D20 vs D6

[Greyfeld85]

Most tabletop systems run either D20 or D6 as their core die. As somebody who's had lots of experience with the former, but not much with the latter, I'm hoping to pick the brains of those who have played a lot of both.

What does each side bring to the table? Is D20 better at some things than D6, and vice versa? Are they basically the same thing?

This is a meta question and I'm hoping for a meta discussion. Feel free to answer by using different games for framing context, but please don't turn this into a pissing context about which systems you like better.

[Totally Guy]

What dice you roll means a lot less than you'd think. It's why you roll them that's important. That includes what the roll means and when you roll.

[snoopy13a]

If by "d20" you mean D&D and by "d6" you mean the old West End Games system then the difference is of one between a skill-based system and a level-based system.

But the dice themselves don't mean as much.

[Tholomyes]

In my experience with games in which d6s are rolled, there are usually multiple dice being rolled, which means the dice are usually more normal of a distribution (depending on the resolution mechanics, this may not be the case, but it often is). As a result, the difference between success between a modifier x and a modifier x+1 is more significant in rolls in which success is roughly a 50/50 outcome, but means much less when the chance for success is closer to either 100% or 0%. With d20s the probability curve is flat, so a difference between a modifier x and a modifier x+1 is always equal.

Generally, games are made and balanced with this in mind, so looking at individual resolution mechanics are usually necessary to determine the functional differences.

[Greyfeld85]

If by "d20" you mean D&D and by "d6" you mean the old West End Games system then the difference is of one between a skill-based system and a level-based system.

But the dice themselves don't mean as much.

D20 means D&D, M&M, Modern/Future D20, Star Wars D20, and any other D20 system under the sun.

Similarly, D6 means any game that primarily uses D6 dice, like Star Wars Edge of the Empire or FFd6.

Alternatively, if you think another system does it better, like WoD's D10 system, that's definitely worth discussing as well.

For the record, I know that "why you're rolling the dice" is the biggest part of any system, but there has to be inherent pros and cons to using specific dice as the core of your system. If there wasn't, every system would use the same type of dice.

[Scow2]

I've had too little experience with d6-based systems.

d20 is about a flat, 5% increment percentile chance, and increasing your ability to reliably do a task also increases the chance of pulling off more difficult tasks by the same amount.

From what I've seen, d6 tends to offer Bell-curves for results, so better skill dramatically increases the chance to do moderate-difficulty tasks, but greater-difficulty tasks increase by a lesser degree.

There's also "Dice Scale" systems, like Savage Worlds and Ironclaw, where skill is represented by a die size vs the target number - Greater skill increases the chance of success of routine tasks moderately, and brings greater chance to pull off crazy stuff without ever completely negating failure chance.

[Ozfer]

(Haven't read any other replies, so forgive me if I recap)

I am currently writing a system, and chose D6s, but I believe both have their merits:

D6s:
-Are easy to purchase, and are available to purchase (and therefore roll) in large quantities
-Have a small sample set, making it easy to organize your system (You can do the exact same thing with D20s, but it makes more sense to say, "When a 6 is rolled", as opposed to "When a 17-20 is rolled")
-They are satisfying to roll in groups. I personally wouldn't want to be rolling 5D20s

D20s, on the other hand allow for greater complexity and options, but I simply feel that D6s have a more clean, down to earth feel.

In other words, the reasons I prefer D6s are, for the most part, completely irrational.

EDIT, rephrase: I think d20s offer a more mathematical, calculating feel, while D6s, for some reason, feel more like a tangible result to me. Especially with a Success/Failure system (1-3 fail, 4-6 success).

[Tholomyes]

There's also "Dice Scale" systems, like Savage Worlds and Ironclaw, where skill is represented by a die size vs the target number - Greater skill increases the chance of success of routine tasks moderately, and brings greater chance to pull off crazy stuff without ever completely negating failure chance. It's a shame, that I don't like Ironclaw as much as I should. The resolution mechanics are mathematically and practically brilliant. It's not just the Anthro thing (I'm not a fan, but it wouldn't be a deal-breaker to refluff), but it's a much Lower-fantasy system than I'd like.

[Rainbownaga]

Normal distributions (multiple dice) are probably the more realistic, since small differences are more relevant (as a proportion of possible outcomes*) when the chance is around 50%, but extreme differences become even more unlikely as the value reaches the extremes (18 on d6 is close to 0.5%)

Realism is not necessarily more fun though- It would get pretty boring if things could only hit you on a 17+ (1%) let alone an 18.

