Mike's Standard Rant #5: The Myth of Opposed Rolls

When RPGs started resolution rolls were all of the "Roll vs. Target" type. Soon thereafter, however, somebody invented the idea of the "Opposed Roll". Where does this come from? As one moves from the early completely combat based resolution systems, and skills get introduced, you get situations occurring where characters are testing skills against each other in conflicts outside of combat. Well, with no target number, the obvious thing to do is to simply roll and see who gets the better result. Thus you get the dichotomy where skills used against things that are not rated via skills are rolled against a target number, and skills used against characters with similar skills are rolled using the "Contested Roll" method.

This all seems well and good until you consider that you've just created two entirely different resolutions systems for a single game. Now I can go on and on about the benefits of only having single systems for resolutions, but the advantages should be pretty obvious. The question becomes why do you need to have two systems? Bound by this particular tradition, most designers have to date continued to include dual systems for resolution. If you ask them how many resolution systems they have in the game, many will proudly claim one. But in fact, there are two similar systems. If you point this out the designer will say that it's just a modification of the single system for a certain type of circumstance, and an easy modification at that. The point still stands, however, that it's extra rules. Why is it needed?

The tradition hinges around that age-old dichotomy between active and passive opponents as mentioned above. Somehow, it seems intuitive that if you are playing chess against an opponent that both sides should get to roll. While if you are jumping a chasm, you out to only roll against a target. What is missing in the analysis is the fact these methods are, effectively, the same statistically. Neither method produces more information than the other, and neither can be claimed to have any more accuracy statistically a priori.

Let’s look at an example. In a particular system you roll d10 + Skill to get over a target number to succeed, the level of success being the number by which you succeeded. The corresponding opposed mechanic is to have both sides roll d10 + Skill and subtract the lower from the higher, leaving the higher roller with the difference in levels of success. What's the difference between the two methods? At first glance there might seem to be a difference but consider this. What if the "opposed" system were to have one person roll Own Skill + d10 - d10 target is the Opponents Skill. This is the exact same system as the regular opposed system, statistically. What this points out is that it does not matter who rolls the die or dice, effectively the systems are the same. In the end there are randomizers, and you get an output number. All rolls in all systems are opposed by their difficulty. There are no “unopposed” rolls. The dichotomy is a myth.

What is different in the opposed system is the number of dice being rolled. What effect does this have? Well, the opposed roll, consisting of two dice gets you the pyramid curve (flat bell). As opposed to the unopposed roll which gets you the flat curve. Which means that the opposed roll is much more likely to come out with an "average" result than an unopposed roll. Yep, that's right, actively opposed conflicts are more predictable, than working against a stationary object. Counterintuitive, no?

On the other hand, this is just a contrived example, and you probably could find a way around this particular problem. But the point is that there are problems with such dichotomies. Which is silly when they can be eliminated simply be ignoring the dichotomy altogether. Yes, that's right, just ignore the whole passive/active thing altogether, and just use one system for all situations. At this point I won't even suggest that you go with a target system, or a both-roll system or anything in particular (despite having my own preferences), just that you have only one system for everything.

This seems at first glance to cause all sorts of problems. When I say this people inevitably point out certain things. Let's say we're talking specifically about going to an all-target-number system. The objection is usually, "What do I use for a character's target number?" Usually, in the example above, they'll point out that if I roll Own Skill +d10 to exceed Opposing skill, this gives the advantage to the roller. And they're right. What they should do is use that +d10-d10 thing (many games like FUDGE do just this). Or some other balanced method. Then ratings work just fine as target numbers. If, on the other hand, we're looking at going to the all-opposed method, they'll point out that a stone has no resist lifting skill rating. To which I'll reply that it should. Or rather, you'd have to set a difficulty in a target number system; the same methodology is no more difficult to apply to creating "skills" or ratings to oppose any roll.

The point is that any well-designed resolution system can be used to adjudicate any situation. One does not need to have separate systems for opposed vs. unopposed situations as long as the method is designed properly. Which takes no more effort than creating a system that is not designed to handle such.

As usual in my rants, this is not a revolutionary idea, or something that I came up with by myself. Certainly many systems exist in which there is a single resolution system for everything. It’s just one of those problems that I see crop up again and again and the assumption of the necessity of such dual systems irks me. As always I’ll caveat this and say that such dual systems are not broken or unplayable, just that they could have been put together in a more coherent fashion with just a bit more thought.

Or at least that’s how I see it.

Mike

Mike,

I like your rant - it makes sense, and it's definitely something to think about when designing games.

I'd like to throw an example system at you and see if it works for you. I have to admit self-interest - it's the system for Paladin.

Players roll a number of dice equal to an attribute (1-5) and attempt to roll 5 or 6. Each 5 or 6 is a success. You need a certain number of successes to perform a task (1 for most things, 2 for hard things, and 3+ for superhuman tasks.)

However, when competing against an opponent, you both roll. You still need the minimum number of successes to perform the task. However, the person with the most successes "wins." The nature of this, of course, is determined by the task, but is usually measured in speed.

Example: To run a mile, one would need only 1 success - it's an average task. If running against an opponent, if you got 1 success and the opponent got 2 successes, both of you ran a mile. The opponent ran it faster, however.

Does this system have the same problems? (I have to admit I'm not mathematically inclined enough to know.)

Awesome Mike, you never fail to impress.

Mike Holmes wrote: When RPGs started resolution rolls were all of the "Roll vs. Target" type. Soon thereafter, however, somebody invented the idea of the "Opposed Roll." Where does this come from?

...Let’s look at an example. In a particular system you roll d10 + Skill to get over a target number to succeed, the level of success being the number by which you succeeded. The corresponding opposed mechanic is to have both sides roll d10 + Skill and subtract the lower from the higher, leaving the higher roller with the difference in levels of success. What's the difference between the two methods? At first glance there might seem to be a difference but consider this. What if the "opposed" system were to have one person roll Own Skill + d10 - d10 target is the Opponents Skill. This is the exact same system as the regular opposed system, statistically.

