Along with a lot of other people on the boards I've been thinking about ways to represent skills in 5e that can be used to make training significant and skill results adaptable while still keeping the math of the d20 in a reasonable range. I think I've come up with a way that can do this without necessarily increasing complexity for those who don't want it, or at least doing so very infrequently. Furthermore, the balance implications for using or ignoring it should generally be fairly minimal, which I consider a virtue for a module. This idea is a mix of the traditional binary result of d20 skill checks with some aspects of success-based systems, but I think it combines them in such a way that the game will still feel very much like D&D.
Suppose that there are 4 corresponding levels of difficulty and training in the game that represent large qualitative differences in skills. For simplicity call these Untrained, Competent, Expert, and Master. When referring to difficulty they represent roughly the level of training required to succeed at such a check regularly. Every skill check has a single DC and a result corresponding to at least one of the 4 difficulty ratings. The DC of the check is a measure of how difficult it is for a character to leverage whatever level of training they have, from untrained to master.
The general form of a check involves one or more d20 rolls, each gaining a numerical bonus to the roll in the usual way, with each roll that meets the task's DC counting as a single success. The number of successes required to "succeed" at a given level of difficulty is listed here:
Many checks allow multiple levels of success. A success at a given difficulty automatically grants all the effects of success at the lower difficulty as well. For example, suppose a creature rolls 3 dice on a Diplomacy check that has outcomes for Competent-, Expert-, and Master-difficulty. If 2 dice succeed the results of both the Competent and Expert difficulties occur.
The effect of training is shown below. Unless otherwise specified, all skill checks are rolled with a single d20.
Untrained creatures can only gain Untrained-difficulty results on a check, and may not even attempt checks without an Untrained-difficulty result. Trained creatures may attempt any check and gain the benefit of any result, even those above their current level of training
In addition to training, additional dice can also be obtained by taking extra time, using the Aid Another action, or by various magical means.
Skill Tricks are ways for highly trained creatures to spend "extra" successes. I'm cribbing the name from 3.5's Complete Scoundrel, where I thought they were promising but a little underbaked.
Here is an example of this system which demonstrates essentially every aspect of it. Note that this is probably the most complicated version possible, not what I'd consider (or want) to be a typical check. As I envision it skill training above Competent should be rare, and should skill checks with a minimum level of difficulty above Competent. Consider a 1st level character, a student, that stumbles across a puzzling ancient scroll. He is no slouch, with 20 Int and Competent in the Decipher Script skill, giving him a +10 bonus on the check (for the sake of argument). For this scroll the possible results might look like this:
This student wants to impress his mentor, so he attempts a full translation. Since he is Competent he must spend a full week on this check (1 hour->1 day->1 week) to get the two extra dice that give him any chance at all. He has, roughly, a 1.5% chance of getting all three successes, a 14% chance of getting exactly two successes, and a 42% chance for both 1 success and no successes. Fortune doesn't exactly smile, and he obtains only a single success on his check. Not what he'd hoped, but he's still plenty excited because the scroll seems to be about the location of some item of great power. He brings it to his mentor, who is a Master in such things. The mentor also has 20 Int, so also has a +10 bonus on the check. When his student tells him of his find the mentor's curiosity is piqued, and they spend an hour going over the scroll together.
The student attempts to aid the mentor on this check. Since the student is Competent his extra d20 will only applies to the Competent and Expert levels of the check. His aid might be unnecessary, therefore, since the mentor needs only a single success to fully translate the scroll in an hour. The mentor rolls a single d20 to see if he succeeds at the Master level. A 3. It happens to the best of us. He then rolls another d20 (since he rolls 2d20 for Expert-difficulty checks and he already rolled the first one trying for the Master level) and the student rolls a d20 as well to represent aid. They have, on these two rolls, about a 6% chance of two successes, a 38% chance of 1 success, and a 56% chance of no successes. Finally luck strikes, with a pair of 17s. Having an extra success thanks to his student's aid, the mentor uses the "New Insight" Skill Trick, which lets him reroll a single previous failure if he spends the maximum amount of time re-deciphering this text. The mentor and student rejoice as they read about an ancient order that secreted away a powerful artifact known as a Nether scroll. The author seems to have some idea where this artifact may be...
