While on a long drive with Kainenchen and Stands-in-Fire recently, we started talking about skills and skill challenges in 4e. At its most macro level, the skill challenge is one of the best ideas that 4e has added to D&D as a whole. I have seen skill challenges be an excellent source of non-combat tension that can be resolved with a combination of clever thinking and multiple dice rolls. (Before I get started here, it has to be noted that Rob Donoghue has done some really clever things here (and in several subsequent posts) and the next time I run 4e, I'll be playing with his ideas.)
The actual usage of skill challenges in-play is tough, though, and some of the problems specifically remind me of problems I had in previous editions. So let's talk about class balance between in-combat and out-of-combat situations, because no edition of D&D has gotten that right.
There are games where the focus of the whole game is on the narrative whole rather than the actual action. The games (Song of Ice and Fire comes to mind, as does every edition of World of Darkness-based games) are designed so that combat is when some players shine, and out-of-combat is when other players shine. It is possible to run a 3.x D&D game this way, though I can't recommend it: I'll note that a violent conflict can easily take 1-2 hours of session to resolve (and some examples stretch far beyond that; I have heard of campaigns in which a single battle took 8, 16, or even more hours to play through). That is a lot of time for one subset of players to have fun, and another subset of players to be observers. Even complicated non-combat conflicts seldom last much beyond half an hour. My premise: "my character contributed something useful to solving the group's problems" is where "feeling cool" comes from in tabletop games, particularly D&D (not so much in, say, Amber).
In terms of roleplaying, most GMs would love it if characters treated violence as an option only when all others had been exhausted, and looked at a broad range of ways to resolve various conflicts. That's great and all, but D&D's rules for combat are more interesting, and every character as some amount of combat ability. When all you have is a tactically interesting hammer that gives experience points when used... Anyway, I mention all of this to explain why I don't favor a split between combat and non-combat characters in D&D.
Skills in 4e have applications both in and out of combat, and players and GMs are encouraged to interpret them as loosely as possible. You'll recognize this technique from such games as Spirit of the Century and World of Darkness; it represents a big change in D&D away from the narrower definitions of skills (and larger skill list) in 3.x.
Some classes legitimately need a variety of skills to accomplish their core concepts. Rogues and bards are the main examples here, such that 3.x D&D gave the two classes wildly more skill points than any other classes. For 3.x, I'm willing to agree that rogues needed lots of skill points to cover the wide variety of skills that 4e rolls into thievery, acrobatics, and athletics. Where 3.x rogues got four times as many skill points as fighters, 4e rogues get only twice as many trained skills as fighters. This frustrates me in 3.x non-combat situations, because I'd like to know who decided that broad, general incompetence was a guiding theme of fighter archetypes. Even increasing my fighter's Int score wouldn't help this problem, because an extra skill rank every other level in one of those cross-class skills does not fit my definition of "help."
As I play a pretty decent number of fighters, this still bugs me. A fighter's best stat is going to be Strength followed by Dex, Con, or Wis, depending on his build. Fighters have no options for class skills that are Dexterity-based, and two skill options that are Charisma-based. Of the core classes, only the Cleric has as many of its class skills in skills relying on stats that the class doesn't otherwise use. Ultimately, this means that when it comes to a skill challenge, a fighter is skilled in very few things, and is not as good at the things he is good at doing. This probably has something to do with the otherwise-inexplicable decision to make scale armor the one type of heavy armor that doesn't impose check penalties.
Judging by the revisions to Skill Challenges introduced in the DMG2 (which do help, don't get me wrong), I feel comfortable guessing that setting DCs has been a bit of a trick. I'm going to do a little armchair design and guess that it has a lot to do with the ranges of scores that characters are likely to have. A starting rogue is going to have scores of +9, if not better, in his two most important skills (Stealth and Thievery) as well as (probably) Acrobatics; Charisma rogues are also going to be somewhere in the +7 to +9 range with the many Charisma-based rogue skills. As so few skills cater to a fighter's strengths, fighters are much more limited in their usefulness during skill challenges, and each individual effort is less likely to succeed.
