Slightly less randomness - 2d10 skill checks [d20 systems]

[Altair_the_Vexed]

1d20 is linearly random. There's as much chance of rolling a 1 as a 20, or any other number.

If I have +5 to my skill, and you have +0, then for any given roll of the d20, I'm more likely to achieve the task at hand than you... but - and this is my point - I'm only 25% more likely to achieve it.

+5 to a skill is not a trivial amount of learning. It isn't exceptional, but it represents a significant investment in learning to do this task.
And yet it only gives me 25% more chance than someone who has very little idea what they're doing.

I've been thinking that it might be a little more realistic to roll 2d10. There's a curve to the probability of 2d10. A 20 is less likely than a 10 - there are lots of combinations that get you 10, but only one that gets you a 20.

I'm aware that there is a variant in Unearthed Arcana that proposes 3d6 for all checks, but I think that goes just a bit too far. Too much has to change from the standard rules.
While giving us a little more of a curve to the randomness of skill use, 2d10 lets us still Take 10 and Take 20. The DCs don't need to change.

(I would leave attack rolls and saving throws as d20 checks - it's simpler to roll one die, and the critical system is made on the basis of a single d20.)

Does this seem like a reasonable thing to do? Do you think it would slow down play? Do you think it would make the game better?

[Dhavaer]

The advantage of the 3d6 system is the average roll stays the same; 10.5 for 1d20 or 3d6 compared to 11 for 2d10. You have to change to 'take 11' and maybe slightly alter some DCs.

[Knaight]

Or keep take 10, putting it slightly below average, and leave things as they are while giving a slightly improved chance at skills, martial abilities, and saves. It works out well overall, and isn't worth changing much, giving a tiny advantage in skills across the board, and a trivial tweak in favor of martial against spell casters.

[Draz74]

If you want a curve that's a little less "pointy" in the middle than 2d10, yet covers the same range (min 2, max 20, average 11), you can also use d12+d8. Give the poor little d12 some love.

Under d8+d12, results of 9-13 are equally likely, then the results taper off at higher or lower numbers, until you reach a probability of only 1/96 for a 2 or a 20. (Pretty close to the 1/100 chance for the same numbers with a 2d10.)

[Altair_the_Vexed]

Taking 10 on a d20 check is still marginally below average - the average of a d20 roll is 10.5.
It's normal outside of D&D to round to the nearest integer. For 0.5, we round to 1. If we round 10.5 under normal conventions, we get 11. :smallbiggrin:

I was tempted to suggest d8+d12, but it feels clunkier than 2d10.

[okpokalypse]

I've been using a d10 Based Custom System for a LONG time. After years of using it these were the adjustments we made when replacing d20 with 2d10:

1. Adjusted Critical Ranges for Weapons. It was a little homebrew, but it worked very well.

2. Made Double 1's and Double 0's (10's) Auto-Fail / Auto-Pass. Normally we had disused the Auto-Anything System, but with chances reduced from 5% to 1% on each end of the spectrum, it was much more acceptable.

3. Any Feats that allowed for Taking-10 became Taking-11.

[Altair_the_Vexed]

I've been using a d10 Based Custom System for a LONG time. After years of using it these were the adjustments we made when replacing d20 with 2d10:

1. Adjusted Critical Ranges for Weapons. It was a little homebrew, but it worked very well. ...
Good to know the general practice is working for you, but I wanted to keep d20s for combat and saves. It seems to me that there's more inherent randomness in shooting and swing a sword about under stress than in trying to hack a computer or craft a canoe.

[Flickerdart]

This increases somewhat the power of taking 20, unless you change it to taking 100 times the time.

[Swordgleam]

I've played tri-stat d10, which uses 2d10 for every roll, and we didn't have a problem with the probability ranges making anything less fun. There were still botches (double 10s; high is bad in tri-stat) and criticals (double 1s) every session or so, but not a couple times an encounter like with a d20.

As for switching out a d20 with 2d10, I don't see any problems aside from the ones mentioned. Keeping take 10 as a 10 seems reasonable - you give up a slightly higher average in order to be sure of your roll, but you're avoiding less randomness by not rolling 2d10 than by not rolling a d20, so you kind of deserve to take the extra hit on your average.