Over the next week the master carefully goes over the scroll, just as his student had done earlier. Normally this would grant only 2 additonal d20s because he already rolled once at the master-difficulty level that first day. However, due to New Insight he can roll all three, and so the probabilities are the same as his student a week or two earlier. Of course, he only needs a single success to decipher the scroll, which has a 58% chance of getting. His first roll fails, but the last two both succeed. Since he again has an additional success (thank you New Insight!) he uses his Flawless Memory Skill Trick, which grants him perfect recall of the original text as well as his translated version.
In the coming months he and his student publish the non-sensitive elements of their great discovery in a prestigious journal, causing jealousy among their peers, and the following year are awarded a travel grant to seek out their prize. After only two weeks they unceremoniously travel into the gently crushing fists of an Adamantine Golem because the Nether Scrolls are not to be trifled with. The End.
Here are some potential examples of Skill Tricks that illustrate how they might work. I've given no thought to balance.
Finally, I think this system is largely orthogonal to several other interesting aspects of the skill system which would need to be defined in the released game, and would work well with any of them. These include the specific skill list (if there even is a predefined skill list) and whether or not skills are tied to particular ability scores. All the skills listed above are only meant to be illustrative. In addition, I haven't addressed at all how training may or may not be limited by level, although I think that is an important issue.
-------
I wrote a lot of analysis and it turned into a bit of an essay so I'm putting it into a spoiler block. The meat is all above, and I think the basic idea is clear, but if you want to delve further read on.
[SBLOCK]
In my view the default assumption of the game should be that training beyond Untrained/Competent is rare, and so are checks where the lowest levels of success have Expert or Master difficulty. That way the game isn't constantly being slowed down by very difficult checks, unless the PCs specifically pursue such things. The presence of multiple levels of success, extra dice on many checks that require only a single success, and Skill tricks is already pretty awesome. Not every lock needs to be of Master-difficulty, even at high levels, for the rogue to show off. That being the case, I would not make training beyond Competent a default part of character advancement. A high level character that wants to be a true master should strive for it, and I think the benefits outlined above are worth expending character resources to obtain. If there is a numerical bonus to trained skills just for leveling up it should be modest, something like +1/5 levels at most. That way the DCs can be kept in a reasonable even as some advancement is observed.
First, I want to talk about playability. It's clear that this is a system where most of the time a skill check is still just a single d20. Much of the extra complexity of the system only shows up when the player opts-in. Most players, I think, will not object to rolling another d20 or two when they get to use their awesomest skills. In fact, I suspect they will relish it. Or when multiple characters collaborate each one contributes at most one die, and quite possibly a die they didn't roll on their own turn in order to aid a skill check they regard to be of value. So the number of dice thrown isn't much increased, and again the players chose to make it happen. Likewise when a character takes extra time making a check, the slow-down at the table isn't very large and the player has a goal in mind, although I admit it is slower than taking 20. This tendency to need only a single success with training also sidesteps an annoying tendency of some success-based systems. Namely, when one needs 3 successes but gets exactly 2. What a heartbreak! That sort of thing should only happen in this system when the players are beyond their normal capabilities, and they know it. It might still smart, but it just won't happen that often.
Another aspect of playability is when a check has multiple levels of difficulty. A master character could, in principle, need to see if they rolled at least one success (of three) for Competent, at least one success (of two) for Expert, or at least one success (or one) for master. However, that looks a lot slower than it really is because a success at Master, for example, grants all the lower results as well. So the process on this check is this: Roll a single die. If it succeeds you get everything. If it fails, roll and additional die. If it succeeds you get Competent and Expert levels of success. If it failed, roll an additional die. If it succeeds you get the Competent level of success. It is easy to see that this is equivalent to rolling for all three checks at once, where the dice that correspond to each check have been labeled beforehand, without actually requiring the player to think about it at all. Just keep rolling until you succeed or run out of dice. Of course, if you don't want the player to know how many dice to roll (say there is a Master-difficulty result but you don't want the player to know that for metagame reasons) this is tougher to do. One could simply make the player always role a single d20 at a time, or have them select a die randomly after the fact, but either will slow down play a bit. A better solution might simply be to roll behind the screen. I think such cases would be rather rare, however.