[Frathe]

Realistically, the difference is that d20 systems are derived from D&D ("d20 System" is in fact the generic name for 3rd edition D&D's generic dice system), while most other systems use d6s because they're the easiest type of die to get ahold of. You have to sell d20s specially (which WotC probably likes), but using weird dice would slow adoption for a smaller system. Some smaller systems also use d10s because humans like base-10 numbers.

[king.com]

Realistically, the difference is that d20 systems are derived from D&D ("d20 System" is in fact the generic name for 3rd edition D&D's generic dice system), while most other systems use d6s because they're the easiest type of die to get ahold of. You have to sell d20s specially (which WotC probably likes), but using weird dice would slow adoption for a smaller system. Some smaller systems also use d10s because humans like base-10 numbers.

The real dice king is the d10. Percentile dice all the way. Everyone loves knowing exactly what their chances are at any give time with a d10. Also yea, base 10 numbers are nice for people.

I agree that the dice itself is not really important. You use it to generate a series of numbers within a range and then mix and match dice to try and get the right swing. What is judged correct is going to vary on what your doing.

[Rhynn]

What dice you roll means a lot less than you'd think. It's why you roll them that's important. That includes what the roll means and when you roll.

I don't quite agree. I mean, philosophically, sure, but the math changes a whole heck of a lot. You get very different systems mainly depending on whether you roll single dice or multiple dice.

Single dice get you an even distribution. Every result in the range is equally likely, be it 1/6th, 1/10th, 1/20th, or 1/100th. This gets you an easily calculated linear probability of success. (Except, of course, when you're rolling dice pools and trying to get X successes, which is a perfectly cool approach. The calculations just get a bit more complicated.)

Multiple dice are distributed on a curve. 2D6, 3D6, 2D10, whatever. The average is the most likely result, and the further you go from it, the less likely it is. (On d20, 20 is 5%; on 2d10, 20 is 1%.)

I prefer multiple dice to a degree because it lets me easily create systems where average people are more likely than 50/50 to succeed at average tasks, yet skilled people don't jump up to 100% at those tasks; see GURPS, for instance (although GURPS also uses auto-failures on 17-18). 3d6 is probably my favorite resolution approach. (Although RuneQuest's d100 with skills harder to increase as they get higher works out in practice, too.)

Like Tholomyes says, this also affects how important modifiers are. If your skill is average, on 3d6, a modifier of +1/-1 can change your chances of success by as much as 12.5%, but if your skill is very high (or very low), a modifier of +1/-1 can be, at its smallest, 0.46%.


Realistically, the difference is that d20 systems are derived from D&D ("d20 System" is in fact the generic name for 3rd edition D&D's generic dice system), while most other systems use d6s because they're the easiest type of die to get ahold of. You have to sell d20s specially (which WotC probably likes), but using weird dice would slow adoption for a smaller system. Some smaller systems also use d10s because humans like base-10 numbers.

Not really true at all, IMO. Many games use d20s and mechanically have no resemblance to any form of D&D. For instance, Pendragon uses d20 as the basic resolution die, but it's essentially BRP with skills divided by 5 and nothing to do with D&D.

[Frathe]

Not really true at all, IMO. Many games use d20s and mechanically have no resemblance to any form of D&D. For instance, Pendragon uses d20 as the basic resolution die, but it's essentially BRP with skills divided by 5 and nothing to do with D&D.I've never heard of Pendragon. Or BRP, for that matter (though I looked them up). I'd consider that a fringe case, because it's really a d100 system that was divided by 5. And some other systems use d20s, I'll admit, because icosahedrons are nice regular shapes. It's just relatively rare, and usually associated with D&D.

[TypoNinja]

I've played games that do d20 d10 and d6's, and I have to say d10 systems are my least favorite.

d20 systems I have experience with (D&D, D20 modern mostly) are about increments. You roll a d20 and try to get as many modifiers stacked up as possible so that even a low roll is still a high enough result to get the job done, you tend to spend your time trying to take chance out of it.

d10 systems are possibly my least favorite (WoD, vampire, werewolf, and mortal) Success on an 8 or higher, with no modifiers leaves you with really, really, really, really random results. Its not uncommon for a character whos invested a lot into a skill, has a dice pool of 10 or more, rolls them all and one roll you might see 1 success, while the next is 15 (exploding dice). Or worse, you roll your 13 dice, pull three successes, and somebody who's only marginally invested in the same skill rolls 5 or 6 dice and gets the same 3 successes. The lack of any normalizing factor is a major downside in my opinion. Somebody who's supposed to be much better at a given skill won't be consistently better at it.

