Opposed rolls

[Triff]

Time this topic got its own thread!

Roll low, roll high, etc. Go on!

I prefer to roll low always, which is why I prefer the RQ:AiG method.

SGL.

[fmitchell]

First imagine a character with a 90% skill vs. a character with 30% skill. Basic probabilities are as follows, ignoring criticals and fumbles:

Ninety wins, Thirty loses => 0.90 * 0.70 = 0.63 or 63%

Ninety loses, Thirty wins => 0.10 * 0.30 = 0.03 or 03%

Ninety loses, Thirty loses => 0.10 * 0.70 = 0.06 or 07%

Ninety wins, Thirty wins => 0.90 * 0.30 = 0.27 or 27%

So the question is really about the last case, which is either second-most probable or most probable (e.g. if both characters's skills are above 50%)

BTW, "Roll high but not over" is mathematically equivalent to "Roll under for the largest difference", and requires one less subtraction.

I'll crunch some numbers for percentiles, but in the meantime take a look at my HeroQuest probability table. HeroQuest uses opposed d20 rules, with 1 a Critical and 20 a Fumble; if both characters succeed or fail, the one with the lower absolute roll loses. The "Marginal Victory" and "Marginal Defeat" columns represent the last case; the 18 vs. 6 row is roughly equivalent to 90% vs. 30%.

[frogspawner]

(Regarding opposition to opposed rolls: )

I can understand the "keep it moving" mentality, but there is a problem with roll and move on. If a PC makes a stealth roll and sneaks past the guard, no problem. But if an NPC assassin makes a stealth roll and sneaks past the PC they are going to wonder why they didn't get a chance to see the person. Particularly if it results in the death of a PC.

"But I have a Spot of 100%! How could he sneak past me?"

If you only roll the Sneak, a character with 100%+ will almost never (96-100) be spotted no matter how good a lookout you are.

You modify the assassin's stealth roll. And/or allow a modified listen roll for the prospective victim. Or do a POW:POW roll or other resistance table roll. Or treat it as an ambush (it's not necessarily going to be fatal, even if the assassin hits...this ain't D&D). Lots of ways or combinations of ways, and they all work without bogging things down with a counter-intuitive resisted roll, and they create tension and fun just the same. Hey, whatever works.

I liked the RQ:AiG version. A successful sneak/hide halved the listen/scan chance of the guard (more if the guard was not expecting anything).

I'd prefer to keep the (Sneak/Spot) rolls entirely separate, and not affect each other at all if possible. Something more like the attack v parry system in combat, perhaps? (without opposed rolls, of course!)

Or maybe somehow use a concept of "layers" of success: Just sneaking by/away is easy (one success required); but sneaking close to someone to pick pockets is harder (two successes); and assasination is even harder (a third success required)?

Currently I use a system where each side makes it's rolls (Sneak/Hide and Listen/Spot respectively) but the success levels just contribute bonus/penalties to a final perception-type Idea Roll by the spotter. Still not really happy with this method though.

[frogspawner]

I think the "Hide/Spot duel" is the classic example that makes people perceive a need for opposed rolls. So if we can come up with a good system for it, using a sequence of normal rolls, then we can forget the whole Opposed Rolls issue... and the hard maths!

[NickMiddleton]

I think the "Hide/Spot duel" is the classic example that makes people perceive a need for opposed rolls. So if we can come up with a good system for it, using a sequence of normal rolls, then we can forget the whole Opposed Rolls issue... and the hard maths!

The "hard maths" bit baffles me:

Better level of success wins (but losers success ameliorates the winners success a bit).

If success levels are equal, best roll wins (I use margin i.e. target-roll, as a personal preference, but higher roll is equivalent).

The only "quirk" (assuming that the amelioration ONLY occurs after determining who wins) is that on a tie of normal successes, the win is cancelled by the losers right to down grade the winners success by one level from normal success to normal failure; but that's basically what happens with a normal successful attack vs a normal successful parry anyway. One could then rule in that case that the winner achieved a "partial success".

