I've been following the advice of Matt Colville from YouTube, and introduced a Skill Challenge to model the party's escape from an exploding dungeon. Using some of the guidelines from his video (and the 4e rules), I came up with the following skill challenge for five 6th level characters....
10 successes before 3 failures. DC 15 checks. Accepted skills included: Perception, Athletics, Acrobatics, Survival, Nature, (or others if you could make a convincing argument). Failure was 9 points of damage unless someone could negate the failed check with a successful check.
So the group didn't come close to the 10 successes. After setting up the conditions of the skill challenge, it became impossible to "walk it back" even as things were clearly going bad. However, after letting the dice fall where they may, I tried to be accommodating for letting the players bring in new characters or to bring back their previous characters with as little setbacks as possible.
I guess my question is ... do any of you think skill challenges are worth having? Do you have any rules of thumb when designing them?
i recall that video and came away thinking "nope" myself but hey...
At the start in my 5e game i setup that we would resolve any extended task with a three-way or trios save race to three just like how death saves are handled except that actual skills and such are involved.
Since i also establish that the definition of failure in the PHB for ability checks will be used - and so i can as Gm decide any failed skill roll is still "some progress with setbck" that leaves a lot of room for how things pan out in the race to three. I myself have not seen any real gain in going mix-n-match on the number of successes thing, given the difficulty, types of skills, time for each check and so forth are all variable... i mean at some point you do need some part of your resolution system to be a foundation the rest is built-on right?
So, i lock down that we will resolve with the race to three and then let the other variables be the way the specifics of the event play into the scenes.
FYI - back in the days of VtM 1st ed, they used task checks with all three of the following varying by task:
Number of dice rolled.
Threshold on each die needed for success
Number of wins needed for "success".
that made for a very frankly busted system where at times the very broken math got in the way.
You really need one element of your resolution process to be locked down as a foundation IMO for the system to be playable with reasonable play, IMO. otherwise you get surprised more than you would like, in my experience.
In play in 5e this has worked out great. Each paretial success gives them a lead, often failures are "you get farther but now that is a tapped resource and you have to look elsewhere or try something else - different skill" so the resreach got you only so far at the church but it did lead you to the options of a hermit or the larger library in another town or maybe some ancient caves nearby to try for more answers.