[erikun]

Under a bell curve, characters with higher bonuses will succeed more frequently while characters with lower bonuses will fail more often. Unless a character is putting full skill ranks into a skill, that skill will become useless at an earlier level than usual. Also, remember that most challanging roles assume making a reasonably good roll (15+) for an optimized character; something that is a lot more difficult under a bell curve.

[Wonton]

Just an idea... what if it was something like 2d8 for untrained skill checks, and 2d10 for trained?


This increases somewhat the power of taking 20, unless you change it to taking 100 times the time.

I agree with this though, seems like a fix is needed.


that skill will become useless at an earlier level than usual

Another very good point - skills don't get used enough as it is, as this would make them less useful, sooner.

[sofawall]

Also, remember that most challanging roles assume making a reasonably good roll (15+) for an optimized character;

Or rather, a reasonable adept character. Making something Hostile into Fanatical is supposed to be very hard. Doing at as a standard action at roughly level 13 is optimized. Making someone Friendly into Fanatical at the intended level 21-ish is focused, but not optimized.

[Jarawara]

I have considered this issue from time to time, but I've never really gotten around to tackling it in earnest.

Two issues I had considered:

First is that the d20 range is too great a variable. A grandmaster at a skill, having trained a lifetime to gain an almighty +18 on his skill check, can still roll a 1 while a complete and untter newbie could outmatch him by rolling a 20.

Mind you, that doesn't happen very often, only 1 out of 400 times -- but you'd be amazed how often that actually is. If you were making 10 high-quality widgets a day, then a little less than once a month a complete idiot could come in off the street and make a better widget. Don't bother hiring the master craftsman - just hire 20 idiots to do the job!

Plus, since I typically run low-level games, and the skill levels never get really high, then skill-successes often depend on the whims of the dice over that of the skill bonus.

To address that, I had considered reducing the entire range to a d12. The 'Take 10' mechanic would become 'Take 6'. 'Take 20' becomes 'Take 12'. All appropriate DC's would have to be rechecked. In that system, a mid-level character who focuses in a skill can see himself mastering a skill, far outperforming the newbie. And if anyone does become a grandmaster with a +18, then only the learned skillmasters would have any chance of competing with him.

*~*

Of course, that doesn't change the sheer randomness to the system, as the Altair pointed out. Which brings me to the second consideration I had - which technically was my first idea: Before 3E came out, I was making a skill system somewhat similar in concept - but I used a pair of dice to do skill checks, instead of a single one. I scrapped the idea when I saw that 3E had a finished, comprehensive list. I wish I hadn't now - I wish I had completed the work, as time (and Altair) has proved the failing of the single-die randomness.

The one thing I did differently was that I didn't have skill check bonuses (I had developed the system before 3E, remember). Instead, I had the untrained use only a single die (shoutout to DrWanton, same idea!), then had beginners use 2d4, the more trained use 2d6, more highly skilled use 2d8, and so on. In theory, there was only 5 levels of skill, topping out a 2d12.

This, of course, really doesn't work within the framework of 3E, so I guess I was right to scrap the idea anyway.

*~*

Edit: DrWanton had the same idea, then edited his post to something different, which ends up being the same idea anyway. Is that like a double-fudge Ninja'd or something?

*~*~*

Well, anyway, it's my day off, so I'm just posting my thoughts. Conclusion is: I think you're onto a good idea Altair.

[Nightson]

If I have +5 to my skill, and you have +0, then for any given roll of the d20, I'm more likely to achieve the task at hand than you... but - and this is my point - I'm only 25% more likely to achieve it.


Given an expert with +5 mod against someone untrained.
Probability expert will roll higher: 70%
Probability they will tie: 3.75%
Probability the untrained will roll better: 26.25%

Given a +10 mod
Probability expert will roll higher: 88.5%
Probability they will tie: 2.5%
Probability the untrained will roll better: 9%

Just so you know it doesn't translate to a straight 25% or 50% chance of doing better.