Let's Talk About Core Game Mechanics

[dbm]

Incidentally and just my opinion of course, but for me, trying to model the ends of the bell curve as described here, off the cuff, is where things fall over for me, despite liking the idea of bell curves, a lot. Having to quickly reverse engineer the "feel" of a 16 vs. 15 vs. 12 and then applying mods is probably a level of intuition I never internalized.

Bell curves and so on feel like a better simulation, but linear is extremely straight(heh)forward.

I can totally understand that. One of the tools in GURPS is a set of general difficulty modifiers, so you can just think about how tricky in a narrative or world-sim way you want something to be and apply that level of mod. The calibration should work out at the macro level as people are given guidance on what level of skill they should be buying to model a given level of competence.

 

[Bill Zebub]

I understand that some people have a hard time wrapping their heads around the math of normal distributions, or they just don't like it as a game mechanic (which is a valid opinion.) But "the antithesis of storytelling" or that it somehow "forces heroes to be normalized"? Those don't make any sense at all.

As for this:

output 1: 12, 11, 7, 10, 10, 13, 12, 9, 14, 6, 9, 7, 8, 5, 10, 12, 9, 9, 10, 10, 11, 13, 15, 10, 8, 10, 4, 12, 15, 14, 17, 3, 12, 10, 11, 11, 7, 5, 11, 13

Why is that somehow more or less exciting than a similar string of numbers that a d20 would have produced?

 

[Bill Zebub]

Incidentally and just my opinion of course, but for me, trying to model the ends of the bell curve as described here, off the cuff, is where things fall over for me, despite liking the idea of bell curves, a lot. Having to quickly reverse engineer the "feel" of a 16 vs. 15 vs. 12 and then applying mods is probably a level of intuition I never internalized.

Bell curves and so on feel like a better simulation, but linear is extremely straight(heh)forward.

Given the number of people who have argued that "a +1 bonus only makes you 5% better in combat" I think even a linear 1-20 is pretty hard for a lot of people.

Also, I'm not sure the precision really matters. Nobody will notice the difference If you make something 3%, 7.5%, or 11% easier.

 

[mamba]

Why is that somehow more or less exciting than a similar string of numbers that a d20 would have produced?

it isn’t, a d20 simply is a lot less likely to produce this sequence (one where every result is basically around the middle of the distribution)

 

[Bill Zebub]

it isn’t, a d20 simply is a lot less likely to produce this sequence (one where every result is basically around the middle of the distribution)

What's weird is that it isn't the numbers that matter, it's what the numbers equate to in game terms. Without knowing what each of those numbers determined in a game (A miss? A hit? A crit? A failed spell? A piece of treasure?) there is no meaning. So I find it puzzling that the person complaining about math (technically statistics) "ruining art" would use this argument.

And, anyway, there was a 17 in that list. A 17 with 3d6 is less than half as likely (~2%) as a 20 on a d20. That makes me horny.

 

[Micah Sweet]

it isn’t, a d20 simply is a lot less likely to produce this sequence (one where every result is basically around the middle of the distribution)

I like the idea that most results are near the middle of the distribution. Feels more realistic for most actions. But if there's some specific activity where that's not the case, I'm happy with making an exception for it. That's what subsystems are for. I love subsystems!

 

[dbm]

I find the asymmetry of Daggerheart attack rolls quite interesting. Players use 2d12 (triangle distribution) while NPCs use d20 (linear distribution). The implication of this is that the output of players will be a bit more normalised while the output of monsters will be a bit more varied.

I haven’t tried it yet but that could deliver an interesting feel to combat where the players have more confidence in their characters, and the monsters provide spice through increased unpredictability. The rules also mean PCs will get a crit success more often than NPCs do, as well (PCs crit on any roll of a double - 8.5%. NPCs crit on a natural 20 - 5%).

 

[Gorgon Zee]

If all you care about is binary success/failure than a linear distribution (e.g. d20 vs. DC) is fine.