From the DM side of things I think it's also more playable overall. This system allows a real easy rule-of-thumb to support multiple results, and in fact the DM only needs to select a single DC to do so. In prior editions, if I wanted to do something similar I needed to make up multiple DCs. Usually this was easy (+5 or +10) but sometimes it required more thought. Selecting the DC in the first place may take a bit of practice because we're so used to making the numbers skyrocket, but I think people would get the hang of it quickly. The main idea is this: Whatever is the primary result you have in mind, figure out what level of training should be required to do that regularly (i.e. with a single d20), and then set the DC for a person with that level of training. If the range of DCs is fairly compressed the numbers should be quite moderate, and the probabilities to consider are just those of the single d20 with its familiar and convenient uniform distribution.
Another benefit to playability for both PCs and DMs comes from Skill Tricks. These can relieve from the DM the pressure to make sure multiple successes always have some unique result, since this is in the hands of the player. And for the player it gives them a little bit of control over the shape of their expertness or mastery, so that it can usually behave in a predictable way rather than always leaving it in the hands of the DM.
Next, I think we can see that this tiered system has a promising relationship between its math and the conceptual elements of a character. First, neither ability scores nor Competent training really outshines the other in any systematic way. This very much like low level 3e and 4e, before all the scaling bonuses start becoming an issue. In fact, as presented, Competent training is exactly like 4e without the +1/2 per level. Expert and Master training, however, really start to emphasize the training over the raw ability. Yes, having a good ability score will always be nice, but let's compare a Dex 10 character A that is an Expert in Acrobatics with a Dex 20 character B that is merely Competent. Perhaps a typical DC is such that the B has a 75% chance on any single roll, while A has a 50% chance of the same. On a check with Competent or Untrained difficulty the less dextrous character gets two rolls, and only 1 needs to succeed. This has a probability of 1-.5^2=.75, just like the less-trained creature. On an expert-level check with the same DC A succeeds 50% of the time, while B needs to wrangle up a second die to try it. And if B does get a second die (say from a potion of limberness) he still needs to succeed twice, which has probability .75^2~.56. In other words, they have almost the same likelihood of succeeding, but B had to spend more resources to get there. Consider the same situations where A has a 25% chance of success and B has a 50% chance of success. On the Competent-difficulty check A succeeds about 44% of the time (B has 50%), and the expert-difficulty check A succeeds about 25% of the time (B also has 25% assuming he can get a second die). Furthermore, A might also have access to Skill Tricks, giving him a further benefit over B. So while having high Dex will always be useful, it's still the case that one could make a perfectly competitive Dex 10 Acrobat with this system.
I think this demonstrates that a little training can go a long way, and that characters can meaningfully explore and specialize in pretty much any skill if desired. Furthermore, consider the impact for the believability of NPCs. Any person who spends their life really learning one or two things will get pretty good at it, even if they're just a normal Joe with 10s across the board and scarcely a commoner level to their name.
Another conceptual feature I like is how Aid Another can be used to help on checks a bit above one's own training. Basically, if you have someone to guide you a bit you can be helpful. To me this gives the classic apprenticeship and other forms of training a pretty clear demonstration in the game world. It can also be used for PCs. Maybe the dwarf PC who is an Expert blacksmith is trying to forge a masterwork warhammer for himself, but knows such a thing is probably beyond him at the moment. Well, he might find a master and actually Aid in the creation of his own weapon, which is in fact part of the process by he will eventually become a master himself. That concrete involvement can really sell the training. And on the flip side, it also says that not everyone can aid every check, that there are baby-steps involved. So that same dwarf PC when he was a Competent smith could not have meaningfully aided the creation of a masterwork weapon any more than I could aid a neurosurgeon.
Another aspect of the system to consider is its flexibility. There are a lot of elements here, so I'll only mention most in passing.
