I do not get the various people who say things like the following

QuoteI can't stand percentile systems - even with an 80% jump skill you are still likly to be falling down every fifth bottomless pit, which tends to limit heroics.

I mean, what it is it about a percentile system that inherently leads to this result? I can think of two ways off the top of my head that will avoid or at least limit this sort of thing, and neither is especially uncommon. To wit: (1) Treat normal failures as "do not succeed" (in this case, you can't jump the pit) and only critical failures as "fail disastrously" (i.e., fall). E.g., in Harnmaster, a critical failure is 20% of the failure chance, so an 80% skill will only fall 4% of the time. Safe? No. Within the bounds of heroics? Yes.

(2) Difficulty modifiers giving a bonus to skill for some stuff, penalty for other stuff.

In a similar vein I don't get the difference between roll-under and roll-over--it seems that there are a lot of assumptions that get drawn into those terms which aren't implied by the terms themselves. E.g. GURPS is a 3d6 roll-under system, yes? If I'm not mistaken you could convert everything to 3d6+Skill+modifiers, need to beat 21 for success. Voilà, roll-over.

I do get that certain die methods (such as variable size pools that are summed or where you count "successes, or using tables) make it easier to generate certain curves or difficulty progressions, but wrapping things up in terms of percentile/over/under doesn't computer with me.

Likes and dislikes are just plain irrational, and I treat them the same way. I like coffee. I don't like mustard. I like roll over. I don't like percentiles. Doesn't mean they are meaningless, just means you can't change people's minds with reason. People give "reasons" for their preferences as an excuse, necause they don't like facing the fact that it's irrational.

-clash

It's simply perception.  If you told someone they have to roll 5 or higher on a d20 they would probably think it is pretty good.   The 1 in 5 chance of failure should be addressed to the task being attempted.  I can certainly see that there is some pit width, 10', 12', 15' etc. where an 80% chance of jumping it is generous.

Roll under vs. roll over?  Statistically no difference whatsoever.  It makes less difference than if your dice are blue or red.  I could at least postulate a physical difference in the color additives to the plastic.

On roll under vs. over I just say what works best for your mechanics.

Dice pools are just a nice way to get two or three "bell" distributions out of one pile of dice, a success distribution, a failure distribution and a neither distribution.

Back to the jumping…all or nothing rolls are something I always think hard about.   I like the degrees of success idea, critical success, success, failure, critical failure.  For jumping a pit it might be critical success (leap and not slowed down a bit), success (leap but must take an action to stabilize yourself), failure (barely miss the edge, make another roll to grab the edge or regain your balance to not fall in), critical failure (bad news, you fall, maybe a critical success on grabbing the edge could let you get your fingertips on the lip of the pit).

Well, some percentile games that persist to this day give characters a 0-100 (or 95, whatever) range. As such, the applicability of difficulty modifiers is less clear. After all, wasn't the point of capping the skill at 100 is that's the best you can be? Similarly, any difficulty penalty automatically means chance of failure. Nobody is so good that their skill overcomes penalties.

You take away this mental pitfall, the major problems with such systems go away. But then, you have a classic roll over or roll under system with bigger, harder to add numbers. Again, the system sort of resists modifiers implicitly.

Strangely enough, I think roll-under percentile systems are the "purest" resolution systems around. Regardless of what dice system you are using, be it target number, opposed rolls, dice pools, or whatever, you are just generating a chance/degree of success. In the end, it all boils down to a % chance of success.

"Ok, so you take your Angst pool and roll against the Miasma factor of your target. Then you take the three lowest dice and toss them at the wall, rerolling all 4s. If the Wild Die is a 9, you can draw from the Narrative Purity Deck and apply that to the remainder of your Spiritual Crisis stat..."

"So, you are saying I have a 38% chance of success with a 4% chance of a critical success?"

"Well, yes actually... but my way sounds much cooler."

I agree that it isn't a mathematically required link, but I do see a correlation.  i.e. Percentile systems like Call of Cthulhu and Unknown Armies tend to have very high-variance rolls -- and thus frequent failures in spite of moderate skill.  Of course, a lot of other systems also have high-variance rolls, so it doesn't say that much.  

A few other thoughts:

1) Even if they are mathematically equivalent, addition and comparison are generally easier than subtraction.  

2) Roll-under systems generally have a more deeply ingrained default -- that being +-0 modifier.  That is, within the rules, there will more often be "Make a Health roll" rather than "Make a Health roll at difficulty 20+damage taken".  By not making rolls resisted, you more often have mechanics that don't scale well.

Some people want to see their "good" dice roll of a high number reflected in success at a task.  Rolling two 9's is "good", so they should succeed.  Roll-under doesn't make those people happy.

I only like percentile rolls when you're using the noisy golfball.

Quote from: cmagounStrangely enough, I think roll-under percentile systems are the "purest" resolution systems around. Regardless of what dice system you are using, be it target number, opposed rolls, dice pools, or whatever, you are just generating a chance/degree of success. In the end, it all boils down to a % chance of success.

