Skills have a bunch of problems.
The first and most direct problem is its binary nature. For most skills, when you have to make a skill check, you either success or fail. No middle point. This, combined with a linear nature, makes them pretty unreliable and unpredictable. When you add on top the lack of skill points, you end with one of the following results: either you overspecialize and pass regularly any skill check, or you don't place a single rank and hope you never have to make a skill roll related to that skill. Middle ground is bad, because not being able to make a roll on a 1 or more means, because of IP, that you'll fail sooner than later.
Another problem is the lack of a -real- measure of needed skill bonifiers for different tasks. There are a bunch of examples, but it is still guesswork to discover an acceptable DC for a given situation most of the time. And the fact that most of the time those guidelines DC aren't decently adjusted at all (such as with diplomacy) just makes this worse.
By the way, because of this last problem, non-optimized skills become even more of a no-no. Too many challenges are tailored only for the characters speciallizing in it. This is reasonable for many skills, because the system assumes specialization, but there are a lot of skills which should not have those problems.
This is related to other problem: the difference of improvement on different skills. Most of them increase in power lineally, while some other (very few others, in fact) scale quadratically. This said, this problem depends a lot on what do you expect from the skills. There are a lot of people who wants realistic skills. If that's the case, a level 15 character shouldn't be jump much further than a 1st level character, which is what, right now, the system supports. But this makes many skills not worth to invest in, because the improvement is only marginally benefical.
Another problem is the set of skills the dessigners chose. Too many skills are just redundant, or simply lack usefulness to be a skill by themselves. Rope Use is the worst offender, but Jump and climb have their own share of problems. In fact, AFAIK, jump became mainstream thanks to Battle Jump, Leap Attack and Tiger Claw maneuvers, and, if it weren't because of those, it's a skill no one would put ranks in.
Other one is the high variance due to magic items. Skill ranks only matter for prerequisites, so the only "real" value for skill usage is the total bonus. A +10 bonus to a skill is pretty cheap, compared, for example, to a +3 to attack. This makes the extra skill rank you can give to a skill whenever you level up feel pretty cheap, and makes it even more difficult to asses an aproppiate CD for skill checks. Most of the time, reaching an acceptable degree of proficiency with the skill requires either maximizing the skill and adding a lot of feats, class features and items, or just casting a spell or getting an appropiate magic item. In the end, the value of the ranks themselves is fairly low.
But the most atrocious problem with skills is that they aren't exactly too useful most of the time, and most of the time they just don't offer enough rewards for pursuing them, because skills suffer the same syndrome than the fighter against spellcasters: Too many times, Spells Do It Better.
For example, going back to jump and climb... why should you use them when you can just Overland Flight all the day? Or simply teleport to your target. Why use knowledge skills when you can cast Legend Lore or Contact other Plane?
This is not a problem on the lower levels, when the resources are scarce, but on the higher levels, too many skills lose their usefulness and start working just as PrC and feat requirement fodder. There are some skills worth putting ranks on the higher levels (UMD), but not many.
