Skill Trees, and related problems.

[Lance D. Allen]

Alfryd,

As someone who likes "realism" in his RPGs as well, I'd like to weigh in on the topic you just brought up.

To address your D&D example specifically, it's actually not at all unrealistic. Looking at what Constitution affects in D&D, it is quite possible, and even likely, for person with a high strength to have a low constitution. In our modern day world, it's even common. Consider the strong fat guy, or the body-builder that neglects aerobic fitness, and concentrates on foods that will bulk him up, rather than make a balanced healthy diet. Both of these guys can be exceptionally strong, but will never be able to run a marathon, and may be susceptible to illness. Now, "hitpoints", which reflect the ability to take a beating and keep going may be tied to body bulk, which can be related to strength, but this isn't necessarily true, even in D&D. hitpoints may as easily be reflected as the ability to roll with a punch, which would be more likely to tie into dexterity.

Now, this isn't to say your answer isn't a good one, so long as it serves the goals you're going for in your game. I just wanted to briefly address this topic.

To address this line, now:

Any skill could have any other skill as a 'key attribute'.  (Maybe even several, if it's performance was dependant on/benefitted from multiple factors.)

What do you mean by this? Are you saying that you can, in your system, choose to have fencing be descended from knowledge: general? After all, it could be argued that you learned to fence by studying the written works of the sword masters, which led from your more generalized thirst for knowledge?

[Alfryd]

Thank you for the feedback.

. Now, "hitpoints", which reflect the ability to take a beating and keep going may be tied to body bulk...

Hit points, in my experience, don't translate reliably to anything meaningful in realistic terms, and I'm not fond of them as a game concept.  If, however, one were to define them as 'ability to withstand direct physical punishment', then yes, improving muscle mass will certainly boost your hit points.

Ability to 'roll with a punch' corresponds to agility, flexibility, and combat experience, and should be modelled appropriately using combat skills such as dodge, fortitude(injury,) etc. etc.

Looking at what Constitution affects in D&D, it is quite possible, and even likely, for person with a high strength to have a low constitution.

Yeah, but you've just carefully explained all the secondary factors which would negate the otherwise normal correlation between developing strength and developing fitness- being overweight, having an unusual diet, etc. etc.  In addition, you may be confusing genetic aptitudes with growth through practice.  The majority of normal training activities that improve strength will also improve fitness.  Not as efficiently as direct training in those areas, but all else equal, the link stands, and should thus form a basis for normal rule mechanics.

What do you mean by this? Are you saying that you can, in your system, choose to have fencing be descended from knowledge: general? After all, it could be argued that you learned to fence by studying the written works of the sword masters, which led from your more generalized thirst for knowledge?

An intriguing suggestion, but I was thinking of cases such as the Archery skill, which might be tied to both Artifice and Acuity, and Gymnast, which might be tied to both Jump and Balance.  That sort of thing.

I've been considering whether it mightn't be simpler to abandon the notion of unique parent skills entirely, and just list 'related skills' that confer half their level as a check bonus.  Narrower skill sets could be balanced by a lower XP cost to develop, so you wouldn't have to worry about stacking their value relative to more general skills.  I think FUDGE uses something similar, but, (as with most things in FUDGE,) the mechanics are either open to a fault, a tad subjective, and/or highly informal.

[David Berg]

Hi Alfryd,

I too like skill-tree realism.  Here's an idea for a point-buy system.  If that's not an option for you, this won't be usable, but... maybe food for thought?

Basically, the idea is that buying and advancing a broad ("parent") skill is expensive, and buying a specific ("child") skill is cheaper.

System 1:
Let's say you have 12 points to spend.

Sleight of Hand costs 4 points per rank.
So, you can spend your points to get rank 3.
You can use Sleight of Hand to do various things (including pick a lock) wth a +3 on your roll.

Pick Locks normally costs 2 points per rank.
You can take Pick Locks without taking Sleight of Hand.
So, you can spend your 12 points to get rank 6.  That's a +6 on your rolls to pick locks only.