But if you want degrees of success, then a normal distribution gives you the ability to have extreme outcomes that are much less likely than 5% (which isn't that unlikely). For example, with 3d6 the two extreme ends (3 and 18) are a little less than 1/10th as likely as 1 or 20 on a d20.

For any dice mechanic, there is the distribution of the dice results, which typically are linear (e.g. d100), bell-curve (e.g. 3d6) , or some form of long-tailed (e.g. exploding dice). And then there is the mapping of numeric results to the game-world results, which can be varied, but generally look a bit like this

• target number is all you care about
• target number with a potential critical success
• target number with critical success / fumbles
• various target numbers determining degree of success.

d20 is one of the simplest systems - linear results, target number + criticals. That's why it's good default. The bad thing is that everything is linear, so your chance of a critical is independent of your chance of success, which is not tremendously realistic.

BRP's d100 has a linear rolling distribution, but has non-linear degrees of success depending on your skill, so your chances of good or extreme success are proportional to your skill. Very reasonable, with the only bad side being that you can end up dividing 67 by 5 in your head to work out if you got an extreme result. Rolemaster used a table (of course!) for results, so the mostly linear dice with rare exploding dice mechanic had highly non-linear results.

Pendragon uses the "blackjack" - roll as high as possible but don't exceed - system, which has the same issue as d20, but has a nice feature that you can directly compare the dice rolls to resolve contests, whereas in a d20 system if you wanted to do that, both sides would need to calculate the amount they exceeded their target by -- slowing the game down a little.

The One Ring is an odd example where there are two separate systems being used at the same time with the same dice roll. One is a simple bell curve where you roll a bunch of dice and compare to a target number to determine success, and the second is where you count the number of 6's rolled on the 6-sided die to determine degree of success, conditional on the first evaluation being a success (there are some extra fiddly bits with rolling 1's and 12's). It should be annoying, but in play I've been quite happy with it, even though I have little intuition on what a likely outcome is.

The above are examples of fairly standard systems. In general, linear dice + linear results are simple, but not terribly realistic (d20). Next step up is non-linear dice + linear results or linear dice and non-linear results. The latter I find the sweet spot -- it's easy to know your chances of basic success, and on the relatively infrequent occasions where you roll well, you do a little math based on your skills to see if you get a better than normal success.

If you go for non-linear rolls + non-linear results, it's hard for people to get a feel for what the expected outcome is likely to be, and that is an immersion-breaker as we'd expect our characters to have that knowledge.

And then there are the oddball systems where which do weird things like collecting sets of similar dice (Outgunned, One Roll Engine). For me, they have to be very thematic to overcome the annoyance of not having a feel for what's going on. So the Classic Deadlands mechanic where you roll dice to see how many cards you get and then assemble your best poker hand is an insane, unpredictable, bonkers mechanic -- and I love it for the theme!

d20 and Fate are the only two systems I play regularly that are all linear. Like @Bill Zebub I do like to see non-linearity in results, but I also like the simplicity of linear dice rolling, so for me the sweet spot is a linear dice roll and non-linear results -- and I think BRP has the best approach I have seen for that. Anything more complicated, and there has to be a big thematic payoff for me, otherwise it feels like the system is trying to obscure its mechanics.

 

[mamba]

What's weird is that it isn't the numbers that matter, it's what the numbers equate to in game terms. Without knowing what each of those numbers determined in a game (A miss? A hit? A crit? A failed spell? A piece of treasure?) there is no meaning.

true, but I am not aware of any games where the average result leads to the more unusual outcomes. You either hit or miss, depending on the target value, but you neither critically fail nor critically succeed, so it generally will be among the most 'normal' and definitely most common outcomes possible.

From my understanding that was the OP's complaint, that a bell curve makes the more exciting / interesting stuff less likely to occur compared to a linear distribution.