A couple of recent threads - one in this sub-forum, and one which got moved to Meta - have talked about noncombat resolution and skill challenges. This has come up in the context of what 4e players might be looking for from D&Dnext.
It seems that people are looking for a range of different things out of noncombat action resolution mechanics. What I primarily want is mechanics that will produce unexpected results in play, and that don't require the GM to pull punches. Which is to say, in these two respects at least, I want noncombat mechanics to resemble well-designed combat mechanics.
I don't pretend to know the full range of non-combat resolution mechanics for every version of D&D, but I will assert that they are not that thick on the ground.
In classic D&D (B/X, or the 1977-79 AD&D books), the only conflict resolution rules in are combat ones (including rules for evading/escaping from combat). There are reaction rolls, but these don't actually go as far as the resolution of social conflict. There are task resolution rules for perceiving various things, but when these get turned into conflict resolution mechanics they are limited to the surprise rules for combat (and this is one source of mechanical difficulty in those games - for example, thieves have a move silently ability and rangers don't, but rangers and not thieves get a bonus to surprise - what does this mechanical difference actually mean in fictional terms?).
There are no social conflict resolution mechanics that I know of in 3E other than the Diplomacy rules. There are rules for allocating character build resources to crafting, professinal and performance skills, but other than the mundane crafting mechanics (which I personally don't regard as very signficant to core D&D play) there are no conflict resolution rules that I'm aware of. To put it another way - the rules tell me that a +10 Performer is a better artiste thatn a +8 Performer, but there are no rules for persuading an aggressive owlbear not to attack by calming it with song (unless you start using bardic magic).
My own view, based on this udnerstanding of prior editions, is that skill challenges are the most robust non-combat conflict resolution mechanic scene in D&D to date. (I think they are far superior to 3E's Diplomacy rules, for example, which aren't really action resolution rules at all but more like scene-reframing rules: the player gets to tell the GM, "Instead of an encounter with an unfriendly NPC, let's have an encounter with a friendly NPC!") They are a scene-resolution mechanic very obviously based on similar sorts of mechanics found in a number of other RPGs (the ones I know are Maelstrom Storytelling, HeroWars/Quest and Burning Wheel, but I'm pretty sure that these don't cover the field).
There are points of technical detail one could quibble with in comparing skill challenges to those other systems, although as I recently posted here the skill challenge as a mechanic actually has some strengths over (say) the Duel of Wits from Burning Wheel.
4e also makes it hard to get the maths of skill challenges right, because it has very divergent and escalating skill bonuses. Hence the repeated revision of suggested DCs, plus the somewhat ad hoc "advantage" mechanic in Essentials. But this issue doesn't go to the structure of skill challenges - problems with the maths caused by scaling is a general problem for 3E and 4e D&D, which it is one aim of D&Dnext to correct.
For some examples of skill chalenges that I've run, you can look here and here. The basic dynamic is very similar: I as GM frame the situation, the players engage it via their PCs, skill checks are made and resolved, I renarrate the situation in light of that, and the process continues until either N successes or 3 failures is reached. The trick to the narration is to (i) keep the scene alive, so that the players continue to engage, but (ii) be able to bring it to a close at the requisite time. A good sense of both the evolving fiction, plus various complications that can be introduced to push things in the appropriate direction, is important to running these encounters.
There is also the need to apply "genre logic" in adjudicating players' declared actions (as per the brief discussion on DMG p 42 - Robin Laws's discussion in the HeroQuest revised rulebook is better). For example, a recent brief skill challenge I ran pertained to the reforging of a dwarven thrower artefact, Whelm, as a mordenkrad rather than a warhammer. At a certain point in the challenge, Whelm was thrumming with magical energy, and the dwarven artisans were having trouble physically taking hold of it with their toos. The player of the dwarven fighter-cleric overseeing the process asked if he could shove his hands into the furnace to hold the hammer steady long enough for the dwarven artisans to get a grip on it with their tongs. At heroic tier, I would have said "no". At mid-paragon tier, I happily said "yes" - and the Hard Endurance check was enough for the challenge to succeed, and Whelm to therefore be reforged as Overwhelm. (Had the player failed the check, I would have allowed the reforging to take place in any event, but was going to impose some sort of consequence for the PC on wielding Whelm, as the burns to his hands returned whenever he picked it up.)
I can't profess to knowing every book for 3E/PF, but I'm pretty sure it has no mechanic that I could use to run any of the encounters I've described or linked to above, nor many of the other skill challenges I've run. I know that there are no mechanics to handle this in classic D&D. And this is so in two respects. First, these earier editions have nothing analogous to the pacing dynamic of a skill challenge, which requires the GM to keep the scene alive (including by introducing new complications) and the players to respond to the evolving scene, which continually, at least in the skill challenges I run, produce unexpected outcomes - of which the dwarf sticking his hands into the forge to hold Whelm steady is just one simple example.
Second, they have nothing analogous to the "genre constraint" on permissible actions. So there is no simple mechanic for (for example) evaluating the success of a dwarven fighter-cleric's attempt to facilitate the reforging of an artefact by shoving his hands into the forge and holding it steady. A related consequence of this feature of skill challenges as a mechanic is that they make it very easy to adjudicate the improvisational use of combat powers for non-combat purposes. For, as the 4e PHB notes (p 259, emphasis added):
Chapter 5 describes the sorts of things you can attempt with your skills in a skill challenge. You can use a wide variety of skills, from Acrobatics and Athletics to Nature and Stealth. You might also use combat powers and ability checks.
In the reforging skill challenge, the player of the dwarf used Fighter's Grit (a utility power that permits the ignoring of certain adverse conditions) before shoving his hands into the forge (gaining a +2 to his Endurance check). In another recent challenge, the wizard PC used a power that lets the caster physically discorporate and possess an enemy to try and read an NPC's mind for a password (he succeeded in possessing the NPC, but failed in the Arcana check to extract the information he was after). In one of the examples I've linked to above, the sorcerer PC used Bedevilling Burst, a push attack, to upset some servants carrying jellies at a dinner party (together with a Bluff check to conceal his casting of the spell), thereby demonstrating effectively how one might physciallyd defeat a gelatinous cube (which was a topic of conversation at that time).
I don't think this sort of improvisation is so easily incorporated within a system that relies heavily on the mechanics themselves, rather than the logic of situation and genre, to determine what is feasible and what the consequences of actions are (and what makes this workable, in a skill challenge, is that the push towards resolution of one form or another is provided by the metagame imperative of the "N before 3" structure, rather than the mechanically-determined outcomes of discretely resolved tasks).
For D&Dnext to support this sort of non-combat resolution, at a minimum it would need to give me DC guidelines, some sort of resource system able to play the same functional role as the power system in 4e, and a general approach to scene framing and scene resolution which allows "genre logic" and metagame-driven complication introduction to work (so eg no need to track time and durations outside the context of the scene, which is one enemy of scene-based resolution).