However, if you take Sleight of Hand (parent) and Pick Locks (child), the costs you pay for Pick Locks are now only to advance it beyond your rank in Sleight of Hand.  So if you pay 8 points for 2 ranks of Sleight of Hand, you can still pay 4 points for 2 more ranks of Pick Locks.  Thus you wind up with a +2 to any Sleight of Hand roll and a +4 to specifically picking locks.

System 2:
Sleight of Hand cannot be used to pick locks, but having Sleight of Hand reduces the cost of Pick Locks for that character from 2/rank to 1/rank.  So those 4 leftover points (after spending 8 on Sleight of Hand) can be used to buy 4 ranks in Pick Locks (instead of the 2 ranks they'd buy without Sleight of Hand).

System 3:
Combine 1 & 2.  8 points buy 2 ranks in SoH, 4 points buy 4 more ranks in PL, producing a total of 6 ranks in PL.

System 4:
Any of the above, but with an addition:
Spending points in PL gives you half that many points' worth of SoH.  E.g., spend 8 points in PL, get 4 points' worth of SoH (1 rank).  This means that your lock-picking expert necessarily has some ability at other Sleight of Hand tasks, which seems realistic to me.

Hope this wasn't off-topic...
-David

[Velcanthus]

Wolfen has suggested that seperate skills for fencing and rapier proficiency may be redundant, but that can probably be fixed.

It seems to me that this can only be done if you create a definite body of skills, each of which has the same impact on the game as any other.

What I wanted to redress was the problem that, in many RPGs, two closely related skills/attributes are essentially treated as largely independant. 

I think it's pretty obvious that the various stats that have been used in most rpgs are impoverished. They would not even be able to effectively model someone walking down a road. I suspect, however, that they work because they provide players with a point of difference. They can say that they are strong but clumsy, or smart but suffer from a frail constitution, and so on.

It seems to me that the value of a statistic in a game is mostly so that the player can get a handle on what their character is like, rather than to be capable of generating a bonus accurately . Perhaps it's unlikely that a character would be strong and unhealthy. But, it is easier to grasp, and that makes it easier to play.

I don't think this is just a function of 'stats'. I think it applies to skills or any other special ability that a character might be said to have.

I suspect a game that is real-world accurate is not as attractive as a game that lets a player draw their character with ease.

[Alfryd]

Hope this wasn't off-topic...

No, it's quite germane.  Thanks David.  I'll have to consider the system suggested and get back to it later.

It seems to me that this can only be done if you create a definite body of skills, each of which has the same impact on the game as any other.

I'm not sure I understand exactly what you mean.  Presumably every skill, by virtue of having different effects, will have a different impact?  Are you talking about balancing their relative values, or do you just want to see more specific examples?

It seems to me that the value of a statistic in a game is mostly so that the player can get a handle on what their character is like, rather than to be capable of generating a bonus accurately . Perhaps it's unlikely that a character would be strong and unhealthy. But, it is easier to grasp, and that makes it easier to play.

I should perhaps emphasise that this skill structure is intended to accomodate skill development, without, for the present, considering the character's inherent genetic aptitudes.  I.e, nurture, rather than nature.
What struck me, for instance, as a major deficiency of D&D is that attributes are difficult to develop significantly after character creation (you get a 1-point bonus every 4 levels, IIRC.)  But things like strength, fitness, flexibility and even intelligence or charisma can all be trained extensively and improved through practice.  (Of course, they should also decay or atrophy through neglect, but that's another subject.)

I suspect a game that is real-world accurate is not as attractive as a game that lets a player draw their character with ease.

I'm not sure I really understand your criticism here.  Is there something inherently difficult to understand about general vs. specific skill sets?

[Velcanthus]

It seems to me that this can only be done if you create a definite body of skills, each of which has the same impact on the game as any other.

