Hey everyone. This time we're going to be looking at the roll and keep dice system featured in 7th sea first edition and all editions of Legends of the 5 rings save for the most recent 5th edition. The only edition of L5R that I've actually played is 4e, so when I reference L5R-specific rules I'll be talking about that edition. Roll and keep is definitely a unique dice system, so I think I'm probably going to devote this post solely to exploring it and comparing it to the other systems we've looked at so far.
The roll and keep dice system, when comparing to the other systems we've discussed, is most similar to a dice pool system like Shadowrun's, with two big differences. Those differences being that rather than counting up successes on the dice you actually sum the numbers that the dice land on, and that you only get to “keep” some of the dice. A roll and keep roll can be written as XkY, with X and Y both being digits. The first number, X, is how many dice are rolled, and the second number, Y, is how many dice you get to add up to get the total result of the roll.
Let's look at a quick example to get a better feel for how it works. Let's take the roll 5k3. For this roll, the first number is 5 so we're going to roll 5 dice. 7th sea and L5R both use d10s for their dice systems, so we roll 5d10. Imagine we rolled the dice and they landed on 1, 4, 5, 5, 9. We look back at the original notation for the roll and see that we get to keep 3 dice. Typically, you want to get as high a result as possible on your rolls, so we “keep” the 9 and both of the 5s, then add them up to get a total result of 19.
Now as for how you determine how many dice to roll and to keep on a test, 7th sea and L5R use mostly the same method. Characters in the systems have attribute ratings and skill ratings. Each point a character has in the relevant attribute for a test adds a kept die to the roll, written as 1k1. Skill ratings, on the other hand, each add an unkept die to a roll, written as 1k0. So, if a character had a strength of 4 and an athletics skill of 3, they would get 4k4 from their strength rating and 3k0 from their athletics. You combine these numbers by adding up the numbers preceding the k, and then adding the numbers after the k. In this case we add the 4 and 3 that precede the k to get 7, add the 4 and 0 after the k to get 4, then write out the final result of 7k4.
As you might expect, this method can get a bit time-consuming. Not only are there two important numbers that need to be known for a roll instead of just one, but also the result of a roll is the sum of multiple dice values rather than just counting up successes. The method that the mentioned roll&keep systems use to keep rolls a bit speedier is the concept of dice overflowing. In both systems, there are never more than 10 dice rolled for a single test. In the case of a character having more than 10 rolled dice, the excess dice are “overflowed” into kept dice. The specific implementation of the rule is a bit different between the systems, but the general concept is the same. 7th sea's is a bit simpler, so we'll look to that for an example of the concept.
Let's say a character has a total of 12k6 on a check. This would suggest that we roll 12 dice and keep 6 of them, but because we can't roll more than 10 dice, we overflow the extra two rolled dice. When you overflow unkept dice, they are converted into kept dice on a one-to-one basis. In this case, we add the two overflowed dice to our kept dice, giving a total of 10k8 for the test.
The same concept applies to kept dice, though the way that we overflow them is a bit different. This time let's assume a character has a total of 12k10 on a roll. We go through the normal overflow rules, ending up with 10k12, which obviously doesn't work because we can't keep more dice than we roll. Normally if such a thing happens, like a character having 5k7 on a roll, you just keep all the rolled dice, with no extra benefit from the two extra kept dice, since there are no more dice to keep past the first 5. However, in this case since we have 10 rolled dice and the kept dice total is higher than 10, we get to overflow the extra kept dice. Each of these extra dice is converted into a flat +10 to the roll result.
This might sound quite powerful, and it absolutely is, but in 7th sea is it very uncommon for a character's roll to ever reach this level. The normal maximum for both attributes and knacks (7th sea's name for skills) is 5, meaning a character that has raised both to their cap will have a roll of 10k5. This is still quite powerful, as having 10 rolled dice means that any additional dice, kept or unkept, will be adding directly to the total kept dice, with the possibility of going beyond 10k10 if circumstances line up. L5R 4e has its attributes and skills both max out at 10, but it's overflow rules in the case of a roll greater than 10k10 are much tamer, as well as it being very uncommon for characters to have attributes higher than 6. That mostly covers the core aspects of the system, so now I can talk about what I like about this system, and what I'm not fond of.
