Revisiting this core mechanic: normal 2d10, cautious 3d6, reckless 1d20

[Greywander]

I was thinking about core resolution mechanics and I remembered coming up with this one a while back. The basic premise is that it's a die roll + modifiers, so very similar to D&D, but the default roll is 2d10. However, the player has the option to instead make a "cautious" roll, rolling 3d6 instead of 2d10, or to make a "reckless" roll, rolling 1d20. The reason I think this mechanic is neat is because it gives the players a little more agency over how they roll, and different rolls are more tactically advantageous in different situations. If you only need a middling roll, then it's best to roll cautiously, but if you need a high roll, then reckless is best. The players don't usually know the exact target numbers they're aiming for, so it's up to them to guess which roll is the optimal one, but choosing wrong isn't an automatic fail, just a slight decrease in their odds of success. I like how this turns the swinginess of the d20 from a bug into a feature; if you're outmatched then playing it safe is a losing proposition, instead you should throw caution to the wind and gamble on the d20's same odds of giving a 20 as a 10.

At the time, I wasn't completely satisfied with this mechanic, and one of the main reasons was that it seemed like there was never a reason to roll normally; if the odds favored you then it was better to roll cautiously, and if the odds were against you then it was better to roll recklessly. Having actually looked at the math, though, I've discovered that this isn't actually true. I punched the formulas into AnyDice and found that if your target number is 9 or less then the cautious roll is best; if the target number is 10, 11, or 12, a normal roll is best; and if the target number is 13 or higher then a reckless roll is best. Granted, the sweet spot for a normal roll is pretty narrow, but it's also right in the middle (average roll on 2d10 is 11), so there's an expectation that "level appropriate" challenges will fall into that range with more frequency than you'd expect.

Something interesting is that if we add a mechanic similar to 5e's advantage/disadvantage system, where we roll one additional die and either drop the lowest (advantage) or the highest (disadvantage), then the optimal range of target numbers is almost entirely unchanged. I still don't quite understand the full interaction; my gut says that rolling recklessly will benefit more from advantage, while rolling cautiously will help to mitigate disadvantage, but given that the optimal target number range hasn't changed (mostly), I'm no longer certain that that's the case.

Something else interesting that we could do is play with critical successes and fumbles. Let's say a critical success occurs on a 19 or 20, and a fumble on a 1 or 2. If we roll cautiously, there is zero chance of getting either result. This means if you want a critical success then rolling cautiously isn't an option, you have to take some risk, but also that you can eliminate the possibility of a fumble. If you roll normally, you can't roll a 1 on 2d10, and a 2 will be pretty rare, only 1%. On the other hand, you can roll both 19 and 20, giving you significantly higher odds of getting a critical success, 3%. If you roll recklessly, those odds suddenly jump to 10% either way. So this is another thing that makes rolling normally interesting; it allows for both critical successes and fumbles, but the odds of a critical success are significantly higher than those of a fumble. This means if you stick mostly to rolling normally, you'll see about three times as many critical successes as fumbles.

What do you think of this mechanic? I don't think it would add much overhead to the game, unlike some more complex resolution mechanics. My hope is that a mechanic like this that offers some agency to the player will help them to get more invested in the game. There can be a real sense of desperation as a player throws a d20 as a Hail Mary and either strikes out or gets that high roll they needed to overcome a challenge that was out of their league. Likewise, there can be a sense of escalation as e.g. a player repeated fails to pick a lock cautiously and steps up to rolling normally, then recklessly as a guard is approaching. It completely changes the bland "I guess I'll just try again" to "should I change how I'm rolling this time?"

One thing is that I might need to tighten up the target numbers and the bonuses to rolls compared to D&D 5e. ...Or not. Since you still have the option to roll a d20, and the best time to do so is when you really need a high roll, it might make sense to keep a similar range as D&D 5e. Likewise, the cautious roll benefits the most when you can accrue enough bonuses to comfortably place the needed roll in the lower half of possible results.

[kosh49]

I agree with your general assessment about the ranges where each die roll is better: for 9 or less use cautious, for 10-12 use normal, for 13+ use reckless. I also agree that advantage and disadvantage do not change those ranges.

