I find diminishing returns at the end of the spectrum overrated. I also tend to think that a +whatever situational bonus should be equally beneficial whatever level of skill you are at.

Also, it occurs to me that BRP-type percentile systems actually DO have a bell curve when it comes to increasing skills,  which I would argue is actually where you want the bell curve to be. At low skill levels it is unlikely you will succeed (in order to get an experience check), but likely you will pass the experience check. At high skill levels you will probably get the check but probably won't increase your skill from it. At middling levels you have medium odds of both earning and passing the check. If you plot y = x(100-x), which is the probability that you will succeed a roll at skill x and then pass an experience check on it, you get - guess what! - a nice bell curve.

Quote from: Warthur;731356Also, it occurs to me that BRP-type percentile systems actually DO have a bell curve when it comes to increasing skills,  which I would argue is actually where you want the bell curve to be. At low skill levels it is unlikely you will succeed (in order to get an experience check), but likely you will pass the experience check. At high skill levels you will probably get the check but probably won't increase your skill from it. At middling levels you have medium odds of both earning and passing the check. If you plot y = x(100-x), which is the probability that you will succeed a roll at skill x and then pass an experience check on it, you get - guess what! - a nice bell curve.

Cool!

Quote from: Warthur;731356Also, it occurs to me that BRP-type percentile systems actually DO have a bell curve when it comes to increasing skills,  which I would argue is actually where you want the bell curve to be. At low skill levels it is unlikely you will succeed (in order to get an experience check), but likely you will pass the experience check. At high skill levels you will probably get the check but probably won't increase your skill from it. At middling levels you have medium odds of both earning and passing the check. If you plot y = x(100-x), which is the probability that you will succeed a roll at skill x and then pass an experience check on it, you get - guess what! - a nice bell curve.

Good point, classic BRP does actually have a bell curve at the low end, now that you mention it, although RQ6 and other variants don't restrict XP rolls to passed tests (they're more like an "inverse hockey stick" than a bell curve).

If one really wanted to model diminished returns at the low end realistically, you'd have to account for task difficulty and allow for advancement on failed rolls. Unskilled people learn from simple tasks meanwhile complex tasks are over their heads, whereas skilled people wouldn't learn anything from simple tasks. Burning Wheel sort of does that.

Quote from: Herr Arnulfe;731353Bell curves might change the way a game feels when sitting back and theorizing about it, but when players are rolling the bones they're rarely thinking about diminishing returns at each end of the spectrum.

Actually, I would even argue that a more realistic method of modelling diminishing returns is by increasing the XP cost of skill improvement as proficiency increases e.g. 40KRP games, Ars Magica and BRP. It's not realistic IMO to have diminishing returns at the low end of the spectrum.

I got away with it in a percentile system by using an action table with a hand tuned diminishing return curve.

QuoteThis was present in MRQ2 - and, I think in MRQ1 - RQ6's predecessors. Same designers, of course, between MRQ2 and RQ6. My point is just that this technique existed for a few years before RQ6 was published.

Indeed it was. We tweaked it for MRQ2 and then tweaked a bit more for RQ6. But the concept was also there in Chaosium's ElfQuest and, in a slightly different way, Ringworld. Its not new, but BRP-based games have tended to be a combination of characteristic-derived bonuses and/or a static starting value.

Quote from: Brander;731294I'm surprisingly unfamiliar with those, thanks for pointing them out.  Gone are the days when I owned just about everything.

If you like your d% gaming light d00Lite is really worth checking out.

Quote from: Brander;731294No argument.  Though Unknown Armies did try to idiot proof the concept by stating that the percentages were only for when under stress or the like.

At times, players insist on rolling a trivial task. I sometimes allow them their fun, heavily modifying the roll to increase the chance of success. But yeah, CoC 6E also states that rolls should only be called for when dramatically appropriate.

Quote from: NathanIW;731302(...) The thing about a bell curve is that you still, in the end, have a percentage chance for success equal to the number of successful die results added up.  So if you're rolling 3d6 and need a 11 or better, it's statistically identical to roll under 50% on a d100.

True, though results will gravitate more strongly towards the average.

Quote from: Brander;731336All I was trying to note was that in my experience d% systems were poorly implemented by nearly all the GMs I played with, which has/had negatively impacted my perception of them.  It's an explanation, not a counterpoint of any kind.

Do you think this happens more often with d% than other systems? That would be odd.

Quote from: Omega;731342Or a great example of BGG at at its best.
"I hate percentile dice because the d10 is biased to roll 5.5."
I wish I were making that line up. Really.

:banghead:

Quote from: Warthur;731356Also, it occurs to me that BRP-type percentile systems actually DO have a bell curve when it comes to increasing skills,  which I would argue is actually where you want the bell curve to be. At low skill levels it is unlikely you will succeed (in order to get an experience check), but likely you will pass the experience check. At high skill levels you will probably get the check but probably won't increase your skill from it. At middling levels you have medium odds of both earning and passing the check. If you plot y = x(100-x), which is the probability that you will succeed a roll at skill x and then pass an experience check on it, you get - guess what! - a nice bell curve.