I'm not sure I understand exactly what you mean.  Presumably every skill, by virtue of having different effects, will have a different impact?  Are you talking about balancing their relative values, or do you just want to see more specific examples?

Balancing their relative values.

It seems to me that the value of a statistic in a game is mostly so that the player can get a handle on what their character is like, rather than to be capable of generating a bonus accurately . Perhaps it's unlikely that a character would be strong and unhealthy. But, it is easier to grasp, and that makes it easier to play.

What struck me, for instance, as a major deficiency of D&D is that attributes are difficult to develop significantly after character creation (you get a 1-point bonus every 4 levels, IIRC.)  But things like strength, fitness, flexibility and even intelligence or charisma can all be trained extensively and improved through practice.

How would this system provide a cure for that particular deficiency?

I suspect a game that is real-world accurate is not as attractive as a game that lets a player draw their character with ease.

I'm not sure I really understand your criticism here.  Is there something inherently difficult to understand about general vs. specific skill sets?

I agree that it seems unlikely that a person might not considerably increase their properties with diligence. However, it seems to me that a skill-based system is just at odds with reality as D&D. After all, you can't expect to increase your chance to hit an opponent with a sword once you have worked out how it works.

Beyond weapon familiarity, what determines whether or not you scored a hit is based on your ability to predict where your opponent will be. And, your willingness to hit them. Surely, that is not a function of any particular weapon skill. I've been in a few bar fights. The guy who could juggle snooker balls never worried me. The guy who was quite happy to throw them at people gave me grave concerns.

[Alfryd]

@Velcanthus

Balancing their relative values.
How would this system provide a cure for that particular deficiency?

Well, because things like Brawn, Logic, Vigour, etc. are all treated as skills that can be improved through practice like any other.  Presumably balancing can be handled (at least partly) by differing XP/buy point costs.

However, it seems to me that a skill-based system is just at odds with reality as D&D. After all, you can't expect to increase your chance to hit an opponent with a sword once you have worked out how it works.

I honestly can't see where you're coming from here.  Are you saying that a skilled swordsman would fare no better in combat than someone who'd never picked up a blade before?

Beyond weapon familiarity, what determines whether or not you scored a hit is based on your ability to predict where your opponent will be. And, your willingness to hit them. Surely, that is not a function of any particular weapon skill. I've been in a few bar fights. The guy who could juggle snooker balls never worried me. The guy who was quite happy to throw them at people gave me grave concerns.

I think this relates more to morale effects and knowing who's likely to start a fight, than to skill in combat per se.  But, just for the moment, if we assume that this is true, then you'd just have a general Combat skill, with particular styles of fighting (perhaps with particular weapons) as child skills.  Are are you saying that weapon proficiencies should be modelled as one-shot feats, or that their contribution is relatively small (entailing high XP cost to slow levelling?)


@ David Berg

However, if you take Sleight of Hand (parent) and Pick Locks (child), the costs you pay for Pick Locks are now only to advance it beyond your rank in Sleight of Hand.  So if you pay 8 points for 2 ranks of Sleight of Hand, you can still pay 4 points for 2 more ranks of Pick Locks.  Thus you wind up with a +2 to any Sleight of Hand roll and a +4 to specifically picking locks.

That seems fairly close to what I had in mind.

Sleight of Hand cannot be used to pick locks, but having Sleight of Hand reduces the cost of Pick Locks for that character from 2/rank to 1/rank.  So those 4 leftover points (after spending 8 on Sleight of Hand) can be used to buy 4 ranks in Pick Locks (instead of the 2 ranks they'd buy without Sleight of Hand)...
Combine 1 & 2.  8 points buy 2 ranks in SoH, 4 points buy 4 more ranks in PL, producing a total of 6 ranks in PL.

Actually, that presents an interesting solution to the 'ancestor stacking' problem.  Instead of boosting rank directly, parent/support skills might simply make a given skill much easier to learn.  So, you could pump XP into the child skill, and use the more general parent skills (perhaps at a penalty) in the meantime.