This would be like a fight in which Player R(ogue) has six different attacks he can make, and he expects to do easy or moderately-difficult things 95% of the time, and has only a little trouble (75% success rate) with hard things. Player F(ighter) on the other hand, has three things he can do, one of which succeeds at the same rates as R's, while his other two things succeed moderate tasks 85% of the time and hard tasks 55% of the time - and those are the things they do well. F isn't going to have a decent score in Int or Cha, which cover most of the skills.
What I'd like to do about this is to smooth out some of the skill disparities. I think that the +5 bonus for trained skills is sometimes too much, particularly when combined with "this skill uses my good stat." I can think of a number of different ways to do this, and my ideas here have a couple of common features. The game needs a step between +5 trained and +0 untrained, I think, and for these purposes I'll call it "familiarity." I'm also proposing the introduction of weaknesses in some areas to counteract some of the added strength that this gives characters.
The cleanest of the progressions I worked out is as follows:
Three Trained Skills
If the core rules give your character three trained skills (before racial adjustments or feat expenditure), you have three skills at +4, chosen from your class list. After assigning these, assign a +2 to eight more skills (not the same skills for which you've already chosen bonuses). You must include the rest of your class skills in this list. After assigning these, pick one remaining skill to receive a -2. Everything else is untrained (+0).
Four Trained Skills
Four skills at +4
Seven skills at +2
Four skills at +0
Two skills at -2
Five Trained Skills
Five skills at +4
Six skills at +2
Three skills at +0
Three skills at -2
Six Trained Skills
Six skills at +4
Five skills at +2
Two skills at +0
Four skills at -2
Adjustments for Race and Feats
Humans and eladrin pick one additional trained skill: humans from their class list, eladrin from the full skill list. As a result, humans effectively receive one more +4 skill and one less +2 skill, while eladrin receive one more +4 skill and one less +0 skill. The Skill Training and Skill Familiarity feats go away, and are merged into one feat that grants +3 to a skill and can be applied to any skill regardless of its existing bonus. Jack of All Trades applies only to your +0 and -2 skills, making them +2 and +0 skills respectively. Racial bonuses to skills apply normally.
Why I Like This
This gives all characters a total of 26 points of bonuses (if you count the penalties), with some characters more specialized than others. In theory, every character should have considerable ability to join in on skill challenges, on more-or-less equal footing. Characters with six trained skills aren't better at the whole category of things we call "skills," they're just more specialized. That being the case, you could let players decide which of these they want for their characters: do you want to be a skill specialist or a skill generalist?
To give another impression of how this would change things, the current spread of "skill bonuses from training" runs from +15 (three trained skills) to +35 (seven trained skills) or even +55 (seven trained skills plus Jack of All Trades granting +2 to your 10 untrained skills). My proposal would change this to +26 on up to +40 (base +26, eladrin turns a +0 to a +4, little Jackie grants 5 skills a +2).
And Another Thing
Arcana is the best skill in the game, in my experience. This skill's application can best be summed up as "be a wizard at it," much like invoking your High Concept Aspect for everything you do in Dresden Files. In a quest to include everyone, a majority of skill challenge situations I've seen allow at least one application of Arcana. If not that, then History, without doubt. Admittedly, this may have a lot to do with the kinds of adventures and situations that our games have tended to see, but I think 4e is particularly high on "well, there's some nebulous magic stuff going on here."
Athletics and Endurance are, in combination, the "be a fighter at it" skills, Stealth and Thievery are the rogue's unquestionable go-to skills for any task, and clerics have Religion and... I dunno, maybe Heal. I feel like Religion really needs to be the full parallel of Arcana in its breadth of application, but as it is written it feels more like "just" a knowledge skill. This might be a matter of perception, given that divine-sourced characters have been vanishingly rare in games we've played. The other thing, though, is that Religion is based on Intelligence, which is a secondary stat for one Invoker build and one Avenger build, and that's it. Other than paladins, though, every divine class needs Wisdom, and I really have to ask why wizards trained in Religion are better casters of Religion rituals than clerics or invokers. I would like a cleric to feel like she is in every way as much of a divine spellcaster as a wizard is an arcane one.
So I'd like to take everything that Religion does right now, and make it do for clerics what Arcana does for wizards, particularly in detecting magic. I'd call it Piety, and I'd make it Wisdom-based. This would be especially useful in making divine casters the best at casting divine rituals, not just healing and druidic rituals. (Also good for letting druids and shamans handle religious functions.)