...What is different in the opposed system is the number of dice being rolled. What effect does this have? Well, the opposed roll, consisting of two dice gets you the pyramid curve (flat bell). As opposed to the unopposed roll which gets you the flat curve. Which means that the opposed roll is much more likely to come out with an "average" result than an unopposed roll. Yep, that's right, actively opposed conflicts are more predictable, than working against a stationary object. Counterintuitive, no?

...This seems at first glance to cause all sorts of problems. When I say this people inevitably point out certain things. Let's say we're talking specifically about going to an all-target-number system. The objection is usually, "What do I use for a character's target number?" Usually, in the example above, they'll point out that if I roll Own Skill +d10 to exceed Opposing skill, this gives the advantage to the roller. And they're right. What they should do is use that +d10-d10 thing (many games like FUDGE do just this). Or some other balanced method. Then ratings work just fine as target numbers. If, on the other hand, we're looking at going to the all-opposed method, they'll point out that a stone has no resist lifting skill rating. To which I'll reply that it should. Or rather, you'd have to set a difficulty in a target number system; the same methodology is no more difficult to apply to creating "skills" or ratings to oppose any roll.

I think this common problem jumps up from two beliefs:

• Critical Success/Fumble Rules
"Both people need to roll, because there is a chance either one could critically succeed/fail" In most cases, I think this is a weak argument, but there are many systems that incorporate sliding odds of doing exceptionally well, or bad, based on skill, special abilities, etc. On the other hand, no one wants to start giving rocks weight skills, hardness skills, etc. to make opposing rolls.

• "But I want to roll dice!"
Which is really,"I feel disempowered". I can certainly attest to this feeling with D&D's AC system("I run up the wall, kick off, throwing my cloak into my enemy's face as I somersault over, and land behind him!", "Ok, your AC is still 12..." "What !?!"). While this makes a lot of sense for players, not a lot of GM's really want to roll for stones' resisting actions.

Neither of these reasons are that great, but usually are the line of logic behind the TN/Opposed roll decision.

Chris

PS-Love the rants Mike, keep up the good work :)

But first, Gordon's Impromptu Rant #1 - Mike, a standard numbering system begins at one (or zero, if you want to go THAT route) and ascends from there. Considering these are largely mathematical issues, I'm wondering what demon of perversity possesses you when you number your Rants :-)

OK, that's out of the way. How I see it (probably just a rephrased version of Mike's thoughts) - opposed vs. unopposed (target number) is a myth. The opposed roll situation simply uses a randomizer (die roll) as part of establishing the Target Number. Consider what happens if we don't roll "at the same time", but rather have one side (say, the GM) roll first. He's set a target number, and now the player can roll against that number.

Thinking this way, the issue becomes one of timing (IIEE, perhaps? Or just details of timing with EE?), and establishing the granularity of what is resolved/established by a "roll." I think it likely most opposed vs. unopposed, target number vs. die roll issues are really about these broader subjects.

Gordon

One aspect of game design is involved with keeping people interested. This takes many forms: Ron's five elements of role-playing, snacks and drinks (or their lack), player skill, and many other things. One factor is the interest provided by the mechanics of the game. The board game Trouble, for example, is a bad simplification of Parcheesi except for the plastic dome that you pound to make the dice bounce. In the same way, one thrill of some games – admittedly a cheep thrill – is playing with all the platonic solids. As a personal confession, that cheep thrill is what initially interested me in role-playing games at a tender age past.

Beyond novelty, though, there is a serious role to be filled by mechanics. I will illustrate this with a worst-case example, a kind of reductio ad absurdum. Imagine a system so constructed as to call for opposed rolls in all cases of character conflict. To pick a hypothetical system, let's make the roll '1d10+skill vs. 1d10+skill'.

Suddenly, in a flash of dust and a burst of light, the world is turned on its ear and all those rolls are shifted into half as many 'skill+1d10-1d10 vs. skill' checks. This is statistically equivalent. But lo! This is a game played with a Game Master (as is often the custom), and only the characters run by the Game Master receive rolls. "What's this?" you ask. Yes, the character's sheets all contain numbers, but the players never roll. Whenever a player-controlled character takes a swing at a foe, the GM rolls two dice (adding one and subtracting the other) and explains the results.

Already some of you may be grumbling about this, but consider this in light of player interest and the rest of you may rapidly see my point. A player could say, "My character, Foo, attempts to lodge his saber deep within the infidel's vitals." The GM would roll, and announce, "You fail, but your opponent does manage to hack off your arm in a clean stroke." Remember, the GM rolls all the dice; he rolls for all the players as well as his own characters.

Player interest could rapidly wane in such a situation. Players, no longer drawn into the physical act of playing the game with their dice, could drift off and act only when their character did something, and then only by speaking. Their characters could even engage in combat by proxy; the players would all give auto pilot orders, go out for pizza, and come back to find that the GM had made all the rolls required by the system and determined the outcome of combat for them!

Now, I must concede that there is more to combat that rolling dice. There is drama and adventure and dilemma. But, there is something visceral about rolling a couple of dice that connects you to your character's fate. If nothing else, asking a player to roll - when his character is in peril of a blow to the head, for example - makes the player sit up and take notice; something is about to happen! The player is engaged not just in talking about the game, but also in the game's actual resolution.

I must compare this to a game of chess. One could very well play chess against a computer with only a mouse, in fact many do. But it is not the same thing as sitting down across a coffee table and moving well weighted hunks of ivory and teak across a smoothed and checkered board. Think how much more removed chess would be if you only played by speech, instead of moving a mouse in your hand. This is what is denied to opposing combatants striped if their chance to roll.