As a note, if you assume that all target numbers are equally likely, a player or monster who always uses the optimal dice choice will succeed about 6.8% more often with this than one who always uses 1d20. Advantage reduces that benefit to 2.0% and disadvantage raises that benefit to 9.7%. Obviously, if higher target numbers are more likely the impact is smaller and if lower target numbers are more likely the impact is larger.

One minor note, on rolls where a rogue's reliable talent applies, normal rolling is better for a target of 11 or 12 (except oddly for a target of 12 when neither advantage nor disadvantage apply) and everywhere reckless is better for higher values, and all methods are the same for lower values. So the rogue should always roll reckless far an ability check unless they have a way to know that the number that they need on the roll is 11 (or maybe 12).

The general strategy for repeated events (like to hits against a large number of identical opponents) is to start at normal and shift to cautious or reckless once feedback has made it clear if one of those is advantageous.

If you want to find a group to playtest this with you it might be interesting to try. I am not sure how much this will shift things. You might want to be cautious about handing out magic items that give to hit bonuses though. If the party consistently gets to shift to cautious rolling because they do not need to use riskier rolling that could result in power boost, and how large that will be is hard to guess.

The critical fumbles part is potentially an issue for the martial/spell caster balance. Depending on spell selection, spell caster may not need to make many rolls. This will mean that unless the overall increase in successes and critical hits exceeds the detriments caused by the fumbles things could become a problem. That is a general issue with fumbles though, and is not specific to your proposal.

[Greywander]

I don't know that I'd want to simply patch this into D&D 5e, so things like Reliable Talent or spellcasters not rolling might not be relevant. I think it would make more sense to build a new system from the ground up with using a system like this in mind.

That said, this does raise an interesting question. In vanilla D&D 5e, you can rework how saving throws work so that the attacker is making an attack roll against a static save DC, instead of the defender being the one to roll the saving throw. You can do it so that the math is identical, so all that's really changed is how it's presented. But if if a mechanic like this one was being used, then it would suddenly matter a lot whether the attacker or defender was rolling, since they would be the ones with the agency over what type of roll to use. It would probably make more sense to be an opposed roll then.

Speaking of, opposed rolls would be really interesting. If one contestant has a slightly higher bonus, they might want to roll cautiously, but if their opponent rolls recklessly it could put them at a disadvantage.

[Coeruleum]

I would use 2d8 for normal to balance cautious being 3d6, since otherwise cautious will just be nearly universally worse.

[Greywander]

I would use 2d8 for normal to balance cautious being 3d6, since otherwise cautious will just be nearly universally worse.

Average roll for 3d6 is 10.5.
Average roll for 2d10 is 11.
Average roll for 2d8 is 9.

If I did 2d8 as the normal roll, there would never be a time where it was more optimal to use than 3d6. As it is, 2d10 has a very narrow range where it's only better when you need to roll a 10, 11, or 12, whereas 3d6 is better if you're trying to roll a 9 or less (and 1d20 is better for 13 and higher).

[kosh49]

I would use 2d8 for normal to balance cautious being 3d6, since otherwise cautious will just be nearly universally worse.

2d10 is actually a good mid-point between 3d6 and 1d20. 2d8 completely fails to be useful. If you take the target number (AC for an attack, DC for a save or ability check) and subtract your bonus, you get the minimum number needed to succeed for the raw value you roll on your die or dice. The following chart shows your chance of rolling a particular value or higher with each of the 4 strategies mentioned:

