Russ, not to get too defensive, but is "you succeed the check, but you haven't gotten enough clues yet to spot the crook" any more arbitrary than "you hit him with your sword, but you haven't dealt enough damage to drop him"?
I think your system works from the DM's side, but it makes less sense to me from a player's perspective. The player isn't thinking, "I'm looking for Boris." He's thinking, "I need to scan this giant crowd for suspicious people."
So he says, I scan the crowd, and makes an Insight check. You say, "Okay, there's a guy glancing around nervously." . . . I dunno, it makes it sound like, hey, I succeeded the check and spotted the suspicious guy. I guess I can stop now. If you then tell him, "Well, you're not sure he's a threat," the player will still latch onto proving this guy is a threat.
It creates a different feel: 'find Waldo' vs. 'figure out whether there is a Waldo, figure out what Waldo would be wearing, look for possible Waldos, rule out a few people who just like red-and-white striped shirts, then finally find Waldo.'
I do like some of the stuff from yours, though. Like clarifying that Bluff/Diplomacy/Intimidate are great for interacting with the three dockers to get them to give up the fourth guy. And changing the mood of the crowd based on the round, to give the event more of a sense of rising action and tension.
I drew this from Stalker0's
Obsidian system. I trust his math, where he figured out that to give players a 75% chance of success (40% partial success, 35% full success), you set high DCs, rather than the ones WotC uses.
Instead of the core 4e style where you can use, for instance, Bluff to make your ally's Perception check easier, in this you make your own check to see whether Bluffing helps somehow. Or you just aid another.
Failed checks don't matter. It's "how far can you get before time's up" (as opposed to normal 4e skill challenges, which are "how far can you get without spilling any water from this glass three times").
I actually prefer a system where only a few skills are 'primary,' and the rest aren't as useful. I want skill choices to matter, and if you can solve any problem with any skill, it makes your choices less meaningful. If you can use Acrobatics, Athletics, and Stealth to win this challenge, why even bother having skills like Insight or Streetwise?
Now, if the players don't have good modifiers in the primary skills, they can still try to make checks, and they can get a few successes from secondary skills. But secondary skills alone won't provide enough successes to win the challenge. You
have to use some amount of Insight, Perception, or Streetwise to find the suspects.
Here's how I envision the skill challenge working in play.
[sblock]I describe the situation, you tell me how you act. Once I've got all that, you make skill checks to see how effective you are. A 'successful' skill check DOES NOT mean you spot a guy. It means you were effective at whatever you were trying to do.
"I'm looking for people acting suspiciously." Okay, Insight. Success? You spot a lot of people acting suspiciously. Like seriously, there are seven hundred people here. If you want to narrow it down you'll have to observe them longer, or look for other clues.
"I'm looking for guys with weapons." Okay, Perception. Success? You spot a few people who might have concealed weapons. That in and of itself isn't a crime, but the overlap between the last two checks narrows your list of suspects to maybe 50 people of interest.
"Can't we just grab them all?" Sure, you could, but if you move against the wrong guys, any real threats might be able to hide from you in the crowd.
"I want to look for anyone here who's a repeat criminal." Okay, Streetwise (maybe History if a PC spins it as 'recalling files I read back at the police precinct'). If you succeed, you realize there aren't any noteworthy criminals, but I'll throw you a bone and say that your preliminary research tells you that dockers are most likely to cause trouble at an event like this.
"Okay, I look for dockers." First, another Streetwise to know the signs and fashions. Success means you can give the rest of the party a list of things to look for. Then a different PC can make a Perception check to actually find people who look like dockers. This narrows the list of suspects to maybe 15 people.
"So can we grab them now?" Again, you can go for it, but if they spot you taking people away they might be able to avoid you.
"Fine. I go and discreetly talk to people we suspect, and see how they react to scrutiny." Insight again, or maybe Bluff if the PC is pretending not to be law enforcement. Success gets the party to 6 total, so the PC identifies one guy who's definitely here to cause trouble, and two other guys who you think are his accomplices.
If the party has time to make more checks, they might grab the first three guys, then Bluff/Diplomacy/Intimidate them into telling their plans. Or use Insight to figure out that the guys are hiding a fourth accomplice. Then use a skill to find the guy before he goes through the checkpoint.
But if the party runs out of time, they might only get to interrogate the three suspects after the crowd goes through the checkpoint. When they discover the fourth guy got past them, he's harder to deal with now that he's closer to the VIPs. The PCs can probably handle one guy, but any sort of scuffle could disrupt the event.[/sblock]
Does that make sense? I'm going to tweak it based on what people have suggested.