As was pointed out earlier, this is a matter of empowerment. Not some mere empowerment, mind you, but Empowerment with a capitol schwa. It is the function of the design to buffer the player's experience from the mathematics of the system. We could all play on computers with random number generators weighted to reflect whatever probability system we enjoyed - a perfect bell curve, perhaps - but we chose not to. The games we delight in, Sorcerer, The Pool, even my poor introduction to role-playing, D&D, all use dice. At some level they all make a concession to the people playing at the expense of the mathematics. It gives us a feeling of power to hold the winds of fate and the tides of entropy in our hands, drop them, and forge our characters' destinies.

To conclude, I submit that while it is rigidly proper to condense mechanics in the way discussed above, it is not an imperative of design, and not necessarily even good design, to do so. All of the above admitted, however, I must note that I am no zealot. Any system is a sum of all of its parts and what is lost in one area may be gained in another. I in no way wish to disparage any design with my general discussion of methods. If I wish to comment on any game specifically, I will do so separately.

Mike Holmes wrote: To which I'll reply that it should. Or rather, you'd have to set a difficulty in a target number system; the same methodology is no more difficult to apply to creating "skills" or ratings to oppose any roll.

Fang:

I follow your reasoning, but I wonder if some of the conventions of the mechanic aren't leftover from the starting point (d10 + skill vs. d10 + skill).

In the first place, why the big subtraction? Subtraction is evil in mechanics! (You may not agree, but IME it's generally a concensus.) So, while some type of subtraction might be neccessary in a mechanic without opposed rolls, it should be minimized and isolated as much as possible.

You ended up with Skill - 2d10 to get the result. Why not make that Skill + 2d10. All it does is move the center of the 2 - 20 result range up to Your Skill + 11. The numbers here don't impress me much, but the method does. More below:

In the second place, if you want a system where the roll translates directly to the degree of success using modifiers to represent situational considerations, why do you want a bell curve at all? Bell curves do bad things to modifiers. Seems like it'd make more sense to roll a d20 or a d10.

Let's do something easy like this. Everyone's used to rating thigns on a scale of 1 to 10, right? Why not say that your result is also rated on a scale of 1 to 10, indicated by a d10 roll, adjusted your skill. Let's say skills range from 1 to 5, average being 3. So your roll will be in the 4 - 13 range if you have a skill rating of 3.

Basically it boils down like this: the d10 roll gives you the result quality on a scale of 1 - 10, when compared to your other attempts. Your skill rating puts that result in context with respect to others in the game.

So, if you roll a 3, your attempt was a 3 on a scale of 1 to 10 when compared to your other efforts. But when you compare it objectively, it's a 6.

No need for opposed rolls here. Simply subtract your opponet's skill from your result - "Opponent's skill" meaning anything that hinders you (A chasm being Very Wide [5] for example.)

To address the subject of how engaging these sorts of mechanics might be, I'd say that it will all depend on the what single system you choose at the end. I'm not suggeting one sort or the other, just that you have but one. I certainly never suggested that rolls be taken away from the players. I agree with KR that a system where only the GM rolled would probably give players a sense of disempowerment (despite the fact that rolling does not actually empower the player in any way). So, um, don't do that. Use one of the zillion other methods that can be employed.

A good example of a system that uses this principle is InSpectres. In that game, all rolls are made by the players, and the GM never has to roll ever. Herr Sorensen has made it clear that he sees this as a feature. And I agree, it seems very freeing. In any case, that was his choice, and it works well.

Again, I'm not saying what sort of system I like, specifically, we can get to that in another debate (I do have a preference, and it should be obvious from my designs).

One argument I might accept is the esthetic one. That a player might be more engaged by the mechanics because they seem to make more sense intuitively if you were to have two types of resolution. The only question is does the added complexity and other potential problems cost more than the esthetic enhancement. I suppose there is probably some design that can be achieved that could give you that feel, and yet avoid just about all the downsides. But I think that the esthetic isn't worth even a little damage, personally. But, as with all things esthetic, YMMV.

Gordon, to address your point on my numbering, what would make you think that I am not perverse? ;-) Being a programmer, I probably would start with zero. Starting with three however was just a way of saying on my part that I understand that I talk too much, and that you all can probably ascribe a couple of standard rants to me to fill in the first couple yourselves. Or maybe I'm just saving them for something really important that I have to say. Who knows?

As to your reprasing, Gordon, yes, that's a good way to look at it. Statistically all that usually changes is the randomizers, essesntially. Which seems odd when you think about it. Especially in the case of the common split method that I cited.

Which brings us to Clinton and Paladin. Yes, Mr. Nixon, you have fallen right into the opposed roll trap. And in the classic manner. Your characters will have more predictability against living, breathing opponents than they will against pasive tasks. I would avoid this, personally.

In the case of your game, however, it's eminently simple to fix. Just rate an opposing force as though they were active. Thus a jump across a particular chasm could be a roll against 2 dice while a wider one would be 4. In essence, what you have right now is "Most Things" = 0 dice (expected value 0 successes), "Hard Things" = 3 dice (expected value 1 success), and Superhuman Tasks = 6 dice (expected value 2 successes). And then you have to exceed that number to win, of course. An advantage of going all opposed is that you can assign all the difficulties in between as well.

The single roll method for Paladin would be to do something like the following. Each die rolled is worth: 1-2 = 0, 3-4 = 1, 5-6 = 2. This averages one per die. Then the idea is to simply roll over the target's stat. Such a system could be used in an "only the players roll" system, FWIW. This particular method might not work with your other mechanics (I'm trying to remember the Animus rules), and there are certainly many other ways to do it (Hero System Body comes to mind), but you see my point. You can do all these rolls one way or the other rather simply, and thus only have one system.

OTOH, Clinton, the Paladin system is so simple that the "damage" done is pretty minor. You could probably ignore it and the system would suffer only minimally. Again, dual systems never break a game, they are merely unneccessary. The question is do you have a reason to keep it?