Something that I think is really cool about the roll and keep method is the way it allows for bonuses to a roll to have different “weights”. There's a big different in the average increase to a roll when adding 1k0 vs 0k1 vs 1k1, and I think that makes for a pretty elegant way to simplify roll bonuses and penalties while maintaining a bit of the granularity.
When you account for the overflow rules though, it does sort of create its own “game” of trying to hit the specific breakpoints where you start to overflow on your rolls for something. For instance, having a 9k4 for a test and getting an extra unkept die for a 10k4 makes very little difference in the roll result, adding less than 2 to the average roll. But when you have that 10k4, getting another unkept die will mean it immediately overflows into a kept die, giving a 10k5 on the test which is a huge difference, adding 6 to the average result.
This “minigame” can be enjoyable for some people, myself included, but others may find it a bit annoying. Two characters receiving the same circumstantial bonus but getting drastically different effectiveness out of it can be frustrating, as not everyone wants to be searching for specific abilities that give them the bonuses they need to hit those overflow thresholds. As I mentioned, I love that aspect of the system. I think it makes character creation in the games that use it a bit more interesting, and if you know how it works you can have a good sense of what you should prioritize in terms of character advancement. But I think that for first time players it can feel like you need to be guided through character creation rather than being able to make whatever choices you want to make for the character.
There's a lot of variability of what exactly the numerical benefit of adding 1k0 or 1k1 to a roll is, which makes it a bit hard to give a direct comparison of average roll result to narrative character ability like we were able to with the previous systems. What I do think is kind of interesting with roll&keep regarding mechanics reflecting the narrative is the way that attribute and skill ranks contribute to rolls. Skill ranks adding 1k0 to rolls does a good job reflecting what practicing something would naturally do. That is, it makes you generally better as well as more consistent. The increases in average roll result from adding an unkept die usually aren't large, they start off adding around 3-4 when in the very low skill ranks and eventually end up adding about 1.5-2 when the number of unkept die is much higher than the kept dice. However, adding unkept dice from skill ranks is very effective at making rolls consistently passable. For instance, if we look at 2k2 against 4k2, the chance of getting a pretty abysmal total of 4 or lower on the roll is about 6% for 2k2, but is only 0.2% for 4k2. If we go to 6k2, the chance drops all the way to 0.003%, which is incredibly small. I think this does a good job representing the way people become better at something as they practice that thing specifically. A beginner (2k2) would be able to perform as well or better than the expert (6k2) occasionally, but the expert is basically never going to fail a very simple task in their area of expertise.
I do think that the roll and keep method, when using the skill and attribute rules from 7th sea, does overvalue the attributes side a bit too much. Taking the expert from before, if we compare them to a more naturally gifted person that has only been practicing a skill for a short time (3 ranks in the attribute and 1 rank in the skill for 4k3 total), the gifted beginner actually performs better on average than the expert, a mean of 20.3 vs 21.9 respectively. The only thing that actually keeps these averages close are dice explosions (which I probably should've mentioned earlier as it could be considered a core part of the dice system, so I'll go through it now).
Basically, when any die on a trained roll lands on a 10, that die explodes, meaning that the dice gets to be rerolled, and whatever that die lands on is combined with the original 10 for the purpose of that die's total value. This is a very similar system to how explosions work in Shadowrun 5e, except this time we're using the combined total of the dice results rather than successes. Explosions can be extremely impactful, and they do help to offset the tendency for naturally talented beginners to outpace un-talented experts. Unfortunately, this doesn't exactly solve the problem, and makes character progression focus heavily on getting more kept dice through attribute increases rather than improving skill ranks.
