Brainfold On Python, game development and everything

The problem with CRPGs

Today, two posts for the price of one! This was going to be a game design -oriented post about character development in computer roleplaying games (CRPGs), but turned into a rant. I'll write about game design later.

In my opinion, the problem with CRPGs is that they are either direct ports of, or heavily influenced by table-top roleplaying games (RPGs), especially Dungeons & Dragons (D&D). Like board games, RPGs are designed with the advantages and limitations of human beings in mind. Specifically, we have awesome imagination, poor short-term memory, and don't like complex calculations. Conversely, computers are excellent at making calculations, have perfect memory and zero imagination. It should be obvious that CRPGs should be designed for the advantages and limitations of the computer. It also means CRPGs will never be real role-playing games unless they are completely multiplayer or someone develops a sentient AI.

I particularly don't like the "classes, levels and skill points" style of character development prevalent in CRPGs. It works great in action games like Diablo 2 where it's an excellent reward system for the player. In "real" CRPGs, not so great. Of course, most of the content in most CRPGs is action, and specifically combat, which is a separate problem itself... Some games, such as Morrowind, implement a "learn by doing" system, where you gain skill points in eg. swordfighting by fighting with swords. It's a step in the right direction, but Morrowind's implementation wasn't that good. It smells too much like spreadsheets with all the numbers, skill increase messages etc. and is easily abused (Morrowind was famous for players jumping around all the time to raise their strength stats). I think computers can do better.

I'm all for complex simulations with simple interfaces in (computer) games for a couple of reasons. First, computers are good at simulation, so why limit simulations to what a board game could do? On the other hand, people like simplicity to avoid their brains exploding, so you can't show all the complexity to the player. At least not directly. Second, complex simulations result in emergent gameplay, and emergency is awesome. Therefore, I'm thinking skills should be far more fine-grained and interrelated than they are at the moment.

Let me give an example. Practically every modern and sci-fi CRPG has weapon skills for different types of weapons. There might be eg. pistols, rifles and heavy weapons. But usually this means that if you can fire a pistol, you can't fire a rifle. And you can't be an excellent sharpshooter, but poor at hitting moving targets. An alternative could be "micro-skills" for aiming, burst control, hand-eye-coordination, leading (as in leading a shot), reloading, jam repairing etc, with "affinity skills" for different categories and even models of weapons. All of these would be improved by training or actual experience, and a combination of these skills would determine the actual result. It would be no problem for a computer to handle all this information, especially if all the micro-skills already match an action in the internal simulation, but it would be too complex to show the player. The game interface might simply read "you are good with rifles, and very experienced with the M-16". This is still just a rough idea, but I'd like to try and take it further by designing and perhaps even prototyping it in the future. And that will be the topic of my game design post.