Chris is right on it when he says that there is the worry about the critical success/failure thing. As he points out the reply to such objections is to simple. Either go with "all opposed" so that you can get in your crit/fumbles on both sides of each and every conflict, or to go to the more modern system of success granularity that many systems have nowadays. Like in my example, where success is determined by the margin. In that case you have a much finer granularity of results as opposed to the success/crit-success, failure/crit-failure dynamic. But again, that's my personal bias. You can do it any way you want. But a single system can handle these results either way.

And thanks, Fang for pointing out another good example of one of the many systems that have already figured this out. ;-)

Mike

Paganini wrote: I follow your reasoning, but I wonder if some of the conventions of the mechanic aren't leftover from the starting point (d10 + skill vs. d10 + skill).

Mike Holmes wrote: This all seems well and good until you consider that you've just created two entirely different resolutions systems for a single game. Now I can go on and on about the benefits of only having single systems for resolutions, but the advantages should be pretty obvious.

Mike Holmes wrote: This is the exact same system as the regular opposed system, statistically. What this points out is that it does not matter who rolls the die or dice, effectively the systems are the same. In the end there are randomizers, and you get an output number. All rolls in all systems are opposed by their difficulty. There are no “unopposed” rolls. The dichotomy is a myth.

I was afraid I hadn't made this terribly clear. I'm sorry if breaking this down to a point by point response bothers anyone; I'd like it to be clear I hear what Paganini's saying, but I need to lead the conversation through his misunderstanding.

Paganini wrote: I follow your reasoning, but I wonder if some of the conventions of the mechanic aren't leftover from the starting point (d10 + skill vs. d10 + skill).

Paganini wrote: In the first place, why the big subtraction? Subtraction is evil in mechanics! (You may not agree, but IME it's generally a consensus.) So, while some type of subtraction might be necessary in a mechanic without opposed rolls, it should be minimized and isolated as much as possible.

Paganini wrote: You ended up with Skill - 2d10 to get the result. Why not make that Skill + 2d10. All it does is move the center of the 2 - 20 result range up to Your Skill + 11. The numbers here don't impress me much, but the method does.

Paganini wrote: In the second place, if you want a system where the roll translates directly to the degree of success using modifiers to represent situational considerations, why do you want a bell curve at all? Bell curves do bad things to modifiers. Seems like it'd make more sense to roll a d20 or a d10.

Paganini wrote: Let's do something easy like this. Everyone's used to rating things on a scale of 1 to 10, right? Why not say that your result is also rated on a scale of 1 to 10, indicated by a d10 roll, adjusted your skill. Let's say skills range from 1 to 5, average being 3. So your roll will be in the 4 - 13 range if you have a skill rating of 3.

Paganini wrote: Basically it boils down like this: the d10 roll gives you the result quality on a scale of 1 - 10, when compared to your other attempts. Your skill rating puts that result in context with respect to others in the game.

So, if you roll a 3, your attempt was a 3 on a scale of 1 to 10 when compared to your other efforts. But when you compare it objectively, it's a 6.

No need for opposed rolls here. Simply subtract your opponent's skill from your result - "Opponent's skill" meaning anything that hinders you (A chasm being Very Wide [5] for example.)

Fang,

I think that 'better ratings are smaller numbers' is so significantly more counter-intuitive, that it outweighs the "evil."

If not for other reasons you've articulated, you could call the dice output "ranks" and eliminate counter-intuitivity in one fell swoop.

Paul

Hmmm.

Mathematically speaking, Mike's 100% right. However, I personally don't see roleplaying game design as being purely about obtaining an aesthetically pleasing (from a mathematical POV) system - creating a system which is OOC enjoyable to use is also a factor.

Opposed rolls make sense for character-vs-character stuff, since it makes it feel more like a competition. Unopposed rolls make sense for character-vs-passive-thingy stuff, since it makes it feel more like you are trying to overcome an impersonal challenge to yourself.

Sure, Mike's system's more efficient in terms of dice-rolling, but making an opposed roll isn't that much more effort than making an unopposed role, and it feels nicer OOC. I know I'm being all subjective rather than analysing the situation rationally and objectively, but then again fun is a subjective experience. Some folks might get pleasure out of a mathematically efficient system - I prefer a system which enhances the atmosphere, rather than sterilises it. ;)

Darnit Paul,

Paul Czege wrote:

Le Joueur wrote: I think that 'better ratings are smaller numbers' is so significantly more counter-intuitive, that it outweighs the "evil" [of subtraction].

If not for the other reasons you've articulated, you could call the dice output "ranks" and eliminate counter-intuitivity in one fell swoop.

Wart and Reimer,

I admitted to the possibility of an aesthetic argument. What you call semiotics, Reimer, was exactly what I was refering to there. You have brought the right term to bear; very good, I missed that. From that POV, Wart, you are right, this is a preference. If you are willing to put up with the admittedly small potential downsides, for something you see as being more intuitive, then that's a choice I can't refute. But there are still some points I'd like to make.

When designing, simply deciding to have a single system is not enough to make a good system. It is merely one step. Yes, while what I have displayed only argues that all methods are mathematically equal, I do so to point out that there is no "danger" from going with one system. Once you get past that assumption (and whether you agree with it or not) the next step is to create a system that is semiotically, aesthetically, and otherwise mechanically accessible. Somehow the asumption here is that I would make some kind of ridiculous system to follow on this assumption. I would not do so, nor would I suggest that anyone else do so.

Let's take a look at some systems that follow my assumption, and work very well. Sorcerer is "all opposed", and works wonderfully. As is Dunjon Krawl (and I assume Dunjon, though we've not seen it). Mithras's Zenobia is where I got on the bandwagon originally with its opposed roll combat. Interesting how many games have combat systems without opposed rolls, or strange modifications of opposed rolls, and then straight opposed rolls for everything else (see Rant #3). Inspectres is an odd case because the rolls are entirely unopposed. That is, they are based only on the character's stat, and there are no difficulties, targets, nothing. Still works. It's one exception is Stress rolls which are separated out specifically to highlight their difference. The aforementioned Scattershot. And you'll have to trust me that all my designs that use one method all work just fine. I'm sure there are other prime examples that aren't coming to mind right now, as well.