Value	1d20	2d10	3d6	2d8
1	100.00%	100.00%	100.00%	100.00%
2	95.00%	100.00%	100.00%	100.00%
3	90.00%	99.00%	100.00%	98.44%
4	85.00%	97.00%	99.54%	95.31%
5	80.00%	94.00%	98.15%	90.63%
6	75.00%	90.00%	95.37%	84.38%
7	70.00%	85.00%	90.74%	76.56%
8	65.00%	79.00%	83.80%	67.19%
9	60.00%	72.00%	74.07%	56.25%
10	55.00%	64.00%	62.50%	43.75%
11	50.00%	55.00%	50.00%	32.81%
12	45.00%	45.00%	37.50%	23.44%
13	40.00%	36.00%	25.93%	15.63%
14	35.00%	28.00%	16.20%	9.38%
15	30.00%	21.00%	9.26%	4.69%
16	25.00%	15.00%	4.63%	1.56%
17	20.00%	10.00%	1.85%	0.00%
18	15.00%	6.00%	0.46%	0.00%
19	10.00%	3.00%	0.00%	0.00%
20	5.00%	1.00%	0.00%	0.00%

As you can see with this chart, if the number you need is 13 or more, 1d20 gives a better chance of success than any of the other strategies listed. If the number needed is 12, 1d20 and 2d10 give equal chances of success, and are better than the other two strategies. If the number needed is 10 or 11, 2d10 gives a better chance of success than any of the other strategies. And if the number needed is 9 or less, 3d6 gives a better chance of success than any of the other strategies.

On the other hand, 2d8 is worse than all 3 of the other strategies if the number needed is 9 or more. In the range of 2-8 it is better than 1d20, but it is still worse than both 2d10 and 3d6.

I did look at the advantage and disadvantage numbers for 1d20, 2d10, and 3d6 (defined as roll one extra die and drop the lowest for advantage or the highest for disadvantage). While the percentages of success changed, the only change in the cutoffs was that at 12, 2d10 become better than 1d20 regardless of whether you were rolling with advantage or disadvantage. I did not look at the effect of advantage and disadvantage on 2d8, but given how miserable it was without them and how little effect they had and the relative value of the other strategies, it does not really seem necessary.

[Rilmani]

I think this setup has some potential, though I’m not sure what changes should be implemented to enhance the player choice aspect from the three options. Focusing on critical results (1, 2, 19, 20) is one option, but I think it needs to be expanded upon.

You can use this setup without making it revolve around player choice. A lot of people aren’t going interested in the predictive aspect of trying to imagine the DC they’re rolling against. It will pretty much always run from 10 to 15, unless you have a DM who requires unnecessary rolls- “roll a strength check to pick up the wooden chair, mister Barbarian,” which would have a DC of 2 to 4 if the GM actually stuck to the rules rather than trying to punish a character for acting out. Back to my point- imagine a gestalt character. A Fighter/Wizard. Traditionally these classes would be treated equally. But they don’t have to be. If a new ruleset forced a player to select a Primary class and a Secondary class, then each class could make use of a different set of rolls. A Primary Fighter could be locked into 1d20 rolls for attacks and physical saving throws, with 3d6 rolls for wizardly attacks and defensive actions. With this setup, the character can only critically succeed/hit on Fighter actions and there is a larger range of targets they just can’t reliably hit with Wizardly attacks.


This is an example of how turning Saving Throws from defensive “actions” into offensive actions made by the source/opponent/attacker could be preferable. Now, the fact that wizards and other spellcasters have a wide variety of abilities that don’t require an attack roll or “saving-attack” would require multiple adjustments. But this is just one example. Another option would be to make a character Rank their capabilities like in Shadowrun’s character creation. They could pick 1d20 for their subclass, 2d10 for their class, and 3d6 for their racial abilities. Obviously that this point I’m significantly diverging from base 5e rules, to even consider giving racial abilities and class abilities equal standing.

[Greywander]

2d10 is actually a good mid-point between 3d6 and 1d20.

I did look at the advantage and disadvantage numbers for 1d20, 2d10, and 3d6 (defined as roll one extra die and drop the lowest for advantage or the highest for disadvantage). While the percentages of success changed, the only change in the cutoffs was that at 12, 2d10 become better than 1d20 regardless of whether you were rolling with advantage or disadvantage.

Yup, I did my homework on this. I even saw that 1d20 and 2d10 have the same odds of rolling exactly 12, but 2d10 beats 1d20 with either advantage or disadvantage, which is why I said 2d10 was better if the target number of 12 (because 2d10 is optimal whether or not you have advantage or disadvantage). If you get into double advantage or disadvantage (roll two extra dice and drop the two highest/lowest), the numbers fudge around a bit more.

