In my opinion, the perfect RPG is likely to occur in an online format rather than in a paper format. Players are readily available, mathematical assistance can be built in, and any number of graphical or literary aids can be distributed at essentially no financial cost and minimal time cost. Most current eRPG interfaces suffer from a clunkiness resulting from too strong a desire to 'stay true' to paper-based RPGs. In the physical world, rolling dice and shuffling disorganized papers is fun. In the digital world, sorting through needlessly complex windows on a less-than-panoramic monitor and clicking on poorly randomized dice icons is not fun.
That's not to say I'm advocating MMOs, console-RPGs or even MUDs over classic RPGs. These are dramatically limited (both meaninings) thanks to their static world interface and apathetic GMs (that is, GMs which necissarily cannot care about individual players). I think the classic RPG, 'stepped up' to the digital age, is the route to the perfect.
That said, here's some of my work.
I'm going out on a limb here, because I would like to publish some of the technical aspects of this material in a 'cleaned up' format, and it's hard to defend this sort of copyright. That said, publish this material in a recognizable format and I will own the company that does so. Not in some punkish 'revenge' way, but in a literal, "Hi, I'm your new boss," kinda way.
We'll start on the theoretical side.
Randomness is bad.
What? Yes. Having randomness in an RPG is inherently bad. It lets characters who should not be able to have a prayer of a chance succeed- which is fine in some games, but the flip side of this tossed coin is that it also means characters (both PC and NPC) who play by the rules can fail miserably at tasks they are theoretically in the top of their field for. Even worse is the 'critical success/critical failure' rule used in many games, where an arbitrary fixed number means failure or success no matter what the circumstances.
In D20, a character who rolls a 1 automatically fails, and most DMs interpret this as a failure that is especially catastrophic in some fashion. Take the 16th level full BAB character and the 16th level 'magic' or specialist character. On a 20-sided die, every roll has a 5% chance of failure. My statistics and probability is not good enough to give you exact numbers, but I'm estimating that means a character making four attack rolls a round has something like a 10.5% chance of 'critically failing' each round. The specialist, who is probably being compensated for fewer attacks with higher damage (for instance, casting Fireball-type spells which do all the damage the Fighter's 4 attacks do in a single blow), has half the chance of screwing up his attacks that the presumably battle-hardened veteran of hundreds of melee battles does.
However! There IS a certain amount of randomness in everyday life. I myself go to try things I know I should be able to do without too much of a problem and screw up miserably. Not all that often- and less so in fields I am more proficient in- but it happens. Maybe God doesn't play dice with the universe, but someone is.
My solution: Multiple dice systems are the best, because they establish a normalizing factor. My favorites are Exalted (d10s), Mechwarrior 3rd edition (it uses 2d10, modifiers based on skill), and Shadowrun. I actually hate the Shadowrun dice system for the most part, but it has a really good feature- I think that skill should be represented by allowing characters to reroll a failed check ("Wait, no, that won't work. I do it another way."), and Shadowrun's policy of one success needed for most tasks means that all the dice a character rolls are essentially rerolls.
Critical failures should decrease in probability as a character improves, while critical successes should have a baseline rate or possibly increase.
I have a couple ideas for this.
Levels are bad.
I'm sorry, but they're not real and they're not especially dramatic. Certainly, there are characters in reality or in drama that are 'at a different level', but this is because they've specialized and achieved some sort of very specific foothold into vast power. Bill Gates is not going to wipe the floor with me in a kung fu fight just because he's got a net worth I can't even imagine (although he can probably buy better guns than I can). Drama is filled with 'young upstarts' unsettling the tried and true because they're just damn good at what they do. Anakin Skywalker lived an adventerous and dangerous life before becoming Darth Vader, upon whence he continued to live an adventerous and dangerous psuedo-life. Yet Luke Skywalker held his own in a fight more than once- because Ben and Yoda trained him, and because 'The Force was with him.'
I'm not saying that the concept of levels needs to be entirely thrown out, but IF levels were to be kept, a new level should be attained because the character gets too damn good to be classed where he used to be, not that he should be promoted a level and caused to improve therein. In fact, such levels would probably be something to be avoided as much as they're sought- they'd slow XP gain or what have you if the character didn't live up to his new status. Additionally, levels should probably be linked more loosely- gaining a level of combat ability from being a dedicated fighter isn't necissarily going to cost (or affect in any way) the character's artistic advancement (although, as Musashi has shown, there are links between everything).
Skills improve through USE.
Bold and caps. You know it's gotta be big.
MW3 has this nifty little thing where if you roll a critical success or critical failure on a skill, you mark off a little box and that reduces the cost of improving it. Normally, it can't reduce the XP cost to 0, but in moments of contrariness, this can be houseruled away.