As I say, when success levels are tied I use "best margin" (i.e. target - roll) as it feels more easthetically appropriate to the main BRP paradigm of rolling low on d100 is always better, and frankly it's usually obvious without maths who has the margin, so the subtraction is rarely necessary. But, as pointed out, "roll under, but as high as possibly" is mathematically equivalent, so I really don't see what the fuss about the opposed roll mechanic is - it's simple, straighforward and doesn't involve any significant maths.

Plus there were three optional variants in the playtest draft IIRC...

Cheers,

Nick Middleton

[deleriad]

I have been using opposed rolls in BRP games ever since I read Pendragon and a light clicked on. Not seen the new BRP rules but my house campaigns collapsed specials and criticals into one (as MRQ does) which meant that I was happier with the range of outcomes. I'm in the camp which tries to read the result off the dice without calculation so I use highest roll wins if the result is a tie. I've never come across any resistance to this among players.

"Made by most" is more aesthetically pleasing but when I tried it (it was the first way I tried to run ORs because it's the most obvious way) there were just enough "hang-ons" that they seemed to interfere with the game and once someone pointed out that I could do highest wins then that ran much more smoothly with my play style. I did occasionally find that having criticals, specials and normals felt congested once opposed rolls started to get used a lot, which is why I collapsed criticals and specials together.

I'm also becoming increasingly converted to opposed rolls in combat. I don't use MRQ's method because it's a mess but with critical/normal/partial results it runs smoothly and all you have to do is look at your dice. It feels about the same level of complexity as RQ3.

Personally I like the deep logic behind opposed rolls. You make a skill test when it's just you against a passive resistance and a skill contest when it's you against someone else. In a skill test there is a fixed modifier affecting your result. In a skill contest there's a variable modifier affecting your result.

[soltakss]

There's no Hard Maths involved. Believe me, I've done some hard maths (although not too hard) and this isn't it.

A critical success beats a special, normal, failure or fumble.

A special success beats a normal, failure or fumble.

A normal success beats a failure of fumble.

A failure beats a fumble.

Simple, assuming you can work out whether you've succeeded/failed/fumbled/specialed/criticaled.

If you get the same result (i.e. both Fumble, both Fail, both succeed normally, both special or both critical) then you have to work out who has done better.

I prefer "succeeded by most", other people prefer "highest roll", they are the same. Normally it doesn't take much calculation to work out "succeeded by most" and no calculation to work out "highest roll".

So, where's the problem?

I'm not sure about the loser's level of success having an effect on the victor's levelm of success as I don't have BRP yet. presumably that is to differentiate a critical vs special from a critical vs failure, for example. It's pretty irrelevant if that's the case as BRP doesn't have any meaningful rules for effects based on differences between levels of success.

[soltakss]

By the way, is BRP still using 1/5th Special, 1/20th Critical? What about Fumbles? Are they still 1/20th of the failure chance?

I feel it's time for some concrete examples of skill vs skill rather than wishy-washy probabilities.

[frogspawner]

As I say, when success levels are tied I use "best margin" (i.e. target - roll) ...

There you go - one number minus another! Hard maths! :eek:

[NickMiddleton]

...I'm not sure about the loser's level of success having an effect on the victor's levelm of success as I don't have BRP yet. Presumably that is to differentiate a critical vs special from a critical vs failure, for example. It's pretty irrelevant if that's the case as BRP doesn't have any meaningful rules for effects based on differences between levels of success.

Precisely the point is that the new BRP DOES have "...meaningful rules for effects based on differences between levels of success..." for every skill. The new BRP opposed skill rule is basically the fairly well known fix used by a lot of RQIII fans for the percieved weakness of Dodging (i.e. that a normal successful Dodge was bugger all use againsts a special or critical hit, unlike a normal Parry which had some effectiveness...)

By the way, is BRP still using 1/5th Special, 1/20th Critical? What about Fumbles? Are they still 1/20th of the failure chance?

Certainly was in the playtest draft and from what I've seen / read about edition zero that's still the case. Arguably, the default should perhaps have been the Stormbringer first edition scheme (Fumble / Failure / Success / Critical on 10% thresholds), but AFAIK it's the RQIII scheme that's assumed throughout.

There you go - one number minus another! Hard maths!