Well, but an important issue is how modifiers affect the chance, and also the distribution of degree of success.  Linear systems almost always have linear modifiers.  i.e. So +10% is quite different from +1 on 3d6.  I've got an essay that talks about bell curves versus linear on my System Design page:

http://www.darkshire.net/jhkim/rpg/systemdesign/

Quote from: jhkimRoll-under systems generally have a more deeply ingrained default -- that being +-0 modifier.  That is, within the rules, there will more often be "Make a Health roll" rather than "Make a Health roll at difficulty 20+damage taken".  By not making rolls resisted, you more often have mechanics that don't scale well.

I think this is about the one thing I recognize as significant: for rolls which use a standard number/type of die, followed by comparison with a target (e.g. 3d6 vs. Dex or some other target, or d20, or d%), roll-over seems to be the best engineering choice for rapid evaluation. You add a modifier to the die roll or to the target number (not sure which is better)...very simple. OTOH it messes with my ability to quickly calculate probabilities. Both of these, I agree, probably influence the underlying algorithms by making the designer shy away from certain approaches which come out fiddly in a given die-rolling scheme, but not necessarily in another.

Quote from: jhkimWell, but an important issue is how modifiers affect the chance, and also the distribution of degree of success.  Linear systems almost always have linear modifiers.  i.e. So +10% is quite different from +1 on 3d6.

And yes, this is another factor, in fact a much bigger one once you get past the simple engineering issues. Bell curves have built-in scaling of a sort, as do dice-pools that count successes. Linear rolls could in theory use multipliers and divisors for modifiers, and sometimes they do; for that matter you could do exponents if you wanted, but the math is too icky for actual play.

Quote from: Elliot WilenI think this is about the one thing I recognize as significant: for rolls which use a standard number/type of die, followed by comparison with a target (e.g. 3d6 vs. Dex or some other target, or d20, or d%), roll-over seems to be the best engineering choice for rapid evaluation. You add a modifier to the die roll or to the target number (not sure which is better)...very simple. OTOH it messes with my ability to quickly calculate probabilities. Both of these, I agree, probably influence the underlying algorithms by making the designer shy away from certain approaches which come out fiddly in a given die-rolling scheme, but not necessarily in another.

It's all where you put the modifier.  A good modifier that makes things easier could be...

Roll 3D6, add your modifier and get over 11 to suceed or...

Add the modifier to 11 and roll under that on Roll 3D6 to succeed.

Both appear equally succinct.  Roll over adds modifier to dice roll, roll under adds modifier to target number.  


The thing I like about roll under is it works slightly, slightly better with the statement...

Roll under your skill level + modifier on 3D6 to succeed versus...

Roll 3D6, add your skill level and modifier and get 18 or over to succeed.

For example, skill= 6 modifier =2,  Roll under 2+6=8

Roll over roll 3D6=13, 13+8=21, success (but you had to carry the one)  a slightly, slightly more complicated mathematical procedure than addition that tallies to less than 20, or even 10.

These are slight differences, very minimal I would think in a non-linear 3D6 mechanics as modifier ranges don't normally go very high numerically, at most probably adding 2 or 3 things together.

Roll-Under percentile systems just plain make sense, because we use that sort of system everyday, and most everyone can understand it.  What's the chance of rain?  30%

You don't get people questioning "do you mean OVER 30 or UNDER 30??" cause they just KNOW.  it's common sense.

it's only among rpg forum goers that the meaning is questioned, and it becomes come sort of "paralysis through analysis".

QuoteI can't stand percentile systems - even with an 80% jump skill you are still likly to be falling down every fifth bottomless pit, which tends to limit heroics.

This is a load of fetid dingos kidneys.  The odds are the bloody same no matter whether it's rolling over a 4 or under a 16 on a d20, or under a 13 on 3d6, or an 8 on a d10.  It's still the same bloody odds, the only difference is that with a percentile, you've already got those odds presented in a form that is instantly recognizeable.

Quote from: kregmosierRoll-Under percentile systems just plain make sense, because we use that sort of system everyday, and most everyone can understand it.  What's the chance of rain?  30%

You don't get people questioning "do you mean OVER 30 or UNDER 30??" cause they just KNOW.  it's common sense.

it's only among rpg forum goers that the meaning is questioned, and it becomes come sort of "paralysis through analysis".

Right.  It makes perfect sense.

To find out if it's going to rain, you just a roll a percentage and if it's over 30, it will be sunny.

Wait...roll under 70 and it will be sunny...

No...wait...roll over 70 and it will rain...

Or...roll under 30 for rain?

I give up...

:haw:

You math nerds and your crazy moon language.

Let me break it down for you.

1) Roll-OVER systems are obviously the way to go for all good men. What is more fun: JUMP over the hurdle like a hero, or CRAWL underneath it like some loser wimp? Roll-over is as nature intended, case closed.

2) Percentile systems suck. Reason: the dice. Either you roll that d100 golfball, making you look like an idiot. Or you roll not one but TWO dice. Two dice. To achieve ONE result. When you need to get to the airport, do you call TWO cabs, cracker jack? Pathetic.*

Therefore, d20 roll-over was, is and always will be the way to go.

*Well, not as pathetic as Steve Jackson, who's calling THREE.