Although it's certainly possible to improve oneself through dedicated study, this is another form of use. Practice is just safer application.
I tend to like the way MW3 does it to an extent, because I support the abstract concept of xp; as pointed out above, there are links between everything. However, with enough field use, characters WILL improve despite themselves. Because of this, I feel that the amount of 'general' xp granted for a session or adventure should be very low, but occur in tandem with mechanically controlled improvement in areas the character is actually using.
I've got more, but I'll skip to two technical sections. This is the stuff I'm paranoid about. I'm actually very willing to sell my work, but I don't think anyone's buying just yet, especially since nothing's finished.
I have a few ideal dice systems. One is using 4 10-sided dice, for a perfectly random yet perfectly balanced generator of numbers between 0 and 36 (In most systems involving 10-siders, I interpret the 10 or 0 mark as a 0). This is a little awkward to balance out properly- even though I really like it, measuring everything by degrees of a circle can be difficult at times.
A simpler dice system would be using the standard one die per skill point, with a 10-sider (same deal). Each die represents a method the character can think of to undertake the task. Since a task with a difficulty of less than 1 doesn't require rolling any dice (if the modified target number is less than 1, it's an automatic success), 0s are always failures. Rolling all 0's is a critical failure; the more skilled a character is, the less likely this is to happen. There are no critical successes per sey, but instead, a success-grade system (like in Alternity or certain Storyteller system games) would be implimented. One idea I've had for this is allowing the character more leeway in describing *how* their success works the higher the grade- with a ridiculously potent success, they could essentially dictate the time period of their roll (insofar as it relates to them and their task). For instance, a character who is attempting to repair a console in the middle of a firefight might decide that their character is so completely 'in the zone' that bullets pass by and overhead, but never hit him or his work. This is a dramatic system- but then, real life has moments of drama too.
As part of my skill system project, I've written down all the general categories of skill knowledge I can think of. Notably absent is a Design category for things like architecture/engineering, and a number of other technical skills are hazy, possibly belonging to many or no categories. "Skill Knowledge" refers to the background mental abilities that direct the application of effort.
Persuasion (Persuasion, to me, seems to be the process of exploiting percieved similarities or differences. Thus, persuasion abilities would probably focus on comparing aspects of different targets- for instance, an intellectually sophisticated character might exploit this to get an 'in' with a scholar, or intimidating someone less astute. Alternately, a character could exploit another type of similarity or difference- for instance, racial categories (we stick together/they're not our kind) or special intrest group. A character skilled in persuasion is adept at polishing or even downright faking his abilities to seem to fit in better or stand out more.)
Tactics (This covers everything about combat that isn't actually taking pieces off of the enemy, and also business manuevering, chess, campaign management, and all the classically linked oppositions. Ranged combat skills might partially fall within this category.)
Athletics (Martial arts, including swordplay, also acrobatics, climbing, swimming, and other stuff people hurt their bodies with for fun.)
Aesthetics (The creation and appriciation of artwork, but also more immediately beneficial things like appraisal, disguise and identification of disguises.)
Physics (possibly scrapped into Chemistry and Mathematics)
Chemistry (includes drugs)
Bioscience (by default, I include surgery in here)
Mathematics (includes economics. May include computer science.)
Geopolitics (social sciences including geography and history. This represents understanding, but not necissarily the ability to apply that understanding to a strategic endeavor- a character specializing in a region's geopolitical skills fits in well there, but can't use that to particular advantage by default)
Linguistics (including applied linguistics, such as poetry and songwriting- music and tone arrangement is an aesthetic field, but putting a message in something or taking one out is a linguistic field)
Interaction (This may be subsumed into Tactics, Geopolitics and Persuasion, but basically means the grey area between the three. Bueaucracy and tea parties.)
Dexterity (Fine motor control. Dexterity isn't a stat, Agility is. Dexterity requires specific training. Prestidigitation (including picking pockets), possibly the physical process of surgery, and hanging upside down while picking up pins with one's eyelashes (the Houdini skill.)
Production (Operating tools to make things. Again, a grey area.)
Espionage (Security measures and countermeasures. This is a skill of limited usefulness in low-tech games, unless they have a magical theme, in which case it would be linked to the knowledge of magical effects and causes. It might need to be scrapped and replaced with a 'Technical Knowledge' type field depending on the game style)
Awareness (While I believe a stat should govern the 'in tune'-ness one has with their surroundings, Awareness relates to knowing where to look and how to look. Conversely, it also means where to sneak and how to sneak.)
That's it for now. This is my introduction, so to speak; I wouldn't say that I'm obsessed, but I've grown up so immersed in fantasy worlds that I think about this stuff even when I'm I don't think I'm thinking about it. Quality or flawed, it's natural to me.