None of these systems have any problems with a lack of data (well, InSpectres does, but intentionally). You can always adjust either side of a conflict in either form by raising or lowering targets, or raising or lowering skills. In fact the reason I am partial to "all opposed" is that you get all that information in using the exact same methods for all sides present. You simply seem to be arguing right for my favorite of the two methods, Reimer.

And none of these systems is in any way incomprehensible to anyone. The active/passive dichotomy is illusionary anyhow. Is a stone falling active or passive? What if I pushed it off the edge? How are these situations different from the player's perspective? He only want's to know if he got crushed by it or not. And having only one method, he will be better able to judge his chances, not having to learn more than one way to calculate his odds.

I am not sure what you are talking about, Reimer, with respect to GURPS or D&D. They, as you point out, do have the problem of multiple systems, which you say is bad in the case of D&D (they have improved slightly in theis area, but are still amongst the worst of offenders; it makes me laugh when people describe D&D as simple). Then you point out that GURPS (which also has opposed vs unopposed rolls, and other various types to boot) should be more complicated than it is by having more types of charts. Well, I could argue against that, but the charts in question aren't even part of the resolution system, so I don't see how that's gemain. In any case, I don't see GURPS suffering from a lack of complexity. I would argue that Ars Magica suffers from its differing resolution methods, but you'd probably disagree. I do have the designers of the game on my side, however, who's later games included less and less different resolution mechanics.

The case of Falkenstien's sorcery vs. skill checks I would agree with being a good idea. But this is because, again, there is a good reason for the dichotomy. This is not the mythic opposed vs. unopposed example, which does not merit a difference, but one that does.

All I'm saying is that, assuming that you follow up with a good system, you lose nothing from going with a single system, while gaining simplicity. This is the definition of elegance. Note that I am a big fan of complexity (reference my ongoing battle with Mr. Martin over that one), as long as it provides something. But when things can be made simpler without losing anything this is obviously an advantage.

Mike

Fang on subtraction wrote:
First of all remember, this is an opposed-roll mechanic. That's why it simplifies to Your Skill + d10 – d10 versus My Skill. The subtraction is inherent and unavoidable. Wouldn't you agree that having addition, on top of subtraction, is more "evil"?

Ah, but that would destroy the utility of the 'faux opposition roll.'

If we 'flop' the dice over into positives, we have to 'reinstall' the target number and all of them would have to be calculated as Target Number + 11. (The normative skill rating is 11, added to 2d10 makes the range from 13 to 31! That would require target numbers centering around 22! To me that's counter-intuitive, not only adding target numbers to the mix (beyond the Your Skill Rating), but also making them always land between 13 and 31. Those numbers impressed me!

Fang wrote:

Paganini wrote: In the second place, if you want a system where the roll translates directly to the degree of success using modifiers to represent situational considerations, why do you want a bell curve at all? Bell curves do bad things to modifiers. Seems like it'd make more sense to roll a d20 or a d10.

How did he put that? It was something like 'putting a high-power scope on your rifle doesn't make shooting fish in a barrel as much easier as it does in sniping' (or something like that).

The bell curve (even though, with two dice, the 'curve' looks like the silhouette of a pyramid) creates the 'diminishing returns' effect. When your skill is 11 (in Scattershot, which uses a 'Your Skill Rating – 2d10 = success level' system) then a single +1 is a difference of 9%; when your skill is 14 a single +1 is 6%. Following the scoped barrel-fishing model, a flat curve 'does bad things to modifiers;' so I guess I disagree with you here.

Fang wrote:

Paganini wrote: Let's do something easy like this. Everyone's used to rating things on a scale of 1 to 10, right? Why not say that your result is also rated on a scale of 1 to 10, indicated by a d10 roll, adjusted your skill. Let's say skills range from 1 to 5, average being 3. So your roll will be in the 4 - 13 range if you have a skill rating of 3.

That doesn't make any sense. This says that you can roll well enough to beat difficulties that don't exist. Skills are 1 to 5? Problems are difficult from 1 to 10? If you roll over 6 with a skill of 4 it beats non-existent difficulty levels.

The difference between these Scattershot and Paganini's system is that, in his, the opponent is disempowered (when played by a player) or the sensation of 'threat' is reduced (there's more suspense waiting for the other die roll) because his is strictly an 'unopposed roll' system. Scattershot looks like it is a fusion system when actually the 'faux opposition' is only supplying a variable modifier.

I know this is a really tough concept to grasp, I wouldn't have figured it out if I hadn't been working on a math minor in college. (Well, it could be easy for you, but it wasn't for me.) If you want, I can explain the mathematics of it in more detail in Private Message, if you're still not 'getting it.' That goes for anyone who finds these maths, 'over their heads.'

Mike,

Am I right in thinking this - and your other two rants - are mostly discussions about the scope of mechanics? You seem to warn us against causing unintended results in stat + skill systems, you warn us not to design more than we should in your article on combat, and you warn us not to follow wasteful design trends in your current article on opposed rolls.

If this is the case, I agree with your motives whole heartedly, elegance in game design requires restraint, or at least deliberate decision. I believe James Ernest wrote an excellent article on games with an eye to promoting a measured style of mechanic design. That is not to say that Cheapass games are sedate or conservative, only that they are designed with focused intent.

However, I do not see that this design (rolling opposed rolls sometimes and not others) is terribly inconsistent with good practice. In addition to possibly being an aesthetic decision, there are some kinds of consistency that would result in this sort of system. One consistent scheme would focus on the player and character instead of overall resolution.

To explain with an example, suppose a game had the rule, "whenever a character faces adversity, that character's player should roll 1d10." Consistent expansion of this rule would require that if two characters were at odds both of those characters' players would roll a die for themselves. This is a consistent mechanic, and while it does require interpretive guidance, it seems to be the founding consistency of many systems that use opposed rolls.