Are you really suggesting that given a two rolls against two percentile targets you find it that hard to give an order of magnitude approximation?

"I only made my sneak by twenty odd" "That's too bad, the guard is very alert - he made the spot by about forty odd so he's spotted you". In very few cases will the exact margin be relevant. And, as has been said repeatedly, "highest roll wins on same success level" is mathematically equivalent to the subtraction, so the rule as written DOESN'T require even the terrifying complexities of basic two digit integer subtraction...

Nick Middleton

[Trifletraxor]

Are you really suggesting that given a two rolls against two percentile targets you find it that hard to give an order of magnitude approximation?

It's not _hard_ but it does slows down things a bit.

And, as has been said repeatedly, "highest roll wins on same success level" is mathematically equivalent to the subtraction, so the rule as written DOESN'T require even the terrifying complexities of basic two digit integer subtraction...

And as have been replied, while this is mathematically a good rule, it goes against the "roll low" tradition. It might be the best solution, but it doesn't really "feel" right.

SGL.

[soltakss]

90% vs 30% gives 83%/17%. I.e. the 90% will win an opposed roll 83% of the time.

10% vs 10% gives 50%/50%

10% vs 20% gives 41%/59%

10% vs 30% gives 33%/66%

10% vs 40% gives 26%/74%

10% vs 50% gives 20%/80%

10% vs 60% gives 14%/85%

10% vs 70% gives 10%/90%

10% vs 80% gives 6%/93%

10% vs 90% gives 4%/96%

20% vs 10% gives 59%/41%

20% vs 20% gives 50%/50%

20% vs 30% gives 41%/58%

20% vs 40% gives 33%/66%

20% vs 50% gives 26%/73%

20% vs 60% gives 20%/80%

20% vs 70% gives 15%/85%

20% vs 80% gives 10%/89%

20% vs 90% gives 7%/92%

30% vs 10% gives 66%/33%

30% vs 20% gives 58%/41%

30% vs 30% gives 50%/50%

30% vs 40% gives 41%/58%

30% vs 50% gives 33%/66%

30% vs 60% gives 26%/73%

30% vs 70% gives 21%/79%

30% vs 80% gives 15%/84%

30% vs 90% gives 11%/88%

40% vs 10% gives 74%/26%

40% vs 20% gives 66%/33%

40% vs 30% gives 58%/41%

40% vs 40% gives 50%/50%

40% vs 50% gives 41%/58%

40% vs 60% gives 33%/66%

40% vs 70% gives 27%/73%

40% vs 80% gives 21%/79%

40% vs 90% gives 16%/83%

50% vs 10% gives 80%/20%

50% vs 20% gives 73%/26%

50% vs 30% gives 66%/33%

50% vs 40% gives 58%/41%

50% vs 50% gives 50%/50%

50% vs 60% gives 41%/58%

50% vs 70% gives 34%/66%

50% vs 80% gives 27%/72%

50% vs 90% gives 21%/78%

60% vs 10% gives 85%/14%

60% vs 20% gives 80%/20%

60% vs 30% gives 73%/26%

60% vs 40% gives 66%/33%

60% vs 50% gives 58%/41%

60% vs 60% gives 50%/50%

60% vs 70% gives 41%/58%

60% vs 80% gives 34%/65%

60% vs 90% gives 27%/72%

70% vs 10% gives 90%/10%

70% vs 20% gives 85%/15%

70% vs 30% gives 79%/21%

70% vs 40% gives 73%/27%

70% vs 50% gives 66%/34%

70% vs 60% gives 58%/41%

70% vs 70% gives 50%/50%

70% vs 80% gives 41%/58%

70% vs 90% gives 34%/65%

80% vs 10% gives 93%/6%

80% vs 20% gives 89%/10%

80% vs 30% gives 84%/15%

80% vs 40% gives 79%/21%

80% vs 50% gives 72%/27%

80% vs 60% gives 65%/34%

80% vs 70% gives 58%/41%

80% vs 80% gives 50%/50%

80% vs 90% gives 41%/58%

90% vs 10% gives 96%/4%

90% vs 20% gives 92%/7%

90% vs 30% gives 88%/11%

90% vs 40% gives 83%/16%

90% vs 50% gives 78%/21%

90% vs 60% gives 72%/27%

90% vs 70% gives 65%/34%

90% vs 80% gives 58%/41%

90% vs 90% gives 50%/50%

[Atgxtg]

It's not _hard_ but it does slows down things a bit.