I think this setup has some potential, though I’m not sure what changes should be implemented to enhance the player choice aspect from the three options. Focusing on critical results (1, 2, 19, 20) is one option, but I think it needs to be expanded upon.

There's a tricky line to walk between making a mechanic interesting and engaging and making it needlessly complex, difficult to understand, and time consuming. Anything I can do to make the mechanic more interesting and engaging without eating up too much table time or adding a disproportionate amount of complexity would certainly be welcome. The best mechanics, especially for something like a core mechanic that every player has to engage with, are simple and elegant.

You can use this setup without making it revolve around player choice. A lot of people aren’t going interested in the predictive aspect of trying to imagine the DC they’re rolling against. It will pretty much always run from 10 to 15,

Sure, players aren't going to be able to predict the target numbers they're rolling for. This is where presentation matters; by specifically using the names "reckless" and "cautious", I can help players intuitively understand when they need to use either type of roll. They don't need to know the exact numbers they're aiming for, only whether the situation calls for being cautious or reckless. This can also lean into how a character is roleplayed.

A Primary Fighter could be locked into 1d20 rolls for attacks and physical saving throws, with 3d6 rolls for wizardly attacks and defensive actions.

It seems like maybe you're not understanding how the mechanic works. 3d6 is far more consistent, whereas 1d20 is much more swingy. What I'd expect is for the fighter to get a higher bonus to things like weapon attacks. Because the fighter has a higher bonus, they don't need to roll as high, and therefore can comfortably roll 3d6 instead of 1d20 against most enemies, leading to them hitting much more consistently. A wizard, who doesn't get a bonus to weapon attacks, has to roll recklessly in order to get the best chance of hitting things with a weapon, but while they can hit, they're much less reliable.

The point is to try and stack bonuses towards the things you want your character to be able to succeed at consistently, and then you can just roll cautiously every time and pass the check most of the time. The time to roll recklessly is when you're doing something you don't have good bonuses for. Occasionally, you'll run into an extra difficult challenge (e.g. fighting a dragon), and even the person with bonuses may need to roll recklessly, but they'll still have much better odds than the person without bonuses.

Another option would be to make a character Rank their capabilities like in Shadowrun’s character creation. They could pick 1d20 for their subclass, 2d10 for their class, and 3d6 for their racial abilities.

Wouldn't it just be easier to use the same rules across all three? Even if you just make everything a d20 roll like in D&D, that's better than having to keep track of three different types of abilities that each use different rolls. Your suggestion seems like it's getting rid of all the benefits of the mechanic in the OP while simultaneously increasing the complexity. Sure, they don't have to choose every time they roll, but surely choosing every roll is easier than remembering which abilities use which rolls?

The whole point is to give players the choice of how to roll. If I'm not going to give them that choice, then I'm just going to choose a type of die roll, e.g. 2d10, and build the entire system around using that for everything. Using different dice for different things is a way that leads to madness. D&D only gets away with it for damage rolls because it is limited almost exclusively to damage rolls. Also, legacy reasons.

Obviously that this point I’m significantly diverging from base 5e rules, to even consider giving racial abilities and class abilities equal standing.

Well, as I said in an earlier post, the intent was not to patch this into D&D, though you might be able to do so without too much trouble. If I implement this into an original system, it will probably be classless anyway.

[Satinavian]

After playing years with a system that has something similar, i have to say it works best with some kind of degree of success/failure system. Otherwise there is always the optimal strategy once you know the DC. But with degree of success you can balance greater reward vs greater risk or avoid both.

[Greywander]

Yeah, multiple target numbers with differing effects seems like a good way to build on this system and add more nuance. Now it's no longer about hitting a single target number, but whether you're content to roll cautiously and settle for a mediocre result or roll recklessly to try and get a better result. Even something as simple as "every 5 higher than the target number gives X benefit" would probably be sufficient.

