I'm currently in the process of creating a new RPG system (really for homebrew purposes: The prospects of marketing it are rather slim, although I may freely distribute it if it comes to anything). The exact details of it are unimportant right now, except insofar as it's generic, character point based and uses a roll a d100 + add modifiers, high is good mechanic. It's also heavily skill-based. The following is the skill system I'd come up with so far, which I'd like some help on.

The first thing is the mechanic of buying the skill. A skill has two parts to it: Familiarity, and rank. Rank is just a modifier to the roll of using the skill, and costs a flat rate per +1 (with the flat rate depending on the skill in question). Familiarity is bought before you can acquire any ranks in it, and allows usage of the skill (defaults are possible without familiarity, which is where the Skill-tree).

I'm a fan of the oodles of skills approach that systems like GURPS takes. However, I acknowledge that it's an absolute pain to have to buy every single last skill, and fails to take into account things like general skill in an area. e.g. if one has a good general background in science it might be possible to know bits and pieces about something you haven't actively studied, and so someone who's really a chemist would probably be able to get by in biology or physics. Or an experienced melee fighter who might normally use a sword should be able to pick up an axe or a club and still wield it with some degree of proficiency. Thus I introduce more general skills as well: e.g. science and melee weapons in the above.

So, we say some skills are specialisations of others: Narrowing down to a more specific area. This is the skill tree. At it's base you have 'raw talent', which splits down into things like 'jack of all trades' and 'magic', splitting further into 'education', 'outdoors', 'ritual magic', 'spontaneous magic', etc. Everything above a skill on the skill tree represents more general knowledge that is applicable to that area, and the rank of the more general skill is added to the rank of the specialisation. (Obviously the more general the skill is the more the ranks cost).

However, one shouldn't just be able to buy familiarity with a general knowledge skill and suddenly be familiar with all it's subsskills. If you could, the price of familiarity with "Jack of all trades" would be obscene - many hundreds of times the points value of any reasonable character. Further, it would just be silly. What familiarity with a more general skill gives you is better defaults with its subskills. This gives the second part of the Skill-tree:

Every skill has a level associated to it. The higher the level the more specific the skill. If X is a specialisation of Y then the level of X must be greater than that of Y. If you have familiarity with Y but not X then you may use X as a default with a penalty equal to the difference in level between X and Y. For example, to pull numbers completely out of a hat, say 'Physical Activity' was a level 20 skill and 'Longsword' was a level 80 skill. You are familiar with 'Physical Activity' but not 'Longsword'. You may still use a longsword, but all your rolls will be at -60. (Remember, these are d100s). However, if you were familiar with 'Melee weapons', which has a level of 40, you would only be at -40 to your rolls. Better yet, if you were familiar with 'Swords' (level 60) your rolls would be at a mere -20. (These numbers are awful and wrong. I haven't asigned any levels or skill names yet. This is just an illustration of the idea).

This seems to me like a sane and relatively easy mechanic to deal with. However it has one major problem with it:

Defaulting to similar skills. e.g. if I'm skilled with a shortsword but want to use a longsword, how should this work? Surely some of my shortsword skill should carry over to the longsword, but how?

I've considered several possibilities (not neccesarily mutually exclusive ones), but they're both bad for one reason or another:

1. Backtrack to the nearest commonly held ancestor (e.g. swords). Take the level difference for each of the skills and then add them. (So if shortsword was level 75 then the penalty for the default would be 20+15 = 35)

2. Bonus skills. For each skill keep track of how many character points been spent on specialisations of it. Take 10% (or some appropriate number of this) as bonus points which must be spent on buying familiarity and ranks in it. Possibly with some fiddling that this only applies to the smallest common ancestor of skills so as to prevent this applying to several things from one group of skills and buggering everything up.

3. Metaskills: Like 2, but you don't really get these skills. It's merely an effective skill for the purposes of calculating defaults (which are then calculated as per the main notion).

The main problem with all of these is that they limite the customisability of the Skilltree. e.g. I would like it to be possible for a GM to add in the skills 'European swords' and 'Asian swords' as specialisations of 'Swords'. However in the above this would then change the default of a longsword to the shortsword skill by making them have a closer common ancestor (in 1), or would mean that you now have another skill that's getting the bonus points.

1 has the further problem of not taking into account the modifier for shortsword. 2 and 3 have the further problems that their clunky, hideously broken and inconsistent. They're really only included for completeness as to what I've tried.

If nothing else, defaults to similar things can be added in manually. This I feel ruins the point of the entire idea though, and I'd be forced to start from scratch. It seems to me like there should be some easy way to handle it within this system, but I can't come up with one.

Consider: you say that a general skill that could be used instead of some other skill would be obscenely expensive. If you're looking to limit character effectiveness, I say that this is not so. Let me explain:

Specialization is powerful. A character that is specialized in one thing will invariably get to situations where the specialization is useful. Moreover, the player will feel it more important to be really good in one thing rather than a hair over the mean in many.