I think this is the platform from which I defend design mechanics that appear to be mixed. They are often the result of some other design consideration of equal value to consistency of mechanical interpretation. You mentioned earlier that there was a system in which only the players (separate from the GM) rolled dice. This apparent consistency is the result of an underlying inconsistency: the GM never rolls. Similarly, a consistent underlying mechanic, "everyone rolls," can create other mechanical situations that appear inconsistent. In fact I am hard pressed to think of a successful system that is universally consistent in every way.

Now, it is entirely possible that opposed rolls are included in many games for no better reason than tradition. Shaking people out of that rut is useful and good, and I wish I had the eloquence to do so myself. But I think in this case you have highlighted the negatives of a mechanic somewhat one-sidedly; but that's probably what makes this a rant.

Ring,

I think you have my motives mostly right. That is, people can make better games when they make informed decisions. I only point out the negatives to say that they exist, and so people know that they may be making a system that is not as good as it might be. If Wart decides that the aesthetics of the opposed roll/unopposed roll are important, or you decide that they are semiotically important, despite whatever downsides might exist, then you have at least made a considered judgment. Further, you can take steps to minimize the "negatives" of such mechanics (such as the problem with static actions having less fecundity than active ones), knowing that they exist.

Yes, if there is a deeper design consideration driving a split, then by all means one should go with the split in question. As I said, the Falkenstien split is a good one, as is the InSpectres Attribute/Stress split. The split I am arguing against is merely the traditionally uninformed adoption of the opposed/unopposed mechanic. Usually, what happens is that a particular mechanic is made that addresses either the opposed situation, or the unopposed situation, and then the designer goes, "hmmm, but what if the other situation occurs?" and promptly designs a second method for that case. The process should be to recognize that both "cases" (fictional, really) can occur, and make a single system that addresses them both from the start.

If designing a game from the ground up, one can still have the consistencies that you discuss using a single system. I think that people are looking at this discussion and saying, hmm, how would that work if I tried to make GURPS into a single roll type system? Well, it probably won't. GURPS wasn't designed with this point of elegance in mind, and small adjustments won't be able to bring it in line. You'd have to rewrite most of the resolution (at which point you might as well just write your own game anyhow). These rants are advice to designers making new games, and from that perspective, you can make such systems work just fine (often with less effort than trying to come up with the dichotomous method; at the very least it's that much less writing).

What you call "consistent" is only so from one POV. It is the standard POV of the industry, but not one that necessarily be taken. In InSpectres, for example, there are no simultaneous contested actions. If I attack you, I just roll, and see how well I did. Your assumption springs from the idea that the opponent's ability must be allowed to counteract the acting characters. This does not need to be so to have an effective system. Not that I'm saying that all systems should discount this, but just to show that there are other POVs as well. Even in a system that does have the standard assumptions about opponent's resisting, however, you can just take turns. I attack you; I roll against your DEX as a target. You attack me; you roll against mine. In fact this is exactly how resolution was done in systems like D&D before skills came along, as I describe in the history of the problem.

If you must have simultaneous opposed actions in some cases, then I suggest that all rolls be opposed. Again, this is why I like this method, and you seem to simply be arguing for my favorite system. Put it this way, what is inconsistent about always rolling an opposition to every character roll? Some people find it odd somehow that I would roll to see how well a lock resisted being opened. But it's just a statistical system for determining degrees of success/failure, I'mnot really rolling for the clock as though it were a character. Most people get over such aversions quickly after just trying it. I think the bias only exists at all because of the tradition. Or, rather, it doesn't aesthetically bug newcomers to RPGs in any way. In fact, it makes a lot of aesthetic sense to me, personally; it makes every conflict fell like a contest as opposed to just tasks.

To sub-rant for a moment, there are all sorts of other advantages to all opposed rolls. As Fang points out, you can get what Mega-Traveller used to call the "Uncertain" roll. Where the player only knows half of the result. This can be used to all sorts of cool effect. Also, you can roll fewer dice and still get more complex curves. Just rolling one die for each side, for example, you get the nice pyramid curve (which has a good mix of reliability and flexibility). Also, it allows for very good ranges, ranges unachievable with target systems.

Anyhow, to sum up, by looking at these sorts of things up front, one can certainly get a better result than if one merely trusts to tradition and includes mechanics just because other games have.

Mike

Paganini wrote:

Le Joueur wrote: I can explain the mathematics of it in more detail in Private Messaging, if you're still not 'getting it.' That goes for anyone who finds these maths 'over their heads.'

Yes please.

I've been lurking here(this thread) a while and came up
some questions, just to play Devil's advocate a bit.
Note:I apologize in advance for stupid questions/points.

One thing it seems to me is that this is only a problem
if you have a situation where both an un-opposed roll and an
opposed role make sense. For instance, an unopposed combat roll makes no sense. If all thing are either opposed or unopposed, admitably, you have decide which is which, but historically that isn't much of a problem( is there an opponent who can get a variable result?).
Thus you have a consistent system, in that unopposed rolls
have one distribution, and opposed have another.

Also, I can see aesthetic value in having a contested roll have a different distribution, as someone is activly trying to foil one, this models the 'feedback' of such a situation.

Also of course, there are two types of opposed events.
1)Non-Zero sum.
Frex:Consider an art contest-I make my piece, you make yours, the objectivly better one wins. You got X, I got Y. Say I win, the 'goodness' of your piece does not reduce the 'goodness' of mine. I.e. mine is just as good as if I'd just made it normally. This method I think is valid. Correct?