At the very start I mentioned that this system should be easy to ignore, so it's worth figuring out to what degree that seems to be true. The easiest step is simply to ignore Expert and Master difficulties entirely, which collapses the system down onto something like 4e without the scaling bonus (or with a very small one). If skill training above Competent isn't a baked-in part of character advancement this doesn't require too much retrofitting. One can convert DCs of the Expert- and Master-difficulty tasks by adding +3 and +6, respectively. (These approximately account for the change in expected value when rolling 1d20->2d20->3d20). Likewise, if a character would have gained Expert or Mastery in a skill one can increase their bonus by +3 and +6. This isn't perfect, but since most checks are already (by my assumption) at the Untrained or Competent-difficulty it shouldn't come up all the time. More problematic is the table that doesn't want a skill system all beyond pure ability checks. One can approximate this by doing the steps above, and then also reducing the DC by 5 for any trained check. Mathematically that is easy, but if there are many Competent-difficulty checks in the base game, one might have to do that a lot. Obviously one loses a lot capability (and gains simplicity) by doing either of these reductions, but in either case the entire skill system does not need to be rebuilt from scratch for the game to play out in a similar fashion.
Finally, I want to briefly list the weaknesses I can perceive, as well as the things I haven't really considered at all. The first problems I've already mentioned a bit: the game will sometimes move more slowly due to additional rolls, some types of metagame knowledge are harder to keep from players than in a game where every check is a single d20, and this skill system can't be totally ignored without making some changes to keep the game running the same. I think these are acceptable tradeoffs for my playstyle, but I doubt that's a universal opinion.
Next, this system could give off some non-D&D vibes. In particular, its success-based elements, however limited in typical play, may be strong enough to put off some people. It doesn't for me (at least while thinking abstractly), but this is solidly YMMV territory. It also lacks some aspects usually associated with D&D, principally a role for the natural 20. Neither 3e nor 4e had that for skills either, so maybe that isn't critical. In fact, with success-based elements it might be possible to give it one. Hmm, I'll have to give that more thought.
There are other structural elements that I haven't considered. For example, the right way (if any) to limit the total number of d20s that can be added to a check (particularly from aid another), considering that as defined here only 3 successes are ever required at most. With enough dice anything can become easy. With 3 successes required, the worst case scenario is 1/8000, which is rolling exactly 3 dice that all need to be 20s. On average 60 dice are required to get 3 successes when only a 20 counts, and the median of that distribution (i.e. where there is a 50/50 chance of getting 3 successes) is in fact just 54 dice. And for something more typical, where each roll has a 25% chance of succeeding, the worst case is about 1.5% chance of success. But on average one would get 3 successes rolling 12 such dice. Getting a 10 people to aid a check may or may not be too easy, it would depend on the situation, but a guideline can be useful so the DM isn't caught off-guard.
I also have no idea how I would handle taking 10/passive checks (or its 5e equivalent), opposed checks, and open rolls in this system. Taking 20 is covered by taking extra time, but taking 10 is very powerful since the DCs do not go to such extremes. Opposed checks and open rolls are even stranger. One could simply make them exceptions to the normal rules, but that's not particularly elegant. One could also try to convert them to success-based equivalents, but I'm not sure how that might be done elegantly.
[/SBLOCK]
Thanks for reading, and I welcome your comments.
Suppose that there are 4 corresponding levels of difficulty and training in the game that represent large qualitative differences in skills. For simplicity call these Untrained, Competent, Expert, and Master. When referring to difficulty they represent roughly the level of training required to succeed at such a check regularly. Every skill check has a single DC and a result corresponding to at least one of the 4 difficulty ratings. The DC of the check is a measure of how difficult it is for a character to leverage whatever level of training they have, from untrained to master.
The general form of a check involves one or more d20 rolls, each gaining a numerical bonus to the roll in the usual way, with each roll that meets the task's DC counting as a single success. The number of successes required to "succeed" at a given level of difficulty is listed here:
Code:
Difficulty Successes required
---------- ------------------
Untrained/Competent 1
Expert 2
Master 3The effect of training is shown below. Unless otherwise specified, all skill checks are rolled with a single d20.
Code:
Training Benefits
-------- -------
Untrained None
Competent +5 (or whatever) bonus to checks.
Expert Expert difficulty checks require only a single success.
May roll 2d20 on Untrained and Competent difficulty checks.
(Optional) Access to expert skill tricks.
Master Master and Expert difficulty checks require only a single success.
