I'm trying to figure out the skill system for my game Channeling (a brain dump format of almost all my ideas can be found at http://shadowedmagi.proboards55.com/index.cgi?board=channeldev).
The game is going to be primarily skill based. There will be attributes and the like, and possibly some racial abilities (although i doubt it), but in general, everything will be a skill. Advancement system will be levelless either advance by use or XP purchasing (similar to White Wolf). Also, the current idea for die type is d100% system, although that isn't decided yet... but thats a question for later.
MY original idea was to simply have skills be a series of lines. For example, (for some reason, i always use this same skill as an example.. eh, oh well) the Thieving skill might have been arranged so that when you first got it, you got the ability to pick pockets, if you increased it, you got pick locks, etc, etc...
Now, i'm thinking that this is needlessly reestricting and hard to balance. An alternative was just to have all skills purchasable individually, but for one thing, i wanted specialization to have advantages, and that would also make advancement by use a bit more complicated.
My current idea is maybe having general skills be, for lack of a better term, containers for specific skills. Every time you increased the container, you chose a new sub-skill, and all previous sub-skills increased. Say i have the container "Stealth" with the sub-skills "lock picking", "sneaking", and "pick pocketing". I first get Stealth, and choose sneaking, and i gain a 5% bonus with sneaking.After a few adventures, i manage to increase my Stealth skill. I then choose lock picking. I now have a 10% bonus with sneaking, and 5% with lock picks.
My specific question is, anyone see any problems with that? any advantages i'm missing with my other ideas? and any other alternatives that keep the feeling of learning skill groups but not constraining to skill lines?
Thanks in advance.

I was working on a skill system that sounded a bit like this a while back, but I never finished it. It work by having player's buy ranks in a individual skill. One XP for the first rank, two XP for the second rank, etc. Alternatively, the player could buy a rank in a skill group of a "whole sale price." This was set slightly lower then the number of individual skills in the group. Additonal skill group ranks could be bought at the whole sale price x2 for the second rank, the whole sale price x3 for the third and so on. This made it so a character could learn a group of skills more easily or specialize in a specific skill if they wanted to. However in the case of specialization, it would eventually be in the interest of the character to buy group ranks, because every X ranks of specialization the whole sale price will be cheaper. So a master swordsman would eventually be able to wield axes and staffs, though not to the same degree.

I can send you the skill list and whatever of the write up I got done, if you want.

Lewis, while your system doesn't seem to be exactly what i want, ideas are always welcome, so i would love to see it if you don't mind. You can send it to my email in my profile, or any other way you want to get it to me, i suppose...

Containers
It would probably be better to use more than one 'container' for skills. That is, containers within contairs. Why, lets just have done with it and call it a Skill Tree.

You could have skill tiers- the further down the skill tree, the less it costs. This means if you buy 40% in Rapier, 10% in Parry, etc it is much cheaper than buying 50% in Fencing (Or, indeed, in Armed Combat).

Alternatively, the higher up the skill tree the skill is, the less useful it could be- perhaps it can only get up to to 10% in tier 1 (in the above example, buying 10% in learning would give you +10% towards all learning rolls, even if another skill is involved- but its expensive to buy 10% of it, much cheaper if you want to specialize), 25% in tier 2, etc.

Or combine the two of those.

Example
Hannah the History Student
She buys the Tier 1 skill Academics at 8% for 80 points (10 points each)

She buys the Tier 2 skills (5 points each):
15% History 75 points
10% Geography 50 points
5% Language 25 points

She buys the Tier 3 skills (2 points each):
22% European History 44 points
19% Medieval history 38 points
16% Middle Eastern History 32 points
12% Muslim History 24 points

She buys the Tier 4 skills (1 point each):
40% History of the Ottoman Empire 40 points
32% History of the Republic of Turkey 32 points
20% History of the Byzantine Empire 20 points
15% History of Hungary 15 points
12% World War 1 12 points

She buys the Tier 5 skills (1/2 point each)
20% Life of Suleiman the Magnificent 10 points
6% Life of Ahmed III 3 points

Her character costs 500 points

Wendy and Hannah are sitting under a tree. Wendy mentions Turkey.
Wendy: Did you hear Turkey is trying to get into the EU?

Hannah rolls Academics (8%) Current Events (10%) Republic of Turkey (32%) = 55. The GM decides that this is not 'History', so her knowledge of european and Middle Eastern history doesnt apply- but that her knowledge of the republic of turkey is specific enough to count.
She gets a 95. This exceeds her skill, and is a fail (A very bad fail).

Hannah: What? Are they?
Wendy: Oh yes. You wouldnt think it, a muslim country.

Hannah rolls Academics (8%), History (15%), Muslim History (12%), Republic of Turkey (32%) = 67
She gets a 06, a huge success.

Hannah: Why wouldnt a muslim country get into the EU? Turkey is quite progressive.
Wendy: Oh. I didnt mean that. I mean, arent they in the middle east? Isnt that Asia?

Hannah rolls Academics (8%), History (15%), European History (22%), Republic of Turkey (32%).She cant use her skills in Middle Eastern or Muslim history, as they are in the same tier, so she picks her highest.  82. It would be hard to not know this!
She gets 15.

Hannah: Actually the capitol of Turkey, Istanbul, is on the very border of Europe and Asia.
Wendy: Istanbul? Wasnt there a song about that? 'You cant go back to Constantinople, because its Istanbul?'

Hannah would make two rolls, one to see if she knows the song, another to see if she knows the history. However, Hannah knows NOTHING about popular culture, and so has an automatic failure on that. She rolls Academics (8%), History (15%), European History (22%), Republic of Turkey (32%), Istanbul (20%). 97.
She gest 46.

Hannah: Ive never heard the song. You cant go back to Constantinople because Attaturk renamed it the city to Istanbul. Constantinople was what the Byzantines called it.

Of course, this approach is very bloated and given the other systems you have in the game, probably not suited. You would likely want a rather small uncomplicated skill tree, not the sort of breadth this one implies. It might give you a few ideas though.

On the overall system
Any system like the sort of structure you outline (And especially the one i outlined!) is rather restrictive. With skill categories and sub-skills like that (even if they arent, strictly, trees) there is a major emphasis on progression and advancement and number crunching. Its not a system i would use. Its okay if thats what you want though.

Advance by Use
If this is an advance by use, what defines a 'use'?
Could a player spend 3 hours picking a lock just for practice? Might get boring for non-combat skills. It might also be hard for skills which are really useful, but only used every now and then. Also might lead to combat skills advancing much faster than noncombat skills depending on your games emphasis.

Huh... While your right it would need some tweaking (unless it had the entire system built on top of it), i like that idea a lot... thank you.
The system i'm trying for is mechanical enough ( "crunchy"?) that that would be a bit much for everything, but at the same time, i was planning on having skills be the main mechanic, as in basically everything from attacking, defending, magic, and everything else are all skills. So... the idea might not need as much tweaking as it looks like...

As for your second point, i know it will be at least somewhat restrictive by nature. I want players to be able to heavily customize their character, but at the same time, i want specialization to be rewarded somewhat.

And the question about "advance by use"? I'm not completely sure, which is why i haven't decided if i'm using it or not. If i did do it, the requirements for a 'use' would include something about it needed to matter (there has to be a penalty to failing). And also, different skills would advance at different rates. The 'pick lock' skill would take less uses than the 'sword' skill.