Against the Universal Skill Check
I have a love-hate relationship with skill systems in RPGs. On the one hand they provide several systemic benefits and allow for more character customization than class-based games. On the other they can often feel underwhelming and don’t always work well as typically implemented. I would like to examine the downsides of skill systems, and hopefully fix such systems for my own use.
While I will touch on the underwhelming nature of skill systems, I am foremost a GM and designer rather than a player and am mostly interested in how they are used in game systems. The role of a game system is, among other things, to answer questions the GM throws at it. Almost all RPG skill systems are designed such that the character possesses a numerical skill which is used to weight the result of a dice roll towards the player’s favor, with the result being a binary success or failure. It’s easy to see where this comes from, at least in D&D. Starting in 3rd edition, skill rolls are an extension of to-hit rolls. You either hit the enemy, or you don’t. You either succeed with your skill roll, or you don’t. Some games focus on the binary result as an issue, but I think the problems they’ve identified are a symptom of the larger problem of using a universal resolution mechanic for all situations.
And to be fair, the standard binary skill system works pretty well all things considered. If the GM isn’t sure whether a character can climb a wall or not, rolling with a binary outcome will answer that question. But I find there are a lot of situations where this simple outcome doesn’t provide enough information to be useful or provides frustrating behavior. One such situation is knowledge checks. A simple numerical score doesn’t tell you what your character does and does not know. You can roll to find out, but this can lead to situations where you succeed a roll to recall deeply obscure information only to turn around and bumble basic facts. Crafting is another pain point. If you are trained in the skill of weaponsmithing, this sort of skill system will tell you whether you successfully craft a given item, but not tell you how long it takes, what materials, or allow for differences in quality of the final item. We are applying a solution designed for resolving attacks to very different questions and unsurprisingly not getting the greatest results.
I suppose my issue is that skill checks tend to be used as a “good enough” replacement for systems which would provide more information. I consider combat in most RPGs to be an example of a fully developed system. It is a system which provides more information and choice than skill rolls, despite having proto-skill rolls baked in. On the other hand, something as common in RPGs as stealth is usually regulated to a single roll determining whether the character is spotted or not, carrying disadvantages for the viability of stealthy characters and for groups of people attempting to sneak as a group. Doing this to combat would likely result in a riot.
A few games do go the extra mile and design “full” systems for things other than combat, such as the hacking system in Shadowrun or the magic system in Ars Magica. These specialized systems exist alongside typical skill systems which handle other things the game isn’t focused on, providing further evidence for my belief that skill checks are often used as a patch over more developed subsystems.
Of course this is all coming from my taste and predilections as a GM. I have a tendency to want the rules to provide a lot of information, mostly so that the game can be consistent in how it operates and players can plan ahead. Not every GM or game system works that way, nor should they. Not every game can and should have detailed systems for everything. There is something to be said for the basic binary skill system giving a GM more room to improvise what is happening, or for a game to be able to focus on a single aspect it wants to highlight but leave other options open, if less developed. Regardless I am pulled by the allure of what I’ve been mentally calling “the forever game”, a game where players are able to inhabit roles as diverse as active adventurers, savvy merchants, master craftsmen, or scheming politicians and for each of these paths to have their own overlapping spheres to work with.
While running a game where the characters all go off in different directions with different life paths would probably be a mess, having these rules allows for different types of campaigns to be easily run out of the box and allows characters playing in a focused campaign to dip meaningfully into other aspects of the game world. I have more thoughts on what I’m looking for in this “forever game” which will get their own post.
Besides creating a forever game there are other reasons to consider rejecting the universal skill system or at least elaborating further on certain skills. Giving skills their own subsystems allows characters with different skills to feel very different from one another despite being made from similar building blocks. A character based around crafting skills should, in my eyes, be very different to play than a character focused on combat. By developing the crafting skills further we give that character things to do and a solid place in the game. Another reason to consider developing systems for skills would be to help RPGs branch out into different genres where combat isn’t the focus. So many RPGs have bespoke systems for combat despite nominally being focused on something else or claiming to be universal in nature. I’m sure I’m not the first person to ask the question of what these games could be without combat, but to my knowledge this doesn’t seem to be a particularly explored space.