May roll 2d20 on Expert difficulty checks
May roll 3d20 on Untrained and Competent difficulty checks.
(Optional) Access to master skill tricks.In addition to training, additional dice can also be obtained by taking extra time, using the Aid Another action, or by various magical means.
Taking Time said:
Aid Another said:
Skill Tricks are ways for highly trained creatures to spend "extra" successes. I'm cribbing the name from 3.5's Complete Scoundrel, where I thought they were promising but a little underbaked.
Skill Tricks said:
Here is an example of this system which demonstrates essentially every aspect of it. Note that this is probably the most complicated version possible, not what I'd consider (or want) to be a typical check. As I envision it skill training above Competent should be rare, and should skill checks with a minimum level of difficulty above Competent. Consider a 1st level character, a student, that stumbles across a puzzling ancient scroll. He is no slouch, with 20 Int and Competent in the Decipher Script skill, giving him a +10 bonus on the check (for the sake of argument). For this scroll the possible results might look like this:
Code:
Decipher Script DC 25. Deciphering this scroll takes 1 hour.
Difficulty Result
---------- ------
Competent Subject of the scroll
Expert Gist of the scroll
Master Full translation of the scrollThe student attempts to aid the mentor on this check. Since the student is Competent his extra d20 will only applies to the Competent and Expert levels of the check. His aid might be unnecessary, therefore, since the mentor needs only a single success to fully translate the scroll in an hour. The mentor rolls a single d20 to see if he succeeds at the Master level. A 3. It happens to the best of us. He then rolls another d20 (since he rolls 2d20 for Expert-difficulty checks and he already rolled the first one trying for the Master level) and the student rolls a d20 as well to represent aid. They have, on these two rolls, about a 6% chance of two successes, a 38% chance of 1 success, and a 56% chance of no successes. Finally luck strikes, with a pair of 17s. Having an extra success thanks to his student's aid, the mentor uses the "New Insight" Skill Trick, which lets him reroll a single previous failure if he spends the maximum amount of time re-deciphering this text. The mentor and student rejoice as they read about an ancient order that secreted away a powerful artifact known as a Nether scroll. The author seems to have some idea where this artifact may be...
Over the next week the master carefully goes over the scroll, just as his student had done earlier. Normally this would grant only 2 additonal d20s because he already rolled once at the master-difficulty level that first day. However, due to New Insight he can roll all three, and so the probabilities are the same as his student a week or two earlier. Of course, he only needs a single success to decipher the scroll, which has a 58% chance of getting. His first roll fails, but the last two both succeed. Since he again has an additional success (thank you New Insight!) he uses his Flawless Memory Skill Trick, which grants him perfect recall of the original text as well as his translated version.
In the coming months he and his student publish the non-sensitive elements of their great discovery in a prestigious journal, causing jealousy among their peers, and the following year are awarded a travel grant to seek out their prize. After only two weeks they unceremoniously travel into the gently crushing fists of an Adamantine Golem because the Nether Scrolls are not to be trifled with. The End.
Here are some potential examples of Skill Tricks that illustrate how they might work. I've given no thought to balance.
- Sneak. Force an enemy to reroll a check that would spot you.
- Sneak. Once while sneaking you can run without any penalty.
- Climb. Once when you or another would fall during this climb you automatically catch and hold.
- Climb. May climb at full speed.
- Creature Knowledge. Gain combat advantage on a single attack against the appropriate creatures.
- Open lock. Open lock as a minor action rather than standard action.
- Open lock. Can perfectly reset the lock after opening, leaving no traces of tampering.
- Spot. You do not lose track of a creature you've seen one time when you normally would.
- Spot. Photographic recall of the thing being observed.
- Disguise. Once after the disguise is finished you may conveniently describe a detail that has been there "all along." (Although retroactive description might be too gamist for some.)
- Decipher script. Spend the maximum time re-deciphering a script to reroll a single previous failure.
- Decipher script. Perfect recall of the original and translated version of a deciphered text.
Finally, I think this system is largely orthogonal to several other interesting aspects of the skill system which would need to be defined in the released game, and would work well with any of them. These include the specific skill list (if there even is a predefined skill list) and whether or not skills are tied to particular ability scores. All the skills listed above are only meant to be illustrative. In addition, I haven't addressed at all how training may or may not be limited by level, although I think that is an important issue.