You get a parabola, which might be good enough. However, this all assumes that opportunity to increase a skill = one skill check and one advancement check. Not so: if you take the original RQ as the example, you get as many skill checks as occur in an adventure, but once you have a success, you only get one advancement check. Also, you can increase a skill through training. This has guaranteed results, with strictly diminishing returns in terms of time and monetary cost.

Point is that when it comes to advancement over time (the ultimate resource), BRP and similar games are pretty tricky to analyze. Harnmaster gives an advancement check after any adventure in which you score a critical success OR failure (IIRC), which comes out to a flat 20% per try, and then uses roll-over for advancement. It also has you handle training by awarding advancement checks. Newer RQs versions get rid of the need to succeed on a skill and just award advancement checks like XP, then have a slightly different mechanic for training. And so on.


I would say that as commonly implemented, BRP-like games are a little wonky in terms of handling difficulties. Fudge derivatives are remarkably clear: the dice roll tells you how well you performed, and the task difficulty is objective performance level needed to succeed. In d%, difficulty is a modifier to the roll, which then gives you a very low granularity relative performance measure--usually crit, special, success, failure, fumble.

Quote from: MatteoN;731349Which is also the meaning of "game X's system is not intuitive".

Eh, not necessarily.  More often than otherwise, I translate that to meaning "OMFG, you mean I have to learn the rules to play this game???  What the hell!"

Quote from: 3rik;731364True, though results will gravitate more strongly towards the average.

At what point along the proficiency scale does gravitating towards average become more desirable though. I mean, there's barely a statistical difference between high stats of 15-18 or low stats of 3-6 when using a 3d6 mechanic. We're talking a few percentage points difference. Is it supposed to be more dramatic when there's no meaningful difference between "Spiderman strength" and "Hulk strength", or more realistic?

Quote from: Herr Arnulfe;731353Bell curves might change the way a game feels when sitting back and theorizing about it, but when players are rolling the bones they're rarely thinking about diminishing returns at each end of the spectrum.

I'll agree that most players aren't thinking about anything but "don't fail, don't fail" or similar when rolling the bones.

"Sitting back and theorizing about it" though is the system designer's job in determining what systems to use to get a desired feel to a game.

Quote from: MatteoN;731346** IIRC, Hero accounts for this facet by having the lowest target number for a skilled character be 11- on 3d6.

Quote from: Herr Arnulfe;731353It's not realistic IMO to have diminishing returns at the low end of the spectrum.

These are good points, but most games I know put a floor on general overall ability at about the 40-50% level in systems through some mechanism or another.  To go way back, Traveller used 8+ modifier on a 2d6 roll for to hit, which is ~42% without modifiers.  Hero is noted above and in Gurps it requires a significantly bad attribute to get below 8- on 3d6 outside of extreme difficulty.

Quote from: Brander;731372These are good points, but most games I know put a floor on general overall ability at about the 40-50% level in systems through some mechanism or another.  To go way back, Traveller used 8+ modifier on a 2d6 roll for to hit, which is ~42% without modifiers.  Hero is noted above and in Gurps it requires a significantly bad attribute to get below 8- on 3d6 outside of extreme difficulty.

True, in reality I suppose most 3d6 systems function more like a "half-bell", starting at say 40% then rising and dropping steeply before tapering off gradually at the high end.

Quote from: Warthur;731355I find diminishing returns at the end of the spectrum overrated. I also tend to think that a +whatever situational bonus should be equally beneficial whatever level of skill you are at.

I'm not sure I find it "realistic" or "versimilitudinous" that, for example, a world class sniper is equally impacted by a "cross-breeze" as a rank amateur.  To one it's just part and parcel (and this a very minor issue) and to the other it would be quite a problem (and thus a big modifier).  With a % system this would generally equate to different penalties to each, whereas in a bell curve the same penalty for a "cross-breeze" would have the effect of reducing the better shooter's skill less and the less-skilled shooter's skill more.

None of this is saying one or the other is better, just that they achieve things in different ways.  In the above example, it's more of a GMs call for a d% but it's built in for a bell curve.  I can see good cases for either being the desired feel for a system.

Quote from: 3rik;731364Do you think this happens more often with d% than other systems? That would be odd.

No, I now just think it happened to me more often.  :-)

Quote from: Herr Arnulfe;731373True, in reality I suppose most 3d6 systems function more like a "half-bell", starting at say 40% then rising and dropping steeply before tapering off gradually at the high end.

That's why my preferred bell curve is XdY + trait + modifier (where x > 1), since it creates a floor of "trait"+ Xd1, makes small modifiers matter a lot to lesser skill levels, matter sort of linearly to moderate skill levels and much less to great skill levels.  And if you graph the results it looks a lot like a half bell (on anydice this would be the "at least" button).

The first example I know of this system is classic Traveller combat, though a 2d6 has more of a triangle curve than a bell.

And none of this is to be taken as a slam on d%, it's just a different way of doing things to get a desired feel.  It would be exactly the same to pull the % out of the resulting graph and use a d% roll.