d6 systems I've only played once, a Shadowrun game that a friend ran at Camp one year, that I found superior to d10's because while the d6 pool system gave you lots of dice to roll and each die was its own target number, the target number wasn't fixed. An easy task might only be a 2, a heroic task could be a 6 or more. Many dice gave somebody who was exceptional at a skill the chance to pull off heroic moves, and at the same time made them far better than an amateur at routine tasks. Which is a big positive in my opinion.

[Greyfeld85]

I've never heard of Pendragon. Or BRP, for that matter (though I looked them up). I'd consider that a fringe case, because it's really a d100 system that was divided by 5. And some other systems use d20s, I'll admit, because icosahedrons are nice regular shapes. It's just relatively rare, and usually associated with D&D.

I suppose that depends on how you define "associated with D&D." I would be willing to accept that D&D created the popularity of the D20 system, but beyond using a D20 as the core die type, many systems have many differences from D&D itself. Especially if you consider that you essentially have to discuss *which* D&D version you're talking about, since each edition has basically been a completely new system.

[Zombimode]

Das Schwarze Auge (The Dark Eye) also uses d20's and this system i nothing like a d20 system. And as the biggest german RPG system, thats hardly a fringe case :smallwink:

[Rhynn]

I suppose that depends on how you define "associated with D&D." I would be willing to accept that D&D created the popularity of the D20 system, but beyond using a D20 as the core die type, many systems have many differences from D&D itself. Especially if you consider that you essentially have to discuss *which* D&D version you're talking about, since each edition has basically been a completely new system.

And I was only talking about using d20 as a core die in non-D&D/non-d20 System (a specific system, mostly using the OGL, not just "any game that uses d20s") games. If we want to talk about d20 as an auxiliary die, it's even more common.

Incidentally, d20s apparently predate d10s, because some RPG books tell you how to use your d20s for d10s.


Das Schwarze Auge (The Dark Eye) also uses d20's and this system i nothing like a d20 system. And as the biggest german RPG system, thats hardly a fringe case :smallwink:

Yeah, RuneQuest/BRP is over 35 years old and outside of the US, RQ was one of the most popular RPGs during the heyday of RPGs (Japan and Finland for sure). Pendragon is only the Arthurian RPG, 27 years old and still in print...

[Greyfeld85]

And I was only talking about using d20 as a core die in non-D&D/non-d20 System (a specific system, mostly using the OGL, not just "any game that uses d20s") games. If we want to talk about d20 as an auxiliary die, it's even more common.

Incidentally, d20s apparently predate d10s, because some RPG books tell you how to use your d20s for d10s.

I never said anything about systems that use it as an auxilery die.

[Togath]

From a statistics standpoint; d6s tend to give better curves, since in multiples they have better chances of better average results than a d10 or d20.

D10 have the special property to generate either 10% chances, or a random number from 1-100, but the percentile function requires two.

D20 has each result at about a 5% chance, and is easy to roll/tell the result, which is likely why it's used in things such as DnD, Pathfinder, or Legend more often than the d10/d%

[Rhynn]

D10 have the special property to generate either 10% chances, or a random number from 1-100, but the percentile function requires two.

Don't forget 2d10! It's not as bell-curved as 3d6, but it's a pretty good alternative. Results 2-20, individual odds going from 1% by increments of +1% to 10%, then down again. More average than d20, more random than 3d6.

[Kurald Galain]

Its not uncommon for a character whos invested a lot into a skill, has a dice pool of 10 or more, rolls them all and one roll you might see 1 success, while the next is 15 (exploding dice). Or worse, you roll your 13 dice, pull three successes, and somebody who's only marginally invested in the same skill rolls 5 or 6 dice and gets the same 3 successes.

What you describe is statistically highly unlikely, and sounds more like a single bad experience than an analysis of the math involved.

Actually in a single die d20 system, an unskilled person is much more likely to beat a trained expert than in a dice pool system, and a d10 dice pool is not all that different from a d6 dice pool. The main drawback of dice pools are that they take longer to resolve.

[Eldan]

Of course, the size of the die says nothing about whether you use dice pools or not. You could have a single d6 system, very like d20, where you roll a single d6+skill against target number. You could also have a system where you roll a pool of d20s and count successes over target number.