And as have been replied, while this is mathematically a good rule, it goes against the "roll low" tradition. It might be the best solution, but it doesn't really "feel" right.

SGL.

Two other possible solutionos:

1) The simple "by the most" method: Just use the tens didg and see who makes it by the most. While 7-4=3 is isn't quite the same as 76-49=27 it is close enough most of the time.

2) The "partial success" option mentioned earlier. THe trick would be to let the "attacker" succed, but cut the effect on success down, the way a parry stops some of the damage (if you use parry APs).

For instance, if sneaking past a guard, the "Attacker" is the sneaking character and the "defender" is the guard. If the boths succeed, then the attacker only get's partw way before having to stop and duck behind cover (perhaps the guard thought he heard something?) for the rest of the round. Ot you could say that the attacker didn't any distance at all, but didn't get spotted so he can keep trying.

If you wanted to get technical with it, you could use the movement rates and give the sneaking character a certain amount of distance based on the success level, and then downgrade it by the defenser's spot/listen roll.

Most opposed contests could be handled the same way. Tie results could be treated as a deadlock, as with any for instance, gambling. Same with a climbing contest.

[frogspawner]

Are you really suggesting that given a two rolls against two percentile targets you find it that hard to give an order of magnitude approximation?

Yep. Harder than not doing so anyway. And it's unnecessary. Why must there always be a winner, immediately? Ties happen.

It's not _hard_ but it does slows down things a bit.

And as have been replied, while this is mathematically a good rule, it goes against the "roll low" tradition. It might be the best solution, but it doesn't really "feel" right.

Same here. It's all about the feel. I think it's the way it'd spoil the immediacy of a 'Dramatic Moment' (slightly).

90% vs 30% gives 83%/17%. I.e. the 90% will win an opposed roll 83% of the time.

But a 90% attack will get past a 30% parry about (90x70=) 63% of the time. So the same numbers will give very different probabilities if you use opposed rolls. How can that be right?

...

2) The "partial success" option mentioned earlier. THe trick would be to let the "attacker" succed, but cut the effect on success down, the way a parry stops some of the damage (if you use parry APs).

For instance, if sneaking past a guard, the "Attacker" is the sneaking character and the "defender" is the guard. If the boths succeed, then the attacker only get's partw way before having to stop and duck behind cover (perhaps the guard thought he heard something?) for the rest of the round. Ot you could say that the attacker didn't any distance at all, but didn't get spotted so he can keep trying.

...

Most opposed contests could be handled the same way. Tie results could be treated as a deadlock, as with any for instance, gambling. Same with a climbing contest.

That's great - 'spot on'! All that's needed is a good interpretation of the tied situation, like this! Call it a draw?

[Nightshade]

The "hard maths" bit baffles me:

Better level of success wins (but losers success ameliorates the winners success a bit).

If success levels are equal, best roll wins (I use margin i.e. target-roll, as a personal preference, but higher roll is equivalent).

Yeah. The idea that this is "hard math" just--boggles me.

[Nightshade]

Yep. Harder than not doing so anyway. And it's unnecessary. Why must there always be a winner, immediately? Ties happen.

With the skills this is usually an issue with, a "tie" is meaningless; they're usually things where, by definition, one side or the other succeeds, and where a tie if it happens, is effectively a win for one side. If one person is Hiding and one Spotting, then either the hider succeeds or the spotter succeeds; there's no result space for anything else.

[Atgxtg]

That's great - 'spot on'! All that's needed is a good interpretation of the tied situation, like this! Call it a draw?

Either than or High skill wins on ties. So if both character's roll a 40, and one guy has a higher skill he wins.

Basically the thing with RQ is that combat has a method for handling success vs success (hit but parried for a partial result) but everything else doesn't. In part because RQ was written in 1978 when 90% of RPGing was combat.