Thus I suggest that you take the simplest possible method, and simply add different levels of skill to each other. So a character with the skill "Physical" at 20 will roll every and all physical activities with that bonus. If he has "Sword" at 10, he'll use a bonus of +30. Simple and efficient. To set the prices for different levels of generality in skills you'll just have to consider the likely situational structure of the game:

In a vast majority of games, I won't expect characters to really meet more than twenty fundamentally different situations. It's theoretically possible to top it, sure, but in practice most of all conflicts will happen in swordswinging or whatever. By knowing this you can simply set the price multiplier of "Physical" skills to, say, x5. Sure it's cheap if you expect to do more than five different physical things, but I'd be surprised.

To top it off, you can even consider different general skills independently: if the game is a dungeon bash, you know that "Physical" will indeed get some five or so different uses (avoiding traps, fighting, resisting poisons, climbing, etc.), but "Social" will only get the one or at most two uses (dickering, diplomacy). So use multipliers accordingly! Throw away all notion of "general" skill trees, and rather let the GM customize the price multipliers of different skills based on how many subskills they have. You can even track the points used on a skill independently; then when the players invent a new subskill for that skill, the multiplier can change and you can calculate a new value for the skill from how many purchase points have been used. Sure, the skill value will go down a little, but at the same time it has gained a new explicit subskill.

In practice this kind of system will be very balanced: in actual play the specializing characters will really shine, because they get situation-breakingly high numbers in their skills. On the other hand, the generic character can expect to get to do something the specializing character cannot. I have found no problem with overtly generic skills, rather the opposite: a player who fails to take a really special ability for his character will frequently feel left out, as others have taken all the best niches.

By the by, I'd suggest that you should consider the way you split your skills a little more carefully: it's a really bad idea to do a point buy system where you give the option of gaining combat skill by weapon specialization. This is much different than differing between, say, different sciences: by genre convention sword can always be used for melee, so the character will never, NEVER, need more than one extremely narrow specialization for combat, while he'll sure need more than physics to swing advanced biological work, as an example. Thus you have a disaster in the offing if you try to split skills by weapon. Or, alternatively, embrace the idea for all other skills too, and allow scientists to use any narrow specialization for anything remotely scientific. If that doesn't entice, I suggest differing between combat situations, not weapons: it's much more interesting to have an expert of ambushes than an expert of the culverin mace, anyway. And the former has some meaning for play, as all battles won't be ambushes, while you can use the culverin mace (whatever it is) in all battles.

And before M. J. Young does it, let me mention Multiverser: that game handles the skill tree by making specialization mandatory: a character would start with a general skill, but at certain level he would have to pick a specialization to continue hiking the skill. He would have to further specialize to get to really high levels. Outside his specialties he'd always roll the maximum general skill, unless he'd have multiple specializations and one of them would count.

I want to thank Eero for pointing to Multiverser. As he says, it handles specialization as an inherent part of skill ability improvement. To illustrate, you would have a general skill in something like "striking weapons", which would take you through the "amateur" level (from +11 to +20). Thereafter you would specialize in perhaps "morningstar" for the "professional level, which would give you significant advantage with that type of weapon (+21 through +30) while retaining your ability with "weapons kind of like this" because they are striking weapons (held at +20). At expert level, you're using one particular weapon, "the morningstar given to me by my father", as better than any other (+31 through +40), but still have excellent skill with other morningstars (+30) and other blunt weapons (+20).

The same concept applies to specialization across the board. It's subject to adjustment by the referee to fit the needs of play--you could have "driving" as the basic skill and specialize into narrower types (stunt driving, demolition derby, heavy equipment driving), or you could have these at the basic level and specialize further at higher levels (if for example you expect to have a lot of different kinds of driving situations in play).

Also, the game eschews point-based skill additions; this aids immensely in making play seem a bit more sensible. If a character wants to pick up a new skill, he has a chance to add it in play. Say our morningstar expert has lost his weapon and grabs an axe off the wall. This is an entirely unfamiliar weapon to him, but with one roll he can "teach himself" how to use this as an amateur (starting between +11 and +13, depending on the roll), and immediately add it to his skills list.

There are quite a few other aspects to the system which might be helpful in your efforts; it's probably worth it to take a look.

--M. J. Young

Wow. Thanks for some great suggestions, both of you.

Multiverser looks really intriguing. Unfortunately I have an almost complete freeze on spending on books at the moment (as I'm having minor cash flow issues and have already bought way way too many books), so I can't check it out just yet. However, I shall most probably use what I now know about it as inspiration. :) (Not directly lifting it mind you - I like it, but it's not quite how I'd prefer to do things).

Now, to address some particular points:

Eero Tuovinen, I didn't intend for all general skills to be obscenely expensive (on a brief reread I can't actually find in my post where I suggested that, but I'm probably overlooking it). I think I was suggesting Jack of alll trades would be. I will however definitely take into consideration what you're saying about the prices of general skills and reduce them substantially from what I had originally intended.

