Suppose we substituted 3d6 for a d20 roll?

[Sadras]

I like the table, not necessarily going to use it for a full campaign (maybe a session or two if I like it), but it is still a cool table nevertheless.

 

[5ekyu]

Well, see, having played a lot of different dice systems, let me say that apart from aesthetics - its the DC bot the dice that matter.

"Bell curve" "smell curve" it means nothing without the dc...

Except for the extreme cases you can get the "success" results b y just swapping around the Dcs you need for success fail.

 

[Blue]

Bell curves make the game a lot less granular and makes min-maxing for any extra modifier a much bigger deal.

I'm going to talk about normal bell curves right now, I haven't worked out the math on your red-die modifier but it seems like it won't make a large difference in what I talk about.

With bounded accuracy, the results you need are supposed to be near the middle for on-level challenges. Consider two characters, one that needs an 10 or higher (55% with a d20) and one that needs an 11 or higher (50% with a d20). That +1 grants a 5% difference in success with the current system.

But with a classic 3d6 bell curve, that is a 12.5% modifier. So a +1 weapon or starting with perfectly aligned race ability scores for your class can give you a large leap up over others.

And because of bounded accuracy, these are really the scores we care the most about, the ones near the middle. It's above a 5% difference until you get to the extremes of 15+ or 6+ - in other words for probably 95% or more of the dice rolls needed to make the target in actual play it gives a much bigger modifier then 5%

So in general bell curves decrease granularity, penalizes build "casual" or themed builds because a +1 is now worth much more in the system, and makes +X items a much bigger difference. None of those are positive for me.

 

[5ekyu]

Bell curves make the game a lot less granular and makes min-maxing for any extra modifier a much bigger deal.

I'm going to talk about normal bell curves right now, I haven't worked out the math on your red-die modifier but it seems like it won't make a large difference in what I talk about.

With bounded accuracy, the results you need are supposed to be near the middle for on-level challenges. Consider two characters, one that needs an 10 or higher (55% with a d20) and one that needs an 11 or higher (50% with a d20). That +1 grants a 5% difference in success with the current system.

But with a classic 3d6 bell curve, that is a 12.5% modifier. So a +1 weapon or starting with perfectly aligned race ability scores for your class can give you a large leap up over others.

And because of bounded accuracy, these are really the scores we care the most about, the ones near the middle. It's above a 5% difference until you get to the extremes of 15+ or 6+ - in other words for probably 95% or more of the dice rolls needed to make the target in actual play it gives a much bigger modifier then 5%

So in general bell curves decrease granularity, penalizes build "casual" or themed builds because a +1 is now worth much more in the system, and makes +X items a much bigger difference. None of those are positive for me.

My biggest gripe with bell curves is - your only count middle theory aside - is precisely that it makes modifiers unpredictable. A +2 might be worth less on a roll than a +1 is on another and ir realky depends not on sone predictable theme like "latters more for harder" or "matters more for eadier" but for "matters more when its meh difficult" and so on.

So do modifiers like say cover. Any linear modifier gets oddly wonky - hardly changing things sometime, hugely changing things other.

Its much easier imo to keep calculable measurable on the fly numbers and just rejigger the DCs to get the "ranges" you like.

If you want a task to be 17+ on 3d6 hard, give it a DC that needs a realky high net roll or if it not an instant task a high adC and multiple checks-stages.

For me, i have never seen a situation in a game where i felt there was enough precision in description and skills to see anything more "precise" than 5% increments as needed.

"No, Bob, really, because we used 2/3 of a pinch of fresh bat guano AND the corn meal was ground 37 times, precisely, not 36 but 37, i am certain the chance to resist would be 23.2% not 25%."

 

[Thomas Bowman]

So we would have to alter the magic item list instead of having weapons with magical bonuses of +1 to +5, we would have only +1 weapons and +2 weapons, but you know what, they could still have the same old magical damage bonuses. To give you an example, these would be the sorts of magical longswords one might have

Longsword +1/+1
Longsword +2/+1
Longsword +3/+2
Longsword +4/+2
Longsword +5/+3

The bonus after the slash is the bonus to hit, the bonus before the slash is the bonus on damage.

Also the saving throw bonuses would increase more gradually with level increases, your basic attack bonus would also increase at about half the rate as is currently shown in the player's hand book. And this is assuming you use the table from right to let, converting all the d20 armor classes and DCs to 3d6 armor classes and DCs. Armor listed in the player's handbook would give you smaller bonuses, the main advantage would be certain armors would be lighter and have less penalties for certain activities for a given armor bonus. But if you don't want to convert your D&D dungeon to 3d6 you can simply use the stats as is and follow the chart from left to right using the red die to differentiate between different middle range results.

 

[Psikerlord#]

Pretty sure this is what GURPS uses. 3d6 roll under system. By using 3d6 instead of d20 roll under (which older dnd used), you will get more average results towards the middle, reducing randomness and making the outcome more predictable. It will give you a less swingy game, which may or may not be desirable. Personally I prefer rolling polyhedral dice too much, but having said that, I had good fun with shadowrun and its d6 pools too.

 

[jhingelshod]

One of the subtle things that GURPS does with the 3D6 bell curve is to allow highly skilled attackers to trade 2 points from their attack to reduce an opponents defence (block/parry/dodge) by 1 point.
Take for example an attacker with a skill of 20 fighting a defender with a block score of 14. Ordinarily, the attacker would hit 99.5% of the time (18 is an auto miss), but about 91% of their blows would be blocked. However, the attacker could perform a clever feint, in game terms reducing their effective skill by 6 points to 14 in order to reduce their opponents block by 3 to 11. The attacker still has a 91% chance of landing a blow, but the defenders chance of blocking is reduced by almost 30%.
It's a simple and clever mechanic that wouldn't be possible with a d20 resolution mechanic.

 

[prosfilaes]

My biggest gripe with bell curves is - your only count middle theory aside - is precisely that it makes modifiers unpredictable. A +2 might be worth less on a roll than a +1 is on another and ir realky depends not on sone predictable theme like "latters more for harder" or "matters more for eadier" but for "matters more when its meh difficult" and so on.

The way I've seen it argued is that it helps most when it would it realistically make a difference. If you can't hit the broadside of a barn, +1 can multiply your odds of hitting something several fold (e.g. 0.5% to 1.9%), even if they're still low. If you're aiming at the broadside of a barn, a -1 will multiply your odds of missing it by the same amount, even if you're still talking a 98.1% of hitting it. In the middle, a +1 can change the odds of hitting for 50% to 62.5%; a large change, but not really changing the perceived odds much (it's still about half the time.)

For me, i have never seen a situation in a game where i felt there was enough precision in description and skills to see anything more "precise" than 5% increments as needed.

I could argue for a d10 system, on the grounds that even that's too fine. For Pathfinder, I'd cut the number of modifiers down by half instead of reducing the modifiers, which wouldn't be a bad thing. But I can't see it being worth the work.