2) Zero-Sum
I think this is the thing your objecting to. I.E a better 'loser' roll reduces the degree of success of the winner.
This I believe is the 'invalid' method. Correct? Or in other words this is what you advocate for all-opposed systems.

all-opposed systems have the following problem in my Mind
Variablity of static quantities/Non-reproducability
This isn't a problem when comparing a static to a dynamic quantity. The Dynamic quantity absorbes any wierdness.
However when you compare static to static, strange things can happen.
I can use my board with Very Long[3] to get across the Wide[3] chasm, but that doesn't necessaraly mean I can use the board on the way back. Another way of look at it is if
I Jump with 2 success on the way across, I could have the same number of succeses on the way back and not make it.
Also, if we want to figure out how far I jumped, how do we do that?
An application of common sense can fix things like this, but it is troubling.

Various other comments:
I think it should be noted that GM's can like to roll dice also.

:)

Just thought that I'd point this one out, since this is part of a serious of (very instructive) rants on gaming assumptions...

Here's an assumption:
Multiple resolution types (e.g. opposed/unnopposed, etc.) are a bad thing.

Is this the case? Many of us assume so...

Jake

Jake Norwood wrote: Just thought that I'd point this one out, since this is part of a serious of (very instructive) rants on gaming assumptions...

Here's an assumption:
Multiple resolution types (e.g. opposed/unnopposed, etc.) are a bad thing.

Is this the case? Many of us assume so...

Jake Norwood wrote: Just thought that I'd point this one out, since this is part of a serious of (very instructive) rants on gaming assumptions...

Here's an assumption:
Multiple resolution types (e.g. opposed/unnopposed, etc.) are a bad thing.

Is this the case? Many of us assume so...

Just poking my head and and seeing if that question was a valid concern in this thread. For the most part, I agree with everything I see here, but I wanted to see if we hadn't built up 2 pages on another assumption.

A pleasure reading these, as always, Mike.

Jake

Hi Damion,

Don't apollogise, you've got some really good concerns. Not that I don't have answers, though... :-)

damion wrote: One thing it seems to me is that this is only a problem
if you have a situation where both an un-opposed roll and an opposed role make sense. For instance, an unopposed combat roll makes no sense.

If all thing are either opposed or unopposed, admitably, you have decide which is which, but historically that isn't much of a problem( is there an opponent who can get a variable result?).
Thus you have a consistent system, in that unopposed rolls have one distribution, and opposed have another.

Also, I can see aesthetic value in having a contested roll have a different distribution, as someone is activly trying to foil one, this models the 'feedback' of such a situation.

Also of course, there are two types of opposed events.
1)Non-Zero sum.
Frex:Consider an art contest-I make my piece, you make yours, the objectivly better one wins. You got X, I got Y. Say I win, the 'goodness' of your piece does not reduce the 'goodness' of mine. I.e. mine is just as good as if I'd just made it normally. This method I think is valid. Correct?

2) Zero-Sum
I think this is the thing your objecting to. I.E a better 'loser' roll reduces the degree of success of the winner.
This I believe is the 'invalid' method. Correct? Or in other words this is what you advocate for all-opposed systems.

all-opposed systems have the following problem in my Mind
Variablity of static quantities/Non-reproducability
This isn't a problem when comparing a static to a dynamic quantity. The Dynamic quantity absorbes any wierdness.
However when you compare static to static, strange things can happen.
I can use my board with Very Long[3] to get across the Wide[3] chasm, but that doesn't necessaraly mean I can use the board on the way back. Another way of look at it is if I Jump with 2 success on the way across, I could have the same number of succeses on the way back and not make it.
Also, if we want to figure out how far I jumped, how do we do that?
An application of common sense can fix things like this, but it is troubling.

I think it should be noted that GM's can like to roll dice also.

Hey Mike,

I agree with most of the principles about game design you've been discussing.

However, for the past few days I've been contemplating the supposed counterintuitive lesser "reliability" or greater "randomness" in the classic "unopposed" roll (die roll plus skill vs. a fixed target number) as compared to the classic "opposed" one (die roll plus skill vs. die roll plus opposing skill). And I'm just not seeing it.

"Reliablity" and "randomness" are not mathematically precicse terms, of course, so let me try to define them. If system B is "more reliable" or "less random" than system A, then we should be able to say one or more of the following about them:

1. The probability of success in system B is farther from 0.5, making it easier to correctly predict whether the outcome will be success or failure in system B.

2. Certain specific levels of success (or failure) are more likely to occur in system B than in system A, making it easier to correctly predict the exact outcome in system B.

3. An outcome that falls within a central range, such as -3 to +3 (or in any specific range) is more likely in system B.

4. The overall expected level of success or failure is of lesser absolute value magnitude in system B.

5. In the subset of outcomes in which success occurs, the mean level of success is of lesser magnitude in system B.

6. Ditto 5, for failure.

7. The effect of a given numerical advantage or disadvantage (say, a difference in the character's skill level) on the chance of success or failure is greater in system B, making success or failure easier to predict (more reliable) in system B when one side has an advantage.

8. The effect of a given numerical advantage or disadvantage on the expected degree of success or failure is higher in system B.

Clearly, number 1 is not applicable. Either type of system can easily be set up to produce the desired probabilities of success, so "reliability" can't be based on that alone. It must instead have something to do with the overall distribution of degrees of success and failure, or on how predictably the outcomes are affected by advantages or disadvantages.

To examine the other possibilities, let me set up a simple "system A and system B" as they might appear in a game system that uses both opposed and unopposed-style rolls. System A, with the hypothetically lesser reliability, will be the typical unopposed roll. It's a d10 plus skill vs. to roll higher than a fixed difficulty level assigned by the GM or specified in a table somewhere. The difference between the die roll + skill and the difficulty is the degree of success. System B for "opposed" rolls is d10 + skill vs d10 + opposing skill, with the difference again being the degree of success. In both cases a tie is allowed (success/failure level of 0).