[CarpeGuitarrem]

The idea of a "d6 system", if it doesn't refer to the WEG systems, is ridiculously broad. Like has been noted--you can do massive heaps of stuff with d6s that aren't generally done with d20. Though d6 systems often fall into "add up" (in which case they're just like the d20 system, but with more consistent odds) or "dice pool" (which is an entirely different thing--but it's way easier to intuitively adjust than d20 modifiers, and you can do a lot of fun, easy-to-remember things like triggering an effect on a particular number or "roll X, keep Y"). Same generally applies to d10.

Just to compare: look at Mythender, Tenra Bansho Zero, and WEG Star Wars. Three completely different systems, all using d6.

[TypoNinja]

What you describe is statistically highly unlikely, and sounds more like a single bad experience than an analysis of the math involved.

Actually in a single die d20 system, an unskilled person is much more likely to beat a trained expert than in a dice pool system, and a d10 dice pool is not all that different from a d6 dice pool. The main drawback of dice pools are that they take longer to resolve.

You appeared to have completely missed quite a bit of the content of my posts, simply to critique the math.

Its a problem the WoD d10 system has, since success is on an 8 or higher each individual die has low odds, so adding more dice to your pool doesn't significantly help. Combined with the fact that most tasks only need one success, and the investment cost vs increased performance tends to be really poor after around 5 or 6 dice. Since die roll events, even when you roll lots at once, are not interconnected, a larger pool of dice with low individual odds is going to remain relatively poor odds.

The LARP adaptions for WoD solved this by changing it from many dice, to a single card draw (Ace through 10), and instead of extra investment being a larger poll, its a modified roll, and at certain values over the target number you get extra successes. I think the current system is 7 as a target and every 3 over is another success. So if you translate a 10 dice pool vs a 3 dice pool to this system, the 10 dice pool knows hes got 2 sucesses, and can draw for as many as 3 more. The 3 dice pool needs to draw a 4 or higher or fail. You can see how this adaption normalizes results.

The D6 pools system used in the Shadowrun game I played was superior because the target wasn't 8 or better on a d10. 8 or better on a d10 gives poor odds, on the d6 pool the target number was fluid based on the difficulty of the tast, so while two people attempting an easy task might only both need a 2 or better on a d6 if my dice pool is 15 to your 6, its a safe bet I'll preform it far better than you, but both of us can succeed. On the opposite end of the spectrum, a heroic task might have a target of say, 9. Somebody with a pool of 6 dice will probably decline to even try, since he'd need to roll a 6, then a 3 or better. He might not see a 6 at all, let alone roll at least average on the reroll. On the other hand with a 15 dice pool, I can at least give it a shot. This system has the benefit of preserving the ease of trivial tasks, while letting exceptionally skilled characters dominate easy tasks, and making heroic endeavors out of reach of somebody not invested in a skill.

Also, under a d20 system while its possible for an unskiled good roll to trump a skilled bad roll there are still tasks you simply cannot even attempt without many ranks, and some that are doomed to failure. Its possible for a poor crafter to make fullplate on a good roll, but one below average roll and he ruins the work under construction. A skilled crafter who can sit back and take a 10 to make the fullplate on the other hand has turned it into a routine tast. And of course searching for and disabling traps while relying on luck is a fairly suicidal past time compared to letting someone trained do it.

I would dispute that Dice pools take longer to resolve as well. Especially for new players. Roll a fist full of dice, count the number that rolled above the target. Or roll a d20 and wait while the new player tries to remember what gets added to it, and/or ask the DM what circumstance modifiers might apply.

[Rhynn]

Its a problem the WoD d10 system has, since success is on an 8 or higher each individual die has low odds, so adding more dice to your pool doesn't significantly help.

Hmmm.

Odds of 1+ successes on...
... 1 die: 30%
... 2 dice: 51%
... 3 dice: 66%
... 4 dice: 76%
... 5 dice: 83%
... 6 dice: 88%

The odds of X successes require a matrix and a formula that I actually think I have somewhere on a spreadsheet, if you want me to dig it up?

It seemed to me like your point was entirely about the math. The odds of getting X successes on Y dice are all about math. Sometimes, improbable things happen. You can't use individual statistical outliers to judge the system. (You can use the variance of the results, though - for instance, just rolling 1d6 is very "swingy.")