-------
I wrote a lot of analysis and it turned into a bit of an essay so I'm putting it into a spoiler block. The meat is all above, and I think the basic idea is clear, but if you want to delve further read on.
[SBLOCK]
In my view the default assumption of the game should be that training beyond Untrained/Competent is rare, and so are checks where the lowest levels of success have Expert or Master difficulty. That way the game isn't constantly being slowed down by very difficult checks, unless the PCs specifically pursue such things. The presence of multiple levels of success, extra dice on many checks that require only a single success, and Skill tricks is already pretty awesome. Not every lock needs to be of Master-difficulty, even at high levels, for the rogue to show off. That being the case, I would not make training beyond Competent a default part of character advancement. A high level character that wants to be a true master should strive for it, and I think the benefits outlined above are worth expending character resources to obtain. If there is a numerical bonus to trained skills just for leveling up it should be modest, something like +1/5 levels at most. That way the DCs can be kept in a reasonable even as some advancement is observed.
First, I want to talk about playability. It's clear that this is a system where most of the time a skill check is still just a single d20. Much of the extra complexity of the system only shows up when the player opts-in. Most players, I think, will not object to rolling another d20 or two when they get to use their awesomest skills. In fact, I suspect they will relish it. Or when multiple characters collaborate each one contributes at most one die, and quite possibly a die they didn't roll on their own turn in order to aid a skill check they regard to be of value. So the number of dice thrown isn't much increased, and again the players chose to make it happen. Likewise when a character takes extra time making a check, the slow-down at the table isn't very large and the player has a goal in mind, although I admit it is slower than taking 20. This tendency to need only a single success with training also sidesteps an annoying tendency of some success-based systems. Namely, when one needs 3 successes but gets exactly 2. What a heartbreak! That sort of thing should only happen in this system when the players are beyond their normal capabilities, and they know it. It might still smart, but it just won't happen that often.
Another aspect of playability is when a check has multiple levels of difficulty. A master character could, in principle, need to see if they rolled at least one success (of three) for Competent, at least one success (of two) for Expert, or at least one success (or one) for master. However, that looks a lot slower than it really is because a success at Master, for example, grants all the lower results as well. So the process on this check is this: Roll a single die. If it succeeds you get everything. If it fails, roll and additional die. If it succeeds you get Competent and Expert levels of success. If it failed, roll an additional die. If it succeeds you get the Competent level of success. It is easy to see that this is equivalent to rolling for all three checks at once, where the dice that correspond to each check have been labeled beforehand, without actually requiring the player to think about it at all. Just keep rolling until you succeed or run out of dice. Of course, if you don't want the player to know how many dice to roll (say there is a Master-difficulty result but you don't want the player to know that for metagame reasons) this is tougher to do. One could simply make the player always role a single d20 at a time, or have them select a die randomly after the fact, but either will slow down play a bit. A better solution might simply be to roll behind the screen. I think such cases would be rather rare, however.
From the DM side of things I think it's also more playable overall. This system allows a real easy rule-of-thumb to support multiple results, and in fact the DM only needs to select a single DC to do so. In prior editions, if I wanted to do something similar I needed to make up multiple DCs. Usually this was easy (+5 or +10) but sometimes it required more thought. Selecting the DC in the first place may take a bit of practice because we're so used to making the numbers skyrocket, but I think people would get the hang of it quickly. The main idea is this: Whatever is the primary result you have in mind, figure out what level of training should be required to do that regularly (i.e. with a single d20), and then set the DC for a person with that level of training. If the range of DCs is fairly compressed the numbers should be quite moderate, and the probabilities to consider are just those of the single d20 with its familiar and convenient uniform distribution.
Another benefit to playability for both PCs and DMs comes from Skill Tricks. These can relieve from the DM the pressure to make sure multiple successes always have some unique result, since this is in the hands of the player. And for the player it gives them a little bit of control over the shape of their expertness or mastery, so that it can usually behave in a predictable way rather than always leaving it in the hands of the DM.