I agree with your suggested mechanic for adding skill levels. That's how I had intended it to work.

At first I was intrigued by your suggestion of making the cost of skills dependent on the particular setting and game. In the end however I'm going to have to reject it. There are two (related) reasons for this. Firstly, the style I'm aiming for is a 'one size fits all' game rather than a 'can be adapted to any setting' game. I'd like the rules to be completely independent of the setting you're in (except insofar as the setting may limit the availability of certain traits. e.g. no laser vision in a modern spies game, no high explosives training in a medieval game). Secondly, it ruins transplantability of characters. In principle you should be able to create a character and stick him into any setting in which he would possibly exist. Having variable costs ruins this because a character's cost now depends on the setting you're putting him into. It is however a nice idea, but I think not one that would work in this system.

I must admit, I don't really see a way of avoiding the problem you describe with weapon specialisation. I don't see a reasonable way to implement your suggestion of specialisation in Ambush or the like (or rather, to do so would have to be in addition to rather than instead of seperate weapon skills). Besides which, it makes sense to me that it should be easier to be generally accomplished at bashing things than it would be to be generally knowledgable of all sorts of science - the latter is a far more complex issue. Also bear in mind that the fighter has to buy all sorts of things that the physicist does not. They need to be fast, strong, able to survive damage, skilled at evading being hit, quite possibly skilled in the use of armour or a shield, etc. It isn't as simple as you suggest. The physicist on the other hand needs far fewer supporting skills and activities. However, at the end of the day I'm really inclined to answer this one with "Yeah, it's easier to break stuff than to learn quantum physics. That's life."

To the suggestions from both of you about Multiverser's mechanics:

Firstly, I like the idea of forcing specialisation. I wouldn't implement it exactly like that though. What I'm inclined to do is rather than force specialisation I would limit generalisation. For example, consider the following: You can't have more ranks in a skill than the skill's level. This should be relatively balanced as the skill level is already scaled so that the higher level skills are more specific, and the skill level is already on the order of modifiers (because it's being used to calculate defaults). The rank cost for generalisation can then be reduced a little further. This has the advantage of allowing some degree of generality for relatively little difficulty, but still allowing one to be a complete specialist if you so choose. Having done this I would then ditch the idea of cross-defaulting altogether. If you want to have some familiarity with using a longsword, you really should have taken some points in the sword skill as opposed to focusing entirely on shortsword.

On reflection, I don't think the above can work as I have it. First of all, it eventually caps (albeit at a rather high modifier). You can't really get more specific than, for example, skill with one exact sword. This then limits your skill in it to the sum of the levels of it and all the preceding skills. Secondly, it again suffers when you want to customise the tree by adding in intermediate levels - you can in the above example up your total skill with a wide range of swords by taking ranks in the new 'european swords' skill. I don't think it works well with the familiarity rules either, although I can't quite put my finger on why.

Perhaps I should just stick with the previous mechanic of encouraging specialisation simply because if you're putting all your points into generic skills you're going to suck compared to a specialist.

Actually, another reason to ditch the above idea is that it's imposing a style after the fact. I'm working on the principle that rules which are essentially tacked on to get a desired effect (in particular balance) are to be avoided if at all possible. The style should be inherent to the basic rules, not something added on.

Now, a question of mine:

What do you think of familiarity? I'm rather attached to it, but it's recieved some objections. I like it because it for several reasons. Firstly, it seems to me there is a qualitative difference between knowing and not knowing a skill. Thus the first rank of a skill should in some way be 'special'. Secondly, it seems to capture a nice feel of a character's general knowledge. Thirdly, it makes it difficult to use skills you don't know without having to buy ridiculous numbers of ranks in them before you actually can use them. Finally, it provides something of a uniformity to skill DCs. DCs can be kept relatively constant while instead fiddling with the familiarity cost to provide a certain base level of difficulty.

I'm not going to try and recount the objections, as to be honest I didn't entirely understand them and probably wouldn't do them justice.

My brother suggested an intriguing possibility earlier regarding defaulting to similar skills, which is that default skill ranks are the wrong way to look at it.

You're not mystically acquiring knowledge of how to use a longsword from what you know about a shortsword. What you are doing is using a longsword as if it were a shortsword. So rather than having rules about defaulting this would be a special case of rules for using non-standard or inferior equipment (which one would want to have in place anyway).

Another way of doing it is to realise that skills are based on other skills. It's possible to assemble those sub-skills into other configurations on the fly and create a skill when it's needed in one's own head. For example, in my own experience, I've created a "shooting rifle skill" on the fly from my skill in being still, holding things firmly, knowledge that a bullet takes some time to travel and drops in it's flight and I developed a skill accuracy of around 70%, hitting the target with 7 bullets out of 10 fired, plus I knew why I missed with each of the three misses. On the occasion, this proved equal to an expert shooter's own skill in rifle shooting, 70%, in the same situation. My total experience in fire arms before this point was firing two bullets about 30 years ago, with no training.