Combined with the fact that most tasks only need one success, and the investment cost vs increased performance tends to be really poor after around 5 or 6 dice.

This appears to be mathematically true - the chance of 1+ successes increases slower and slower, and returns on investment clearly diminish at 5-6 dice.


on the d6 pool the target number was fluid based on the difficulty of the tast

Now this is quite true - changing targets numbers are a whole different beast. (I thought SR4 at least used fixed target numbers, though, and modifiers apply to dice pools?)

The Riddle of Steel, for instance, uses pools of d10s with sliding target numbers (in basic skill tests, your skill sets the target number and your ability score determines the dice pool).


Also, under a d20 system while its possible for an unskiled good roll to trump a skilled bad roll there are still tasks you simply cannot even attempt without many ranks, and some that are doomed to failure.

This is very true, but this has nothing to do with using a d20, and everything to do with the "d20+modifiers compared to target number" mechanic.


I would dispute that Dice pools take longer to resolve as well. Especially for new players. Roll a fist full of dice, count the number that rolled above the target. Or roll a d20 and wait while the new player tries to remember what gets added to it, and/or ask the DM what circumstance modifiers might apply.

Again, you're comparing a specific system (which, in the case of 3.X, definitely does have that issue exactly), but I generally agree that dice pools are not hard to resolve. (Although I'd love some special dice with X blank sides and Y marked sides for some games, so you don't even have to look at numbers, just see how many "success" marks turned up.)

[Kurald Galain]

You appeared to have completely missed quite a bit of the content of my posts, simply to critique the math.

No, I really haven't. Stating that "a dice pool of 10 or more, rolls them all and one roll you might see 1 success, while the next is 15 (exploding dice). Or worse, you roll your 13 dice, pull three successes, and somebody who's only marginally invested in the same skill rolls 5 or 6 dice and gets the same 3 successes" means you have misunderstood the math involved. Sure, it probably happened to you once, but it's extremely hugely vanishingly unlikely.

Especially compared to a simple d20 system, where a character with +8 to a skill regularly defeats someone with +13 to a skill.

[Rhynn]

No, I really haven't. Stating that "a dice pool of 10 or more, rolls them all and one roll you might see 1 success, while the next is 15 (exploding dice). Or worse, you roll your 13 dice, pull three successes, and somebody who's only marginally invested in the same skill rolls 5 or 6 dice and gets the same 3 successes" means you have misunderstood the math involved. Sure, it probably happened to you once, but it's extremely hugely vanishingly unlikely.

Especially compared to a simple d20 system, where a character with +8 to a skill regularly defeats someone with +13 to a skill.

Yeah, seriously. I mean this is math, we can actually determine the real odds (complicated as though they will be with exploding dice and dice pools and counting successes).

[Greyfeld85]

How do you guys feel about FATE's skill checks? That system seems to put a much heavier focus on skill ranks and situational modifiers. Or the Marvel Universe RPG, which eschews dice altogether in favor of a resource management system that allows you to allocate your effort to different actions during a scene?

[TypoNinja]

I guess I didn't spell it out enough. I thought bringing up the LARP solution would make it clear.

The d10 system is entirely chance based, and the only thing you can do as a player is throw more chance on the pile.

The d20 system is about getting your modifiers high enough that chance doesn't matter.

And the d6 system has sort of split the difference. Its still a target roll, but the roll can be influeced by either the players or the circumstances. Sniping that man in a 3 +1 for distance, +1 for the head. You've got a good scope though so -1. You need 4's. roll some d6's.

The points I was going for wasn't about the math, but about how you interact with it. In the d10 system spending more XP buying up a skill stops being worth it, especially when your 5th dot in a skill costs similar to the first three in a new skill. Regardless of the math players can start feeling like specializing stops being worth it, which is bad design in my opinion.

In the d20 system, there is a wide swing, but its easy to lean on it, and standardized DC's mean there are soft caps all over the place where you can be satisfied with your skill at a task and never fail it again for the most part. (taking a 10). You can easily establish your baseline skill at something, and stay good at it. You can also know if a task is even worth trying at a glance. Climbing a cliff? What's your climb check? Nope, can't roll that high, don't even bother trying.

The d6 system has split the difference, were back to target numbers on a collection of dice but those are fluid depending on what you want to do, increased skill at a task is back to more dice instead of a bigger modifier, but now the target is allowed to move depending on the difficulty of the task, and if your player has prepared you have options to bring the target number back in your favor. The flat target is also a 4 on a 6 sider instead of an 8 or a 10 sider, and number of successes matters more that the d10 system. So its easier to pull at least one success out, but only one is likely to only be partially effective.

