I was in fact talking about "inherent" bonuses to BAB, saves, and skills. If you start talking about additions from magic, the delta gets worse. Magic weapons are limited by the rules to a +5 enhancement bonus. Skill bonuses start at +5 and amp up from there, and a +5 skill bonus costs about as much as a +1 attack bonus. Saves bonuses are likewise much smaller. Spell bonuses to skills obtainable at the same level outstrip bonuses to attacks and saves. There is a game design reason for this, and it's to keep up with the skillmonkey. But there's nothing stopping the skillmonkey from picking up a skill-boosting item to increase his dominance. Which means adventure design has to ramp DCs to keep up with the magic items and spells even faster than if you just look at the gains by character level. And adding skill points to everyone, or even just the non-rogues, doesn't help this. This can be fixed by jiggering the magic item rules and the spells, of course, but the magic item design for skills is set up because of the awesome variability of skills. Jump gives you a +10 bonus to jump checks, whereas magic weapon, a spell of the same level, only gives a +1 bonus. And the Jump spell ramps at 9th level to +30! These are all consequences of the skill system as it stands, with the huge variability of skill levels.
One of the complaints I saw (and I don't remember whether it was here or the WotC forums) is that SWSE skills sytem means we get superheroes, not heroes. We already have that problem, it's just that nobody notices because skills are often trivialized (you can run a campaign with skill importance, but the general cases in game desgn trivialize certain skills). And skills are trivialized because it's too hard to design for all cases when writing up a mass-market edventure; there has to be either a "path to success" for every possibility of skill sets from the maximum-single-skillmonkey in each skill to the possibility that the party might not have anyone with a single maxed skill, or even anyone with a particular skill at all. That means, among other things, there has to be an escape hatch that doesn't depend on game-mechanical skills (as opposed to pure roleplay) at all either "kill-them-all" or a macguffin from a previous encounter than allows the party to depress the DC of the skill check(s) low enough that an unskilled party can pass it. And if that skill-depresser macguffin is there, then the party with the skillset to pass the test beforehand can take advantage of it and not have to worry about the test at all.
Let me give an example of what I'm talking about. I'm an adventure designer, writing up a module to be published to the wide world. At a certain point, I have a primary encounter that the entire party needs to be at (a "boss fight", say). And I want to set up a set of barrier encounters to make the party work a bit to get to the boss fight. Say the "boss" is in a castle holding a party, and the PCs have to figure out how to get into the castle to disrupt the party and kill and/or destroy the influence of the host. I want to set up one of those "multiple path to victory" scenarios. The party is (nominally) 10th level, and this is calibrated to be a challenge worth XP for the party. Some of the paths to victory can require use of spells, but if the party doesn't want to use or hasn't got the spells prepped, there has to be non-magic-based challenges. What skill sets do I require, and what DCs do I set those challenges at? For the purposes of this example, I don't want a "kill them all" escape hatch. I have to pick skillsets that most parties will have, and I have to challenge both the single-skill full-maxed-monkey, the broad-skill half-maxed-monkey, as well as the party that has holes in their skill coverage that they are covering with magic. Lets say one of the challenges is a wall that has to be climbed, but there are other ways to get in. What do I set the climb DC at such that it is not both a no-brainer for the cat-burglar-concept character and too hard for the fighter who has been dropping skill points into climb every other level because he has to also balance it with Ride and Intimidate, because the party rogue has been going for social skills rather than athletic skills (or the rogue multiclassed to wizard a while back and his climb skill hasn't kept up, or what-have-you). Not only do I have to accoutn for the fact that one character can be +2, +3, or even +5 better than another who has been maxing the exacct skill depending on feat choice, but I also have to account for the skilled character who has decided to split his skill ranks between 2 skills (so is half-maxed in both). At 10th level, the maxed charater with both skill feats has a +18 skill mod before stat or magic boosts. If he picked up magic items to help him focus that skill further, it could be a +23 or +28. Whereas the half-maxed character (who we have to consider, since the big benefit of the 3.5 skill set is flexibility in assignment of skills) could only have a +9 check, and may not have chosen to pick up any skill-boosting items because he's a generalist, not a specialist. 2 characters who have been training in the skill, who are both much better than the rest of the party. And you can't even challenge both of them because the delta after magic items is right around +20 I certainly can't assume every party at 10th level has slippers of spider-climbing in adventure design; but a tthe same time I hae to take that into account.
Repeat across all skillsets, and you very quickly find out that depending on the party, either none of the skill challenges are actually challenging because the party can bypass them due to a skillmonkey with a high enough check, or they are all too hard, because there is nobody in the party with high enough skill checks. And anything I put in to make the lesser-skilled party have an easier time with the checks can also be used by the more-skilled party.