Next, I think we can see that this tiered system has a promising relationship between its math and the conceptual elements of a character. First, neither ability scores nor Competent training really outshines the other in any systematic way. This very much like low level 3e and 4e, before all the scaling bonuses start becoming an issue. In fact, as presented, Competent training is exactly like 4e without the +1/2 per level. Expert and Master training, however, really start to emphasize the training over the raw ability. Yes, having a good ability score will always be nice, but let's compare a Dex 10 character A that is an Expert in Acrobatics with a Dex 20 character B that is merely Competent. Perhaps a typical DC is such that the B has a 75% chance on any single roll, while A has a 50% chance of the same. On a check with Competent or Untrained difficulty the less dextrous character gets two rolls, and only 1 needs to succeed. This has a probability of 1-.5^2=.75, just like the less-trained creature. On an expert-level check with the same DC A succeeds 50% of the time, while B needs to wrangle up a second die to try it. And if B does get a second die (say from a potion of limberness) he still needs to succeed twice, which has probability .75^2~.56. In other words, they have almost the same likelihood of succeeding, but B had to spend more resources to get there. Consider the same situations where A has a 25% chance of success and B has a 50% chance of success. On the Competent-difficulty check A succeeds about 44% of the time (B has 50%), and the expert-difficulty check A succeeds about 25% of the time (B also has 25% assuming he can get a second die). Furthermore, A might also have access to Skill Tricks, giving him a further benefit over B. So while having high Dex will always be useful, it's still the case that one could make a perfectly competitive Dex 10 Acrobat with this system.
I think this demonstrates that a little training can go a long way, and that characters can meaningfully explore and specialize in pretty much any skill if desired. Furthermore, consider the impact for the believability of NPCs. Any person who spends their life really learning one or two things will get pretty good at it, even if they're just a normal Joe with 10s across the board and scarcely a commoner level to their name.
Another conceptual feature I like is how Aid Another can be used to help on checks a bit above one's own training. Basically, if you have someone to guide you a bit you can be helpful. To me this gives the classic apprenticeship and other forms of training a pretty clear demonstration in the game world. It can also be used for PCs. Maybe the dwarf PC who is an Expert blacksmith is trying to forge a masterwork warhammer for himself, but knows such a thing is probably beyond him at the moment. Well, he might find a master and actually Aid in the creation of his own weapon, which is in fact part of the process by he will eventually become a master himself. That concrete involvement can really sell the training. And on the flip side, it also says that not everyone can aid every check, that there are baby-steps involved. So that same dwarf PC when he was a Competent smith could not have meaningfully aided the creation of a masterwork weapon any more than I could aid a neurosurgeon.
Another aspect of the system to consider is its flexibility. There are a lot of elements here, so I'll only mention most in passing.
- When a task might involve many checks and/or a long time, this system can let you abstract them into a higher difficulty level in a consistent way. Crafting and long-term research come to mind, but it might also be a way to handle things like stealth, forgery, etc. that might normally involve lots of opposed rolls.
- The DM can easily adjust a situation for substantial player creativity or blunders by adjusting the difficulty of the check. For example, convincing a hesitant Duke to commit troops might be an Expert check, but if the PCs investigate and find that a man he made an oath to long ago is in danger, that fact alone might make it Competent-difficulty diplomacy check.
- The DM can adjust for unique resources in the same way. If the party finds the lost library they can make knowledge checks against the ancient menace as though it were one difficulty level less than normal.
- This gives a consistent way for the PCs to attempt things that are essentially beyond their normal capability, and maybe even succeed in some partial way. I'm thinking of things like understanding artifacts, lock-picking the door to the national treasury, etc.
- Skill Tricks leaves the room open to expand the skill system in focused ways, while leaving a natural space for DM and player improvisation. With a little work it could form the basis for a unified stunting system.
- It could provide a natural mechanism for scaling rituals. Raise Dead might be Competent-difficulty, while True Resurrection the master-difficulty version. One could also use the Aid Another notion to make rules for mass rituals with tens or hundreds of participants. No one wants to roll 100d20 so clearly not everyone lends dice (or they do but all but the last 10 are abstracted away). Either way, sometimes it's cool to think about exactly what it takes to make a mythal, and to see how it connects to the rest of the game's mechanics in a natural way.
