A Skill-based character attribute system

I may not have a lot of experience making videogames, but I do know RPG-systems! And it seems rather odd to me that so many CRPGs that have been developed over the decades very clearly attempt to emulate the mechanics used by pen and paper games to determine if an attack hits or a skill checks succeeds. Sure, it works. But those kinds of mechanics are designed around the necessity to introduce randomness by rolling dice and the ability to make all the calculations involved in your head in a second or two. These are limitations inherent to the medium, but in a videogame there really is no need to abide by them. You are needlessly restricting your own design. When the very first CRPGs were being made, and it was still completely unknown what form such games would actually take, I do see some wisdom in those early developers choosing to use a rules system that they already knew does work, and concentrating their work on the technical aspects of having it run in software. And of course, Baldur's Gate is a licensed D&D game, so obviously it had to mirror the game mechanics of D&D. But overall, as someone who understands pen and paper system design, designing the mechanics of a videogame as if it was one seems like a really dubious design decision.

After I had roughly defined for myself what kind of game I would want to make, I immediately got into thinking what kinds of game mechanics I would want to use for it? What would the rules system be? Should my game have character classes? Should it have levels? Should characters instead be defined by skills? And in that case, would it even still make sense to have Experience Points? Do I even need to have health points?! (I quickly realized that yes, I absolutely should use health points. But that's still a conscious design decision, not something that a game automatically would have to have.)

Coming from a pen and paper background, and more importantly a game designer perspective, one thing that got me really excited about the idea to try making a videogame is the possibilities for trying out the kind of game mechanics that sounded interesting on paper but have turned out to be too impractical and unwieldy in pen and paper games.

Entirely out of a gut feeling, I have decided that I want to go with a percentage based skill system for nearly all character attributes. It's always been an intriguing system in the Basic Roleplay system, and I really enjoyed how it plays in Kenshi.

Most RPG systems have the main parameters of characters set by four to eight main ability scores. But in most games, these scores don't actually do that much. D&D having the scores go from 3 to 18 came about because rolling six dice was a good way to create a normal distribution when six sided dice was the only thing they had. But you don't actually use the scores. The scores simply reference a row in a table that lists a number of modifier to various secondary attributes. You don't actually need ability score. And in a skill based system, the only thing they would do is to determine your skill values at character creation. And you don't need ability scores for that. Therefore, my game will not have them.

The majority of character stats are going to be skills. Everything that can be a skill will be a skill, such as hand weapon attacks and ranged weapon attacks, or avoiding getting poisoned. And of course negotiating, sneaking, opening locks, and all the other things that are more traditionally understood as skills. But skills should be limited to something you do. The amount of injuries a character can take would be one stat that doesn't really work as a skill.

Skills are percentage based, though that will remain hidden behind the curtain. The game interface will simply display a skill rank of 16, but for the actual game mechanics, that will be a 16% chance of success. If a task is particularly difficult or easy, for example an enemy you are trying to hit wearing armor, those circumstances modify that percentage up or down. Because of this, skills can actually go higher than 100. If your skill is at rank 120, but the target of the skill attempt adds a -40% modifier, the chance for success will be only 80%. Also, once all modifiers have been added to the character's skill rank, the final chance for succeeding the action can never be higher than 90%. Even if the skill ranks and modifiers come out to a total of 110%, the actual check will still only be done at 90% success.

The reason I want to go with this approach is because of the skill improvement mechanic you can combine with that. Any time you use a skill, you have a chance that it will increase by one rank, which is the inverse of your base chance for succeeding on the skill. If your skill is at rank 30 and you have a 30% base chance, then using that skill has a 70% of raising the skill rank to 31. While skill ranks could go to 100 and higher, the chance of increasing the skill rank stays at a minimum of 1%.

The result of this is that the lower your rank in a skill is, the faster it will increase when you start using it. Even if a skill were to start at rank 1, it would be pretty fast to raise it up to 50. I've done some quick estimations on this, and getting a skill from 1 to 99 would on average take somewhere in the range of 550 attempts. And from 50 to 99 still about 450 attempts. Which I think looks like a pretty good pace for the kind of gameplay I have in mind. What this should produce is characters that are pretty well rounded out among their skills, and it should provide decent flexibility in changing your playstyle over the course of the game.

One thing I like about this approach for increasing skills is that it avoids having a progress bar on the skills and the players won't know how much longer it will take to reach the next rank. If players will like that is something that will have to be seen if I will be able to get this game to a sufficiently advanced prototype for testing. There is a certain randomness to it, but at lower skill ranks the ranks will go up almost every time or every second time that you use them. And over the course of 200 to 300 skill uses, the probability of overall progress speed being close to the statistical average becomes very high. Two characters who uses the same skill 300 times should end up with very close ranks.

All of this is still at the very early ages of pre-production, done pretty much entirely as a few scribbled pencil notes. But I think this is a very solid character ability mechanic that shouldn't be too terribly difficult to implement. But obviously, it will have to be seen how it will actually hold up in playtesting.