X successes before 3 failures incentivises letting a single character do all the work. You can force the players' hands, but the game mechanism is fighting against you rather than helping you.
It only creates that incentive if the GM frames the fiction in such a way that
nothing bad is going to happen if no one else acts. Given that D&D is a party-based game, we don't generally design and adjudicate combat encounters in that way, so why would we do so for non-combat encounters?
The mechanism X successes before 3 failures doesn't look anything like combat. Attacking and missing doesn't bring you closer to losing the combat than just being stunned for the round unable to do anything. It is the opponents' successes in hitting you that bring you closer to losing the combat.
But those successes are a function of attacking and missing, because if you attack and
hit then you take the opponent out of the fight.
Sending the fighter into combat alone without any help, will almost certainly end with a dead fighter.
Not if the fighter is allowed to take on the orcs one-at-a-time, in a series of gentlemanly duels.
But it is taken for granted, in the design and adjudication of combat encounters, that they won't unfold like this. If the same principle is applied in the design and adjudication of non-combat encounters, then even players not optimised for whatever is going on will declare actions for their PCs.
Imagine how D&D combat would look if it was just a series of static, sequential rolls to hit (in the fiction, that is my line of duelling orcs). Of course groups would send the fighter in to make those rolls. But no one frames combats like that. (Sometimes players manufacture such a situation, eg by having the fighter hold a chokepoint. But that is by dint of their play; it's not generally just handed to them on a platter by the GM.)
If non-combat challenges are framed as a series of static, sequential rolls against a predetermined fictional element of a predetermined difficulty (like a monster's AC, or a lock's DC) then players will send in the bard, or thief, or whomever to make all the rolls. In general, I think that is not good encounter design out of combat any more than it is good design in combat.
Quickleaf gave the example: "A trapped treasure vault rigged to blow if wrong answer is given. In this scenario an X successes before Y failures is appropriate." I agree and as a player in that scenario I would certainly send forth the most knowledgeable character while the rest of us hid around a corner. Preferably with the half-orc barbarian gagged and bound.
That doesn't seem to me to be a very interesting skill challenge - it is really just a complex skill check, and it is the sort of thing that, in 4e, works best integrated into a broader situation which engages the other PCs (examples I've used include shutting down demonic gates, or unlocking doors to safety, while the rest of the party hold off onrushing hordes).
A skill challenge that is meant to engage the whole party needs to exert pressure, in the fiction, upon each PC, just as is the case in the typical combat. In combat, players are expected to use their resources to change the fictional situation so that their PC can work effectively rather than be ineffective - for instance, the player of a melee fighter is expected to declare actions that bring his/her PC into melee contact with the monsters/NPCs (and similarly the player of an archer or wizard who finds his/her PC in melee is expected to declare actions that restore to his/her PC the advantages of range). Or players (and their PCs) can help one another - eg the wizard teleports the fighter into melee, or the fighter holds off an enemy while the wizard or ranger falls back.
Similarly in a skill challenge: if the players have PCs who are strong at A but not B, and the GM is confronting them with a B-situation, then the players should be bringing their resources to bear to turn it into an A-situation, and hence one in which they can do well. Players who don't try and change the fiction, but just hurl themselves futilely at the GM's initial framing hoping that they get N successes before 3 failures are like players who leave their wizards in melee and their melee fighters throwing stones rather ineffectually from range.
Because there is no action economy in a skill challenge comparable to D&D combat, the onus falls on the GM to be fair and reasonable in the way that s/he applies pressure. For instance, if the NPCs strike up conversation with the low-CHA fighter, and the player rollls and fails some appropriate social check, and then another player declares an action that succeeds and that takes the low-CHA fighter out of the social situation, the GM shouldn't just turn the pressure back onto the player of the fighter (assuming it's not a two-player game) - there are other players whose PCs are there to be engaged. I don't adopt a strictly round robin policy - I try to be even-handed in my applications of pressure, but I also follow the lead of the fiction, which includes letting the players enjoy the benefits of their successful checks. For instance, if the fighter is successfully extracted from the social situation, then it would be unreasonable to apply further
social pressure to that PC unless the fighter reenters the social situation. (Which s/he might do for any number of reasons, the most obvious of which would be to avoid looking anti-social.)
In short, in an interesting skill challenge the GM will be applying the pressure to all the players (via their PCs), and the players will be using their action declarations and their PC abilities to make their own luck, just as happens in the typical D&D combat.
(and I STILL flirted with death more than once!)