If we put some sort of staged success like I suggested in option 2 we can get the nice an easy resolution method used in combat. Since both characters get something for their success rolls, both will be happy.

[badcat]

I'd like to note that it is not about 'speed' so much as 'quality' of play. For me, the less distracted I am by unnecessary convoluted mechanics the better game I am going to run. I suspect the same for anyone. Claiming ability to deal with complications under pressure and possessing it are usually two very different animals.

[Nightshade]

The issue, however, is perceiving this as complicated in the first place. At least to my view, its not particularly, at least the quality to quality reduction, since it uses a mechanic I'm long used to dealing with anyway (as I expect any of the actual RQ players are).

[badcat]

No, that's not an issue at all. There is a difference between 'complicated' and a 'complication'...check your dictionary. I have no trouble with opposed rolls, there are simply better ways to run the game. Any game.:cool:

And hey, I never was a RQ player, so that doesn't apply either.

[fmitchell]

Wait, so the only argument is between "highest without going over" and "greatest difference in the same success level"? "Absolute lowest in the same level" was never a consideration?

Can we argue about something more relevant, like whether we crack open our morning eggs from the big end vs. the little end? (BTW, I like my eggs scrambled.)

[Nightshade]

A complication is a problem, in the end, because it makes things more complex; otherwise it'd be largely irrelevant.

And I'm aware you weren't an RQ player, but its one of the reasons the complaints on this issue are as foreign to some of us as it is; it seems to present as a problem we've never seen as one, and have trouble understanding why others do.

[Rurik]

Wait, so the only argument is between "highest without going over" and "greatest difference in the same success level"? "Absolute lowest in the same level" was never a consideration?

Can we argue about something more relevant, like whether we crack open our morning eggs from the big end vs. the little end? (BTW, I like my eggs scrambled.)

Absolute lowest affects the odds compared to absolute highest and favors the lower skill. That is why when comparing opposed rolls published rules always go with higher roll wins, even in systems where low rolls have traditionally been good - the odds are exactly the same as "Makes roll by most" but there is no need to calculate how much everyone made their roll by.

What about cracking eggs in the middle? And is it really necessary to specify morning eggs? Does anybody change their egg cracking habits at midday?

[NickMiddleton]

Yep. Harder than not doing so anyway. And it's unnecessary. Why must there always be a winner, immediately? Ties happen.

Because in the vast majority of circumstances that one would model via opposed skill rolls the outcome would be to the advantage of one side or the detriment of the other, if only marginally. Noughts and crosses (or tic-tac-toe) regularly ties - most situations one would be attempting to emulate in an RPG DON'T: someone wins the duel, the guard spots the intruder or the intruder sneaks past the guard, one person performs the best poem in the bardic competition, one person catches more food...

And the opposed roll rule allows for ties with the rule as written anyway: both rolls fail, thus achieving the same degree of success...

Precisely the point about opposed rolls is that whilst one wouldn't use them all the time, there are some (relatively common situations) where the best description of what is happening is that there is a direct competition between two skills - and BRP has never (in any prior Chaosium edition) had an explicit generalised rule for resolving such contests of opposed skills. And whilst it may not be to some people's taste the rule in the new BRP looks remarkably serviceable to me (it is as I said a close variant of my own house rule) and is a fairly common "fix" to the "Dodge problem" in RQIII.

Cheers,

Nick Middleton

[Atgxtg]

No, that's not an issue at all. There is a difference between 'complicated' and a 'complication'...check your dictionary. I have no trouble with opposed rolls, there are simply better ways to run the game. Any game.:cool:

And hey, I never was a RQ player, so that doesn't apply either.

Cat, you missed his point. He said "The issue, however, is perceiving this as complicated in the first place."

It's all about perception. Something that you might perceive as as unnecessary, someone else might perceive as important. If often boils down to personal preferences, rather than absolutes.

For instance, many BRP players consider hit location to be an unecessary complication. Others feel that the benefits that hit locatives give outweight the added complexity. Same with the major wound rule, category modfiers, SAN rules and pretty much everything else that's been in one of more versions of the system.