L5r 4e does have it a little bit better in this area, as skill ranks going up to 10 allows a character to get to the 10 rolled dice breakpoint more easily without having to increase attributes. L5R also has more strict overflow rules, in which it takes 2 overflow rolled dice to get a single extra kept die, which ends up making it still pretty hard on low-attribute characters. I do think this is a bit of a shame, as I think the elderly master and the hard-worker archetypes are cool character concepts that unfortunately don't get much mechanical support from the system. That is a bit of a nitpick at the end of the day though, as for the most part I do think the system does a good job representing most other aspects of progression.
When comparing to the other two dice systems we've looked at, roll&keep is definitely most similar to the dice pool successes system, so let's compare them directly. One of the big differences in these systems is how they evaluate the result of a roll. With the success-based system each die is binary, either it is a success or it isn't. Roll&keep instead has you add up the numbers that the dice land on, so each die can vary in value anywhere on the scale of the die. For an in-person experience, I would definitely rate the success-based system higher, as it speeds up the actual evaluation of the rolls a lot. Though, when using dice rolling tools or playing online, I like the total result of roll&keep a bit more. Each die having some value makes rolls less swingy, as every die that you keep is at least contributing something to the roll total, even if it isn't much.
Roll&keep having the concept of kept dice is something that I also see as an interesting innovation over the standard dice pool system, at least in a vacuum. I think the concept of keeping only some of the rolled dice adds a lot of depth to the system, as rolled and kept dice can be improved somewhat independently. Shadowrun 5e does actually have a mechanic sort of adjacent to kept dice in the game, called limits. These limits are, as the name suggests, a limit on the number of successes a character can obtain on a roll. This essentially functions like kept dice, but typically in shadowrun limits are high enough for them to rarely matter and can be bypassed entirely when spending edge to push the limit on a roll.
Comparing roll&keep to the d20 system is going to be mostly a repeat of last time when we compared the dice pool successes system and d20 system, so I don't think there's too much of a reason to delve back into that at the moment. I'm also going to start varying the topics of the entries a bit after this post, which means it's a good time to give a bit of an overview of my thoughts on this topic as a whole.
In general, I think that any dice system can work in a TTRPG, provided that the rest of the system and the tone match up. I'm not a big fan of d&d 5e, but I do recognize that WotC have created that system with specific goals and considerations in mind, and the system does definitely create a cohesive experience. Similarly, Shadowrun 5e's dice system meshes well with the mechanical complexity of the game as well as the more professional super-criminals vibe that is expected of the standard shadowrunner team. We could theoretically swap the dice systems for these two games, and they would both still work well enough, but would end up with that small disconnect between mechanics and narrative.
This is actually does come back around to explain one of the gripes I have with d&d 5e's specific implementation of the d20 system, where modifier progression is relatively small, which has always left me with a strange impression as a player. It's also an area I think pathfinder 2e does a lot better, as they modifier progression does give the feeling of increasing ability and importance for a character as they level up, which I think aligns perfectly with the novice-to-demigod narrative progression of level 1-20 characters.
Outside of meshing with the rest of the game, each dice system comes with its own benefits and drawbacks regarding proper narrative representation, time complexity, and learning curve. Certain aspects of a dice system are going to appeal to certain types of players, which should be considered when choosing a dice system for a game.
I think that brings us to a good stopping point for the moment. This ended up being a pretty fun series of posts to do. Forcing myself to really think about what I do and don't like about these dice systems has been a good way to give myself a better insight into what I value in a game, and why I have certain feelings about specific games. For the next post, I think I'm going to give a bit of an introduction to the TTRPG that I've been sporadically working on creating recently. I've had a lot of scattered ideas for it, so I think writing about it will help me a bit in organizing my ideas and reaching decisions about how I want to implement things into the game.
Until next time.
-JGDev