I recall looking into an armor revamp for 5e, which I could probably adapt in some form to an original system, where armor has three stats: deflection, coverage, and absorption. Deflection is basically a bonus to AC, and represents a glancing blow that connects but doesn't do any real damage. Coverage adds to your AC to create a secondary DC; if you roll high enough to hit, but not to beat the target's coverage, this represents a direct hit on the armor, reducing the damage. Absorption is how much the damage gets reduced by if hit within the range of your coverage. If you beat a target's coverage, then that represents hitting them in an unarmored spot, so they take full damage.

With a system like that, there are already multiple target numbers built in, with predetermined penalties for failing to hit the higher target number. It would be tricky to expand that into a universal system that works for any kind of skill check and not just attacks, but I'm sure something could be worked out. I'm a fan of the Rule of Three, so it might make sense for every skill check to have three different DCs built into it, each offering a different degree of success. This could even replace critical successes and failures as a result of the natural die roll (i.e. instead of getting a critical hit on a 19 or 20, you'd get a critical hit if you pass the highest of the three DCs).

[Baron Faey]

First, I really like this idea, and played around with something similar for a homebrew system I'm working on. Due to difficulty with getting the math to be both quick and balanced, I decided to make the risk/reward choice be on the "damage" end rather than the accuracy. I'm calling it "raising the stakes", and it changes the cost or reward rather than the chance of success or failure. For example, in a fight, they can rush in agressively to deal extra damage but the target can take advantage of that and deal extra damage to them as well, or when trying to grow their bussiness they can make a risky investment for the chance of greater money, but lose more if the venture fails.

Returning to D&D, in an interplanar campaign I ran a few years back, each plane had its own rules, and for a lawful plane one of the rules was that all d20 became 2d10s. This helped create the mood I wanted for the adventure (more regular rolls, few critical hits, no fumbles), but also slowed down play.
I was also planning on having a chaotic plane adventure where d20s became 1d10x2 (treating 1s and 10s as 1s and 20s), but two of the 4 players moved away before we got there.

For my Luckspinner class I played around with d20 modifying abilities. The main challenge I found was that not all D20 rolls are made equal, and some classes make more rolls than others and use them differently (which has been discussed above).

I'll also second Satinavian's comment about degrees of failure/success.

[animorte]

Value	1d20	2d10	3d6	2d8
1	100.00%	100.00%	100.00%	100.00%
2	95.00%	100.00%	100.00%	100.00%
3	90.00%	99.00%	100.00%	98.44%
4	85.00%	97.00%	99.54%	95.31%
5	80.00%	94.00%	98.15%	90.63%
6	75.00%	90.00%	95.37%	84.38%
7	70.00%	85.00%	90.74%	76.56%
8	65.00%	79.00%	83.80%	67.19%
9	60.00%	72.00%	74.07%	56.25%
10	55.00%	64.00%	62.50%	43.75%
11	50.00%	55.00%	50.00%	32.81%
12	45.00%	45.00%	37.50%	23.44%
13	40.00%	36.00%	25.93%	15.63%
14	35.00%	28.00%	16.20%	9.38%
15	30.00%	21.00%	9.26%	4.69%
16	25.00%	15.00%	4.63%	1.56%
17	20.00%	10.00%	1.85%	0.00%
18	15.00%	6.00%	0.46%	0.00%
19	10.00%	3.00%	0.00%	0.00%
20	5.00%	1.00%	0.00%	0.00%

Thanks for the visual.

Yeah, multiple target numbers with differing effects seems like a good way to build on this system and add more nuance. Now it's no longer about hitting a single target number, but whether you're content to roll cautiously and settle for a mediocre result or roll recklessly to try and get a better result.

Precisely, this has proven quite sufficient for me. More could benefit from using this method whether strictly using d20 or not.

[Rilmani]

It seems like maybe you're not understanding how the mechanic works. 3d6 is far more consistent, whereas 1d20 is much more swingy. What I'd expect is for the fighter to get a higher bonus to things like weapon attacks. Because the fighter has a higher bonus, they don't need to roll as high, and therefore can comfortably roll 3d6 instead of 1d20 against most enemies, leading to them hitting much more consistently. A wizard, who doesn't get a bonus to weapon attacks, has to roll recklessly in order to get the best chance of hitting things with a weapon, but while they can hit, they're much less reliable.