Spending points in PL gives you half that many points' worth of SoH.  E.g., spend 8 points in PL, get 4 points' worth of SoH (1 rank).  This means that your lock-picking expert necessarily has some ability at other Sleight of Hand tasks, which seems realistic to me.

An intresting notion, but shouldn't there logically be a kind of cascade effect back up the skill tree?  (i.e, 16 ranks in Fencing grants 8 ranks in parry grants 4 ranks in Melee grants 2 ranks in combat grants 1 rank in Artifice...)
I'll have to think about it.

I don't suppose anyone has suggestions about how to handle skill scaling?

[Velcanthus]

@Velcanthus

However, it seems to me that a skill-based system is just at odds with reality as D&D. After all, you can't expect to increase your chance to hit an opponent with a sword once you have worked out how it works.

I honestly can't see where you're coming from here.  Are you saying that a skilled swordsman would fare no better in combat than someone who'd never picked up a blade before?

No, but neither do I think it contributes a lot to combat effectiveness, especially as the character progresses.

Beyond weapon familiarity, what determines whether or not you scored a hit is based on your ability to predict where your opponent will be. And, your willingness to hit them. Surely, that is not a function of any particular weapon skill. I've been in a few bar fights. The guy who could juggle snooker balls never worried me. The guy who was quite happy to throw them at people gave me grave concerns.

I think this relates more to morale effects and knowing who's likely to start a fight, than to skill in combat per se.  But, just for the moment, if we assume that this is true, then you'd just have a general Combat skill, with particular styles of fighting (perhaps with particular weapons) as child skills.  Are are you saying that weapon proficiencies should be modelled as one-shot feats, or that their contribution is relatively small (entailing high XP cost to slow levelling?)

I think that modeling this realistically means that weapon familiarity gives you a reasonably large starting bonus, which quickly plateaus. It would behave like a logarithmic function. However, the ability to predict what someone is likely to do is somewhat more linear. This is not a morale effect. This is, I suppose, the general property of someone knowing a lot about combat, as you say.

If I were to model this, I would let the combat skill advance linearly at, say, 1:1. However, for bonuses that contribute to weapon familiarity, I would do something like this:

skill	bonus
1	1
3	2
6	3
10	4

But, then, I would only be doing this because I was interested in realism. It seems to me that the most important job that a game can do is provide a means of describing things to everyone involved in a clear and concise manner.

People who can model human interactions with the world accurately would not be designing rpgs. They'd be working for NASA.

[Alfryd]

I think that modeling this realistically means that weapon familiarity gives you a reasonably large starting bonus, which quickly plateaus. It would behave like a logarithmic function.

Ah.  I can see what you're getting at now.  ('Morale effect' refers to "willingness to hit them.")  What about parent proficiencies for broad categories of weapons?  Would you say these stack with the child bonus before diminishing returns, or apply seperately?

But, then, I would only be doing this because I was interested in realism. It seems to me that the most important job that a game can do is provide a means of describing things to everyone involved in a clear and concise manner.

This doesn't strike me as a major difficulty provided the mechanics involved aren't mandatory aspects of the game.  I mean, the rules for weapon proficiency won't concern you unless you take the combat skill AND a particular proficiency skill.  Any given player only needs to concern himself with rules pertaining to his character's area of specialisation.  My main concern is streamlining the basic mechanism for skill progression, which, I agree, should be kept relatively simple.

People who can model human interactions with the world accurately would not be designing rpgs. They'd be working for NASA.

This doesn't mean you can't come up with a reasonable approximation for a given degree of complexity.  There are areas where strict adherence to realism isn't vital for me- such as the incidence of combat-related fatalities, for obvious reasons- but I would like to keep results that are at least plausible, if not probable.

[Velcanthus]

What about parent proficiencies for broad categories of weapons?  Would you say these stack with the child bonus before diminishing returns, or apply seperately?

I'd just lump all contibuting bonuses together and push them through the aggregate series.