words, there isn't any 'reality modeling' reason to make resolutions more calculable.  This bears back on Myth #2 and the 'predictability' of either opposed rolls or unopposed rolls.  What value is 'predictability?'  If it has to do with these 'estimates' you mention, then it doesn't need to be simple to rigidly calculate out; reality isn't.  On the other hand, why 'should' either opposed rolls or unopposed rolls be more predictable?  And to whom?  The implication is that opposed rolls are not more predictable; yet Mike has demonstrated, along real world estimables, opposed rolls provide more 'stable' results.  Meaning they're more predictable, just not terribly calculable.

All of that describes a value judgment made about predictability and calculability.  Such is clearly a subjective issue and therefore there is no myth; the statement is neither true or false, it is only an opinion.  (If you look closely at myth #2, there's a big should in the middle; that's a warning that it can only be an opinion.)  I'm personally not too clear on whether this thread supports unopposed only, opposed only, or mixed only, resolution systems; I just wanted to point out that these myths are not only a matter of opinion, but also skewed on the idea that 'realism is best.'

And that isn't realistic game design.

Fang Langford