Dice and I have never gotten along particularly well in the context of roleplaying games. I appreciate them as fun Platonic solids (other than the black sheep d10), they feel nice in the hand, and I’m not immune to that variable ratio schedule behavioral reinforcement that comes with a critical success. However, every game I’ve published until now doesn’t include them. They’re just too random: they make clever plans fail to work, interrupt dramatic moments with incongruous outcomes, and very often make competent player characters look like bumbling fools. I spent many years working very hard to avoid resorting to them, mostly by eschewing randomizers altogether, sometimes by adding cards instead. This past year or so, though, I’ve reevaluated that position. All of those criticisms are correct, but trying to remove dice has revealed the situations where they shine.
What are dice good at? Well, the most obvious thing is that winning a roll feels exciting, it’s a little hit of gambler’s dopamine. It’s so expected from most RPGs that when I’ve tested diceless games, players would sometimes report that it didn’t feel like they were actually taking an action when the moment lacked the binding agent of the roll. Perhaps more crucially, they take heat off the GM: players will readily blame dice for their misfortune, even at low probabilities of success, but their ire may lodge with the GM when that’s the person who keeps telling them they failed. Third is the basic yet endlessly useful function of rolling on charts to procedurally generate people, places, and events, which also ease GM pressure by giving them a semi-automated tool to refer to.

The Six-Sided Solution
So if dice cause as many problem as they solve, how do we retain the good aspects while excising the bad? By rolling first and saving (for) later. I’m calling this new mechanic the “dice bank”, and it works like this:
Instead of rolling a die when you need to make a check, roll six dice ahead of time and put them in the center of the table. Whenever a player would roll a die, they instead choose one of the dice from the table and apply it to their check as though they rolled it (adding modifiers versus a target number as usual), removing it from the pool. Once all six dice have been expended in this way, roll a new set of six dice and repeat. The GM uses their own separate pool of dice for NPC checks.
Everything else stays more or less the same. You can slot this system into any otherwise standard one-die resolution mechanic – although I wouldn’t recommend it for systems without penalties on failures, as otherwise you could easily eliminate low dice by calling for meaningless rolls. I’m using d6s for all actions in the game I’m working on to test the concept, but it would work just as well with d20 or otherwise. You’re still always rolling, but you can peer a few steps into the future and pick when you want that high roll and when you can accept a low one. Therefore you’re always making a decision when you go to pick up a die, which isn’t necessarily the case with the normal roll-to-save paradigm.
I’m calling it a dice bank as a generic term, but in a given game it might easily be given a diegetically flavored term like “fate”, or described as the opportunities presented to you. It’s inspired by the video game Citizen Sleeper, in which each day you roll a set of dice and assign them to the actions you take during that day. Although in that game the allocations only modify probabilities up or down, with the bank you directly use the pre-rolled dice for actions.
What this means is that, although you still get the full range of successes and failures at randomized rates, you’re making a meaningful choice every time you make a check. You’re asking yourself (and often, collaborating with the group strategically) how important it is to win this time, enough that you’re willing to fall on your sword next time. Input randomness instead of output randomness, a melding of skillful play and dramatic flow.
It also creates some unique design space in how abilities might manipulate the bank: you might be able to change the size of the bank to “see the future” a little further, toss out dice you don’t like, refill the bank up to maximum sooner than waiting to use it all up, store a die and carry it over into the next pool, or even opt to do an immediate new roll instead of selecting a die from the pool. Perhaps most interesting is the possibility of abilities you trigger by spending a specific rolled number, e.g. a spell that can be cast only with a 5. Such an ability would be especially valuable if it wanted low values like 1 or 2, because it would be an efficient way to get rid of dice likely to create failures in normal checks.
Dice might also be spent on things that aren’t strictly skill/attribute checks: in theory they could also be applied to encounter rolls and the like. Essentially anything the system procedurally generates (events, reactions, hex contents, treasure) could be handled in this way. By default I imagine that the GM would use their own dice bank for these, both in-game and out: it would turn pre-session prep into a sort of minigame in which rolled dice became a sort of budget to work with, assigning them to different things. Alternatively (or on top of that), players might be allowed to influence procedural rolls with their own dice, explained as, say, the ability of a hunter to track down certain prey and avoid creatures they aren’t interested in.

Further Considerations
There are several dials to be tweaked here that may significantly alter how the bank feels in play: first of all, how many dice go in the pool. I’ve chosen six to roughly match with the upper end of the number of players around the table, but it could be higher or lower, or simply be set equal to the number of players.
Second, each player could have their own pool of dice instead of drawing from a communal reserve. This would have the upside of removing the chance of arguments about who got to use which dice, but the downside of limiting the teamwork aspect of sharing dice. Collaboration could be reintroduced by making an assistance action that hands people dice from your pool to represent your aid. However, for the sake of parity the GM would then potentially have to maintain dice pools for individual NPCs, which would be much more of a pain. This may not be a popular opinion, but I strongly favor GMs using the same rules as any other player at the table, for consistency.
Third, it wouldn’t have to be dice: you could easily use a deck of cards instead. In fact, if pools were individualized as above, the system would simply become having a hand of cards that you played from, a well-established concept everyone is already familiar with. I suggest dice first due to the prevalence of dice as the medium of randomization in TTRPGs.
Fourth and finally, the dice could be rerolled on a particular schedule, such as whenever you make camp, or at the start of an encounter. This approach has a certain appeal, but I’ve decided against it for because it gives you more opportunities to throw out low dice. The choice of when you allocate your low dice is the most interesting aspect of the system, and not something that should be easily avoided.
Although fairly simple on its face, the dice bank could manifest cascading effects throughout the rest of the rules once actually put into practice, so I’m going to get a few rounds of playtesting in and report back in another post. I shall redeem dice on my own terms.