The point is to try and stack bonuses towards the things you want your character to be able to succeed at consistently, and then you can just roll cautiously every time and pass the check most of the time. The time to roll recklessly is when you're doing something you don't have good bonuses for. Occasionally, you'll run into an extra difficult challenge (e.g. fighting a dragon), and even the person with bonuses may need to roll recklessly, but they'll still have much better odds than the person without bonuses.

Fair; I explained myself poorly and didn’t stick to feedback for your main proposal. Since the start I had an exaggerated picture in mind for the consequences of critical successes and failures, an order of magnitude more influential than in typical games. In order to justify the “reckless” part of the reckless option.

Wouldn't it just be easier to use the same rules across all three? Even if you just make everything a d20 roll like in D&D, that's better than having to keep track of three different types of abilities that each use different rolls. Your suggestion seems like it's getting rid of all the benefits of the mechanic in the OP while simultaneously increasing the complexity. Sure, they don't have to choose every time they roll, but surely choosing every roll is easier than remembering which abilities use which rolls?

The whole point is to give players the choice of how to roll. If I'm not going to give them that choice, then I'm just going to choose a type of die roll, e.g. 2d10, and build the entire system around using that for everything. Using different dice for different things is a way that leads to madness. D&D only gets away with it for damage rolls because it is limited almost exclusively to damage rolls. Also, legacy reasons.


Well, as I said in an earlier post, the intent was not to patch this into D&D, though you might be able to do so without too much trouble. If I implement this into an original system, it will probably be classless anyway.

Right… if I try and stick to the actual benefits of this system… Let’s imagine a setup in which enemies have reactions, but they don’t have actual turns. A blackguard might have reactions against enemy spellcasters and enemy swordsmen. When a creature fails to affect the blackguard (a missed attack), the blackguard uses a reaction to deal damage. If you miss his AC by 3, he deals 3d4 damage. If you miss his AC by 8, he deals 8d4 damage. The Blackguard could be designed with a low AC or a high AC. I think generally this sort of enemy would use a large damage dice if they have low AC and they’d use d4s if they have a high AC.

This is a pretty clear example of why choosing the right roll type is important. I suggested that some players aren’t interested in trying to predict a target number. Your rebuttal about framing, combined with the fact that players are likely to suss out the AC of an individual creature/boss like this within two rounds, makes me think I was wrong. With that dismissed, we have the following scenario:
Each character is cautious, balancing whether they should take X action with the idea that the enemy can respond and punish them for it. Each player is considering their roll types carefully- can they afford a terrible roll? Will they die in a moment if they don’t roll an outstanding success?

So. Does the skeleton of this fight make for fun gameplay? My “the blackguard only takes reactions instead of actions” isn’t completely unusual for a tabletop rpg, but it is a weird scenario. All the same, it does outline the dangers of a failed roll VERY clearly. Whether or not a less obvious penalty for failure will click with players is hard to say. And of course the bonuses a character will have for their rolls will play a major role as well.

Let’s modify the scenario. The cost of a failed roll is very obvious- lost HP. Do we want degrees of success like we have degrees of failure? Should success be a static state (you roll your normal damage dice) or should exceptional successes have the ability to turn the tide of battle (for everyone 1 by which you exceed the target number, roll one additional damage die)? Or perhaps you could recover a resource (Hp, stamina…) based on how successful you are.


Each of these are options. More complicated options will slow down gameplay, but complexity is a very useful way to hook people into a game. When done well. I think the value of this scenario depends entirely on what the average fight looks like. If massive groups of enemies are more common, then the party of players aren’t likely to treat each enemy as their own puzzle. If you are in a hectic spaceship war dealing with around 4 repeating enemy types in small groups across a whole campaign, then you can afford to give each enemy’s defenses some depth. Another factor- some rulesets are quite terrible to build enemies with. Hours of work, rather than “plug and play.” Some depth to an enemy’s defenses, but not too much…