And the PC is thinking "If I ride well enough, fast enough, I can escape! Oh wait I failed to go the right way! If only I knew the terrain of my home town better / was more perceptive / <better at whatever other excuse can be provided as to how a gorge is a reasonable failure for a riding check> I wouldn't be at risk to get caught! If I live, I better improve my ability to spot gorges!!!"
So, where's the character feedback? The agency of the PC is continuous n
ot scene framed.
Where was the PC agency again? How was the PC allowed to react and adapt to the world around him? The PC reacted to the failure to the best of its ability and that reaction was wrong because the game feedback was wrong -- the gorge wouln't be avoided because "spotting gorge" skills were improved; it'd be avoided because Riding improved.
We still seem to be missing each other. Take my anecdote above. Extend it to any martial exercise (and one non-martial exercise):
- How is it that an excellent pitcher with his best stuff (James Shields has done this untold times this year) and command suddenly and inexplicably lose it after cruising for 5 innings?
- How is it that skateboarders, divers, floor exercisers cannot stick the landing every single time on a trick/dive/routine that they've practiced an enormous number of times?
- How is it that a professional basketball player (primarily a jump shooter) cannot consistently put away jump shots at the same general percentage clip in a game-in/game-out basis?
- How is it that a professional soccer player can completely miss the net on a penalty kick?
- How is that Rafa Nadal (2008, 2010 winner and 2011 runner up) can manage to not even advance past the second round in this year's Wimbledon?
-How can a world renowned geophysicist and a father of paleogeography be on the wrong side of the "Continental Drift" hypothesis when facing off against a meteorologist/climatologist by training?
Funny things happen in singular moments. No one reproduces their acumen with perfect delivery and sometimes other oddities interfere to change the course of the moment. However, after a significant body of work is established (number of data points are collated) and then regressed to the mean, their level of proficiency will be clear.
That is merely an explanation on your own terms. I can think of any number of sensible outcomes if required but I don't even agree with your premise, to be honest. I've played a considerable portion of my gaming life whereby every single check statically produced the very next "physical micro-moment" binary outcome. With respect to the creation of dynamic, interesting fiction (the design aim here...not process simulation) I have found this unbelievably unfulfilling (as have my players) over the course of time. Having the next fictional happenstance not be an "(i) immediate, (ii) PC perspective (iii) micro-evolution" from the last check allows for a considerably wider scope of fiction and a much more interesting game for us (you can actually reproduce Indiana Jones Chases and Star Wars Cantina moments with some measure of reliability). If you allow it to (i) be a few moments removed from the last (if you must), (ii) from the perspective of "is this compelling fiction", (iii) macro-abstraction, you will have much greater control on the the aimed-for level of dynamism and a greater chance at satisfactory trope reproduction....both from "check to check" and through the aggregation/outcome of the Skill Challenge. That is our design aim at our table with skill challenges. You can do that and still "simulate" the entirety rest of the gameplay (combat, rudimentary skill checks) till your heart's content if that is your wish.