Obviously I can only speak of the specifies of the systems I'm actually familiar with, which is why I bothered specifying each in my first post.

But again, my point wasn't the math directly, but how you interact with the system, most games have some kind of probability at their core to make it interactive and have the outcome in doubt, but I think they way they go about it, and how its used are more important than numbers behind it.

[tasw]

I've played most of the different dice systems out there and D6's are my least favorite.

The 3d6 bellcurve method reduces randomnes and increases the effect of skill. Which some people like. But for myself I want MORE randomness so this goes against my game goals.

The D6 pool vs target number isnt as bad but it still reduces randomness to a point where I find the D10 dice pool systems to be superior.

Although I agree with what others have said that a way to adjust the target # would be good as well. Old world of darkness rules had that as I recall.

[Kurald Galain]

I guess I didn't spell it out enough. I thought bringing up the LARP solution would make it clear.

The d10 system is entirely chance based, and the only thing you can do as a player is throw more chance on the pile.

The d20 system is about getting your modifiers high enough that chance doesn't matter.
No, that wasn't clear from your earlier post.

It's also still incorrect. In almost none of the various d20 systems is it reliably possible to make modifiers so high that you don't have to roll. In 2E it doesn't exist, in 4E it's mostly avoided (except for easy DCs) and 5E has the explicit goal of making this impossible. Only in 3E can you do that, and then only for certain skillmonkey builds at reasonably high level.

So comparing the two, it is clear that in a d10 dice pool system it is much easier to get your character to the level where you can reliably pass a check 90% or 95% of the time. In d20, failulre rates of 20% - 30% are common even for experts (and intended by the designers, in case of 4E and 5E).

[Rhynn]

So comparing the two, it is clear that in a d10 dice pool system it is much easier to get your character to the level where you can reliably pass a check 90% or 95% of the time. In d20, failulre rates of 20% - 30% are common even for experts (and intended by the designers, in case of 4E and 5E).

Total agreement. In 3.X, some skill checks with set DCs become trivial (0% chance of failure, either on taking 10 or on rolling 1), but many skill checks are opposed, and it does require optimizing. It is not regular or constant, and certainly hasn't been at my table (we started 3E in 2000).

And yeah, with the WoD dice pool (8+ succeeds), 7 dice get you a 91.7% chance of 1+ successes, 8 dice get 94.2% chance, 9 dice get 96%... that sounds like a fairly low number of dice? Isn't it "ability core + skill + modifiers" in WoD?

You can directly and easily calculate the actual chance of 1+ successes in that system: it's P(X) = 1-(0.7^Y), where Y is the number of dice in the pool being rolled.

Boring math off-topic ramble:
You can also calculate the individual odds for every "X number of successes", although I'll be doggone if I can remember how to write out that formula (and probably couldn't using this forum anyway since it'd require the symbol for combinations). However, the OpenOffice (Excel too?) formula is...

=(COMBIN;Y;X)*(POWER(P1;X))*(POWER(P2;Y-X))

X is the number of successes we're calculating odds for; Y is the number of dice being rolled; P1 is the probability of getting a success, and P2 is the complementary event, e.g. 1-P1 (the odds of not getting a success).

This should be right, but it's always possible I messed up; I'm working off a spreadsheet I made when I actually remembered how to calculate this stuff.

So, rolling 12 d10s at 30% success odds, I can see that the chance of 3 successes is 23.97%, for instance.

Of course, this does not account for exploding dice, which always makes calculations much harder, but no doubt there's a way to account for that too.

So yes, in D&D 3.X specifically, there are cases in which there is a literal 0% chance of failure. Some people consider that a strength, some a weakness, but it is not the basic state of things even in that system. Mostly, you're always going to have a chance of success you can calculate (easily, in 3.X). But the odds with dice pools are perfectly calculable and predictable.

I'm using 3d6+ability+skill+modifiers vs. target number in my Fuzion cyberpunk hack. It gets me normally distributed results, and by controlling the range of abilities and skills, and the target numbers, I can set the chances of success appropriately for various tasks. I get a normal distribution, avoid the problem of "0% chance to fail" (3 an autofailure, the smallest failure chance is 0.46%), and emphasize that the main use of high skills is to have a much better chance to succeed at difficult tasks, without making average skills bad at average tasks.