Starting with evenly matched examples, we have in the A case d10 + x against a difficulty of 5 + x, with the outcome distribution:

[code]
SYSTEM A
outcome +5 +4 +3 +2 +1 0 -1 -2 -3 -4
probability .1 .1 .1 .1 .1 .1 .1 .1 .1 .1
[/code]

In the B case we have d10 + x vs. d10 + x, with the outcome distribution:

[code]
SYSTEM B
outcome +9 +8 +7 +6 +5 +4 +3 +2 +1 0 -1 -2 -3 -4 -5 -6 -7 -8 -9
probability .01 .02 .03 .04 .05 .06 .07 .08 .09 .10 .09 .08 .07 .06 .05 .04 .03 .02 .01
[/code]

Okay, now, do any of our suppositions for "B is more reliable" hold up?

1. Already shown to be N/A

2. No. Zero successes (a tie) are equally likely; all other possibilities are LESS likely in system B.

3. No. The opposite is true in all cases.

4. No. The opposite is true. In A it's 2.5, in B it's 3.3.

5. No. In A it's 3.0, in B it's 3.67.

6. No. In A it's -2.5, in B it's -3.67

7. No. In most cases the change makes more of a difference in A. In other words, an advantage or disadvantage is usually more likely to swing the outcome in A. Subtract 4 from the character's ability score, for example, and his chance to succeed is now 0.1 in A, 0.15 in B.

8. No. In both cases an advantage of +1 greater ability adds (this one's easy) 1 to the expected outcome. Same deal for disadvantages.

So, that's my dilemma. I cannot see any basis for calling B "more reliable" or "more predictable" or "less random." But I can see good reasons for calling A so, which would conform to everyone's intuitive expectations.

What seems to happen is that in B, the wider range of outcomes more than makes up for the "pyramidal" concentration of outcomes in the center. B would be more reliable ony if you were to normalize the two outcome distributions to the same absolute range before doing the comparison. But no system I'm aware of does that, nor does it seem a justifiable thing to do under the circumstances.

- Walt

In the case I was displaying it is exactly number one that is the problem (assuming I understand you correctly. We can debate these criteria in another thread, if you like). Yes, it can be avoided, it just usually is not, and is not in the example. There are lots of ways that you can reverse the nature of the curves should it suit you to do so. Roll Skill + 2 dice - 2 dice vs. target for "unopposed" situations, and Skill +1 die vs. Skill +1 die for "opposed". This makes the unopposed more reliable by my definition. Or go with Skill +1 die - 1 die vs. target to make it completely even distributions between both opposed and unopposed if you don't see either situation as being more or less reliable. The problem is again, the addition of a system without a need for one (you'll note in the balanced option that the only real difference is who rolls the dice; why not have each player roll one in every case, the GM for "passive" things as well as NPCs?).

So it's not that opposed rolls always have these sorts of math problems automatically, it's that they often do because people do not realize that such a system has such potential probelms. This is just one problem that can occur with such a system, BTW, the one I happened to throw out there. It's just what tends to occur when you do not apply rigorous analysis to your system, and just trust to tradition to provide you with proper tools. Which is all I'm warning about. My simple solution? Just avoid it altogether by not having the dichotomy. Note that in my favorite version, that I do allow such an idea in, but intentionally and to serve my own design purposes.

Mike

Question and request for assistance from Mike:

1) I was rereading Blue Planets mechanics last night and noticed this issue and one other (see below). I'm considering a rewrite of BP's mechanics to make them a bit lighter; to this end I'm seriously considering breaking their opposed roll system into something like whispering vault, where all rolls are made byt eh players. I feel (without serious evidence, admittedly) that this might reinforce the "you are dependant on your own actions" feel I would like to promote to support a theme of wilderness survival. My questions therefore are a) are you familiar with the system, b) what do you think of the general principle, c) do you think an anomalies would arise?

2) Just as an appendix of your standard rant: given that I was looking at the system in consistency terms, how do you feel about the completely seperated initiative mechanic? This is a common if elderly feature of RPG systems: a different, and usually mor erandom, mechanic for generating action order. Any thoughts on that?

contracycle wrote: Question and request for assistance from Mike:

1) I was rereading Blue Planets mechanics last night and noticed this issue and one other (see below). I'm considering a rewrite of BP's mechanics to make them a bit lighter; to this end I'm seriously considering breaking their opposed roll system into something like whispering vault, where all rolls are made byt eh players. I feel (without serious evidence, admittedly) that this might reinforce the "you are dependant on your own actions" feel I would like to promote to support a theme of wilderness survival. My questions therefore are a) are you familiar with the system, b) what do you think of the general principle, c) do you think an anomalies would arise?

2) Just as an appendix of your standard rant: given that I was looking at the system in consistency terms, how do you feel about the completely seperated initiative mechanic? This is a common if elderly feature of RPG systems: a different, and usually mor erandom, mechanic for generating action order. Any thoughts on that?

Ok, in my "shorthand", BP is:
[1-3]d10 <= dif(skill+stat)

Competence is measured in two forms: number of dice rolled (based on an Aptitude for an area of expertise), and the Stat and Skill levels are numerics, of which stat can be a negative value in the 1-3 range.

Margin of success = diff - roll for comparing opponents

As an initial joust, it seems to me that for opposed rolls, the starting Difficulty (derived from skill + stat) should have a value added to it which represents what would have been the diff modifiers for each party, minus relavent opposed skill+stat. So it would be:
[1-3]d10 <=diff(base + mods - victim abilities)

What this does not carry through is the use of the targets pootentially multiple dice.

Ah well perhaps this is work avoidance behaviour ratrher than real redesign, or at least redesign to no good purpose. Reexamning the system last night, it looks like it may be faster than I had feared, and the only other thing that bugs me is big skill lists. But maybe thats a good thing in such a sim-supporting game.

This would require some considerable work to redesign, if I read you correctly. Interesting that in this game the dice are variable. Messing with that will very much mess up the outcomes. I'd either use an entirely different system, or just use the one provided. As I've said, fixing damaged systems is often not a real solution. Designing them well fromthe ground up, it's the only option.

Mike