- There is a ton of other space opened up by player collaboration using this version of Aid Another. For example, a Competent skill user might assist an expert on an Expert check so that the Expert has chance to use a particular Skill Trick to save the day.
- Another benefit of collaboration might be to allow one, without too much modification, to create something very close to 4e's skill challenges. If I had my way it would have more ebb-and-flow and a less gamist focus, but that's it's own thread.
- If the DM decides there is room for an "Immortal-difficulty" task (say creating an artifact) it is clear that requiring 4 successes is a natural way to do it.
- Partial successes can also serve as a vessel for partial (or total) critical failures instead of the traditional method of failing the check by a certain margin. This isn't necessarily better or worse, just more in keeping with a success-based method.
At the very start I mentioned that this system should be easy to ignore, so it's worth figuring out to what degree that seems to be true. The easiest step is simply to ignore Expert and Master difficulties entirely, which collapses the system down onto something like 4e without the scaling bonus (or with a very small one). If skill training above Competent isn't a baked-in part of character advancement this doesn't require too much retrofitting. One can convert DCs of the Expert- and Master-difficulty tasks by adding +3 and +6, respectively. (These approximately account for the change in expected value when rolling 1d20->2d20->3d20). Likewise, if a character would have gained Expert or Mastery in a skill one can increase their bonus by +3 and +6. This isn't perfect, but since most checks are already (by my assumption) at the Untrained or Competent-difficulty it shouldn't come up all the time. More problematic is the table that doesn't want a skill system all beyond pure ability checks. One can approximate this by doing the steps above, and then also reducing the DC by 5 for any trained check. Mathematically that is easy, but if there are many Competent-difficulty checks in the base game, one might have to do that a lot. Obviously one loses a lot capability (and gains simplicity) by doing either of these reductions, but in either case the entire skill system does not need to be rebuilt from scratch for the game to play out in a similar fashion.
Finally, I want to briefly list the weaknesses I can perceive, as well as the things I haven't really considered at all. The first problems I've already mentioned a bit: the game will sometimes move more slowly due to additional rolls, some types of metagame knowledge are harder to keep from players than in a game where every check is a single d20, and this skill system can't be totally ignored without making some changes to keep the game running the same. I think these are acceptable tradeoffs for my playstyle, but I doubt that's a universal opinion.
Next, this system could give off some non-D&D vibes. In particular, its success-based elements, however limited in typical play, may be strong enough to put off some people. It doesn't for me (at least while thinking abstractly), but this is solidly YMMV territory. It also lacks some aspects usually associated with D&D, principally a role for the natural 20. Neither 3e nor 4e had that for skills either, so maybe that isn't critical. In fact, with success-based elements it might be possible to give it one. Hmm, I'll have to give that more thought.
There are other structural elements that I haven't considered. For example, the right way (if any) to limit the total number of d20s that can be added to a check (particularly from aid another), considering that as defined here only 3 successes are ever required at most. With enough dice anything can become easy. With 3 successes required, the worst case scenario is 1/8000, which is rolling exactly 3 dice that all need to be 20s. On average 60 dice are required to get 3 successes when only a 20 counts, and the median of that distribution (i.e. where there is a 50/50 chance of getting 3 successes) is in fact just 54 dice. And for something more typical, where each roll has a 25% chance of succeeding, the worst case is about 1.5% chance of success. But on average one would get 3 successes rolling 12 such dice. Getting a 10 people to aid a check may or may not be too easy, it would depend on the situation, but a guideline can be useful so the DM isn't caught off-guard.
I also have no idea how I would handle taking 10/passive checks (or its 5e equivalent), opposed checks, and open rolls in this system. Taking 20 is covered by taking extra time, but taking 10 is very powerful since the DCs do not go to such extremes. Opposed checks and open rolls are even stranger. One could simply make them exceptions to the normal rules, but that's not particularly elegant. One could also try to convert them to success-based equivalents, but I'm not sure how that might be done elegantly.
[/SBLOCK]
Thanks for reading, and I welcome your comments.
Last edited:
I may revise the first post if I can present the basic idea more clearly. Let me give it another shot.
Or just make some "skill tricks" universal, with a natural 20 giving an extra one. 