These additional systems have the interesting property of portability. So many games sharing the same universal skill system makes it possible to create bolt-on subsystems which use the skill checks as building blocks. These can then be carried by a GM from game system to game system as they wish.
Circling way back now, the other issue I see with skill systems is their lack of umph. One of the universal skill system’s strengths is also a great weakness: Incrementalism. Small increases in numerical values simply don’t have the same narrative power or draw as gaining a new class ability or feat. Incremental quantitative changes do have a certain appeal, computer RPGs are living proof of that, but qualitative changes are much more engaging besides playing better to the strengths of the medium. I believe a big enough quantitative increase is a qualitative increase, but in general I think character advancement through feats is more fun than skill increases. Feats do a better job of describing who a character is than a +1 to haberdashery. Of course, the issue with feats is that they don’t represent incremental progress very well, which is something nice to have in an RPG system especially as players have been conditioned to expect continuous character improvement. Incremental progress (especially with diminishing returns) strikes a happy medium in my eyes where player characters can progress and character building is rewarded but it doesn’t overtake the campaign itself.
My own untested and theorycrafted solution to this is somewhat inspired by Alexis Smolenk’s sage system. Increases in skills unlock feats at regular intervals, let’s say every three skill levels. Skill levels are used primarily as a counter towards these feat intervals but the intermediate numbers can also be used for skill checks, as inputs to other systems, or for feats depending on the skill. In the style of Alexis’ sage system some feats are granted automatically at reaching a given level of skill. Others must be bought separately with XP but have skill or feat requirements of their own, typically focused on the three level feat intervals. Some feats will need to be found in the world to be learned. I am aware I have essentially just re-invented feat trees, but the idea has a certain shine for me at the moment (especially as it would help banish some of the cracks forming in my current house rules). I think the main thing I like about this skill system is that it provides what I perceive to be the benefits of a skill system without the downsides. Skills allow the advancement system to be incremental, but each skill is used in a different way as the situation demands and qualitative abilities are still provided to the players.
There’s a reason skill systems are one of the cornerstones of modern RPG design. I just think that they need to be broken out of the constraints of the universal skill check in order to reach their full potential.
2023-07-26
Multiparty Processing
Let’s suppose that a GM wants to run multiple parties in the same world. They have two parties A and B which play on different days weekly. Let’s say Tuesday and Thursday for the sake of example. Everything goes well until the following scenario happens:
Party A goes to dungeon X and ends the session mid-dungeon. Before their next session, party B goes to the same dungeon and captures the largest treasure hoard.
There are several ways to prevent or resolve this issue, depending on the priority one wants to put on the real world timeline versus the game world timeline.
If the real and game timelines are tied together, such that each day which passes in the real world causes a day to pass in game, then party A will either have to leave the dungeon by the end of the session or else be forced to spend a week of game time in the dungeon, which is likely quite undesirable. I’m going to call this the “freeing” method, as party A is forced to let others use the resource of dungeon X when they aren’t actively using it. This prevents the problem from occurring in the first place, at the cost of imposing time limitations on the party. There are other problems with this method which will be discussed later.
If the real timeline is held to be more important than the game timeline, then we reject that this is actually an issue. Events happen solely based on their real timelines, so party B can absolutely swoop in and steal the treasure out from beneath party A. I’m going to call this the “ignoring” method. This is a fine solution for many styles of running games, but causes issues for my own style of sandbox games where the game timeline is important. If the GM’s job is to simulate the world fairly and the game timeline is totally subsumed by the real timeline, then when is it fair for the GM to change the world on its own accord? You lose any real ability to answer what day it is in game. My criticism of this approach mostly stems from my own principles and desire to simulate the game world. I feel as though there is more to say here but I’m not in a position to dig into it at the moment.
If the game timeline is preferred over the real timeline, then one must prevent the real timeline from causing issues with the game timeline. In the example, this would mean explicitly preventing party B from going into the dungeon due to party A already being there. I will call this the “blocking” method. This has the advantage of preventing the issue, but also fails my principles for games. Trying to prevent the players from reaching the dungeon without explicitly telling them they can’t go there will likely be immensely frustrating for the players and I would consider this to be classically terrible GMing, but explicitly telling them breaks immersion. Either way this option undermines the reality of the game world. This option could be saved or at least the blow lessened by some diegetic reasoning in this specific example, such as a board the adventurers post their claimed dungeons on, but it’s unlikely that a given diegetic solution will make sense broadly across multiple possible collision scenarios.
