(Note: I put a lot of weight on what characters can do, however it is couched. For purposes of clarity I'm going to talk about skills here, but by skills I really mean "those things that a character can do" so it applies as much to Storyteller or Cortex as it does to a Risus or Over the Edge, or even to some wackier games.)
As noted in my previous post, I'm a big proponent of acknowledging a character's capabilities as a way of making the character feel cooler to the player. A part of that is the idea of niche protection - every character has a niche or role that they play in the game, and when another character can also operate in that role, it can kind of suck. The classic example is the acrobatic thief and the mage - the thief may have really awesome jump, climb and tumble abilities, but if the mage can casually fly or teleport, then the thief can end up feeling useless. This is a situation to avoid, and when designing classes, powers or abilities I consider this a larger issue than simple "balance".
To my mind, skills really break down into two categories - baseline and outliers. Baseline skills are those you expect to see used commonly, possibly by every member of the group. Outlier skills are ones which might be limited to only one or two characters, and which provide opportunities to showcase the character's schtick. Both groups are important, since baseline skills help describe the game, while outlier skills help define each character. Identifying which is which can provide concrete improvements for your game.
Establishing the Baseline
In the show Leverage, each character has a particular specialty but they all have basic criminal skills - pickpocketing, lockpicking, running a simple con and whatnot. Similarly, in a police, military or espionage game there's usually a baseline level of capability expected, and characters can be distinguished for how they excel beyond those parameters.
A lot of games have at least implicit support of this idea. Sometimes it's generic - Over the Edge allows players to roll 2 dice for any action that would be reasonably within their experience, like driving a car or programming a VCR. Often it's more specific to genre. Games may assume certain default starting skill levels, for example, or price skills in ways to funnel competence into appropriate avenues. It can even be handled by the rest of the system: in D&D 4e the assumptions is that everyone can hold their own in a fight, and virtually every rule in the game bends in that direction.
Despite these examples, a lot of games don't address this at all. This is not a scathing criticism because this is usually done in service of player choice. The bulk of point based games start characters from a blank slate and require that you purchase everything up from zero. I have a great deal of respect for that sort of freedom, and it has a lot of advantages, but it has one concrete drawback - because you have to pay your way up to competence, it can feel like a tax on not sucking.
To illustrate, imagine I am playing police based game. As a player, I want to be able to do a certain set of things - lets say shoot, fight, detect and drive. There are other things like interrogate or knowing the streets I might want too, but they're fuzzier, since someone might choose one of those as their signature, so let's stick to the very basics. I will want to buy those skills up to some level of competence and there's a good chance that I've eaten a chunk of my points to do so, especially if I don't have good guidelines. As a player, this is a frustration tax. If I'm a novice player it's even worse since I may overlook some skill I need to reach the baseline, and only discover that gap in play.
There are a few steps you can take to help break out and handle your baseline skills in almost any game.
1. Simply lay out your expectations for baseline competence - what skills you expect people to have at what sort of levels and so on. If players want to be actively lacking in some area, that's great, but make sure they do it consciously. Not only will this be useful for your players, it will help you get a better sense of what skills you need to treat as outliers. Templates can be a great tool for this if they already exist, but don't stress that too much if they don't.
2. Figure out how much the baseline is going to cost your players and cut them a break. The simplest thing you can do is just give them the baseline for free or at some discount (say, half price) so you don't need to worry about someone needing to scrabble for points to pay for things that are mandated by your game (like, say, the merits/advantages that represent the resources and authority of law enforcement).
2a. If your baseline is fuzzy you can also just offer the appropriate skills at half price up to some threshold (so the first 3 levels of guns are half price, for example) and your players will usually just go with the flow. Usually. The problem with this approach is that you still need to deal with things you MUST have (like appropriate merits/advantages to reflect being a cop) and you run the risk of contrary players. The former can be addressed by just handing over the merits, but the latter is more of an issue of knowing your playgroup.
3. Reset your expectations. Depending on the game, you can treat all the characters like they already are at the baseline, and everything they're buying is over and above that. This mostly manifests in terms of what you don't call for rolls for. A common example of this is the Drive skill - what does it mean in your game if a character has no drive skill at all? It might mean that they don't know how to drive a car, but it might also mean that they can drive a car well enough to commute without getting in an accident, but not necessarily well enough for a high speed chase.
3a. This can be mechanically tricky and, to be frank, almost never works well with combat skills since those are hard to gracefully not roll for. It also demands respecting small investments - if a Drive of 0 reflects basic competence, then a drive 1 is noteworthy (as contrasted with 0 == ignorance and 1 == novice). The headaches in dealing with this are a big reason it's nice to just treat outliers as binary - everyone can drive, but bob is the car guy, and that's all there is too it.
Calling Out the Outliers
Once you've established your baseline, it's pretty easy to spot your outliers though the process of elimination. The question then is what to do with them.
Now, one solution to this is to make outlying skills exclusive. If one character can pick locks, then the pick locks skill is now off the table, and he is now the lockpick guy. This is especially satisfying for skills that are far from central to the game - if you're playing a western, then there's probably not going to be more than one guy with experience as a sailor, so if someone takes boating, he's the boating guy.
There's also a good case for weighting the skill prices based on this thinking (and, in fact, the Tri-Stat system does exactly this) - in this case, the guy who purchases "Boating: 1" actually gets "Boating: 4" (or some other arbitrarily high number or multiplier) so that he may very inexpensively claim a niche. Practically speaking, it will mostly be color because these outlier skills don't come up a lot, but when they do, it really puts the character's signature on things.
The best solution in my mind for handling this mechanically is as simple as treating certain skills as merits/advantages/stunts/feats/whatever, and this allows you to easily reverse-engineer this model into an existing system. For example, I could take a Storyteller or Cortex skill list and mark off a bunch of skills as "outliers". Anything not on that list gets bought normally. Anything on that list has a flat cost - something reasonably cheap, but which goes up with each one you buy - to get the skill at 4 dots or d8 or whatever's appropriate.
 - Hitter, Hacker, Grifter, Thief & Mastermind. Why yes, the writer - John Rogers - is a gamer, and he contributes to a wonderful blog over at Kung Fu Monkey.
 - Or sometimes, even within those parameters. While it is easiest to make a character distinct by giving them skills outside the common pool, like the cop who is a computer expert, it is also possible to work within the pool if they're really exceptional. In a police game, everyone can do detective work, but a Sherlock Holmes equivalent can have a niche by doing it at a much higher level. In a military game, everyone can shoot, but one guy might be the shooter. The one trick to keep in mind with this is that the difference needs to be so great that it's really a difference in type. If other characters could do just as well simply by rolling well, then this is an uninteresting distinction - it needs to be clear there are things that only this character could even attempt.
 - Competence is not defined by the color descriptions of the skills, it is defined by my likelihood of success when I roll. It does not matter if the color text says having a skill at level 3 makes me a "Skilled Professional," the real yardstick is the dice. This was one of the great frustrations of the old storyteller system, and one of my favorite things about the new one - competence no longer demands 5 dots.
 - This is a great example of something that can be baseline and an outlier - even if everyone in the group can drive (but haven't bought the skill), the one guy who has actually bought the drive skill is going to stand out when it's time to burn rubber.
 - The real purpose of this solution is that it works within the framework of existing games without needing any houserules in play, since it's purely a chargen hack. It is possible to do even more interesting things in play, but doing so makes it much harder to use existing material.