Another option might be to split the game timeline. Party A ends their session in the dungeon. Party B steals the loot. When party A comes back the loot is still there. I consider this “splitting” solution to be a hack. I see two possible ways to implement this, either having the timelines be totally separate or merging them after collisions. If the timelines are totally separate, I would argue you aren’t running two parties in the same game world to start with. You wouldn’t have any of the gameplay benefits of multiple parties for the players, and as the timelines diverged the prep benefits of having only one world are also lost. This is why I call it a hack solution. The other implementation of merging the timelines would be something like both parties A and B coming back with two identical copies of the same treasure. Frankly, trying to merge the timelines after every conflicting event sounds like a pain in the ass. It could be justified in-game with multiple timelines being a canonical feature of the world, and I think would be more satisfying to me than doing so for the blocking method, but this is a high level cosmological choice which greatly changes the flavor of the game world and perhaps makes the player characters more cosmically important than I think they ought to be.
My own previous attempts at multiple parties have drawn on the previous work of the bOSR and their patron-style play, which I consider to be the canonical example of the freeing method (though not one that Gygax et. al. used, and certainly the not only way to play D&D). As such I have put more thought into problems with this method. Using the patron-style method of fixing the real and game timelines, we can still run into conflict situations. Consider this second scenario which can only exist with linked timelines.
Party A walks to city Y, which takes them two weeks. They buy a unique magic item, Z. All of this happens in one session. The next Thursday, Party B teleports to the city and plans to spend two weeks there, all in the same session. This means they are in a position to buy Z before party A does. Should they be allowed to do so?
(Following patron-style play, the session happens just as a normal D&D session would (i.e. during the session the players play out traveling to the city and taking any actions there), but the characters are unplayable for two weeks afterwards as the events of the game session play out in real time. The “camera” of the game timeline follows them for the session, but afterwards real time has to catch up to the camera.)
As far as I can tell, in these situations the bOSR actually follows a real-timeline first method here, simply resolving actions on a first-come first serve basis. So party B would simply be unable to purchase the item, either for meta or diegetic reasons. This has the advantage of being easy for the GM to resolve as well as being a general solution, but does restrict the possible actions players can take.
A more “pure” linked-timeline method might be to consider all actions which occur in the future to be orders and not set in stone until that real-world date has passed. So party B would be able to purchase the item, party A would be out of luck and have to re-submit orders from the point where things changed. This creates a lot of extra work for the players and the GM, having to retcon and then effectively re-play decisions. It would also be extremely frustrating to say, get a really good roll when fighting a powerful monster only to have that version of reality not happen. I could also see it causing a considerable amount of meta grief with players jockeying for position due to their ability to effectively know the future. I haven’t tried to come up with an example scenario, but I also fear that under this paradigm it might be possible to get stuck in an infinite loop of retcons. So though this is a pure solution it is also a much messier one.
This scenario initially involved party A burning down the city, but this was revised as multi-party PvP and indeed PvP in general is a can of worms I’m not opening in this post. But I do want to consider it, and there are some interesting questions such as whether you can fairly interact with a character whose player is not present even outside of a PvP context.
I’m sure there are some solutions to this scenario I haven’t thought of, and that there are implementations of each of them which reduce the impact of their downsides. I’m probably going to stick with the one-to-one method for now as it avoids most of the problems which conflict with my own personal gaming principles, but for people largely unconcerned with a consistent canon or fixed timelines the ignoring method is probably the easiest and simplest. I may consider adopting a modified version of it myself once I can properly analyze and resolve my own issues with it, as it provides a more casual experience for players than one-to-one time and my own players have complained about being forced to leave the dungeon at the end of a session. If I was designing a new multi-party campaign from ground up the splitting method with merging timelines might work well alongside anti-canon tools, but it would need to be tightly integrated with the setting and my current project is not compatible with that.