Subconsciously, I have to know that it's being adjusted to my performance or rather the expectation of my performance. It's so patronizing. When I 'lose', I have to think, "Am I being herded into another clash?" When I win, I wind up feeling, "What was the point in that?" If the gamemaster adjusts incorrectly, something usually 'magically appears' allowing chara progress.
All in all, that's too conflict, win-or-lose, for me. It's like the old chasm problem. There you are, being pursued. You reach the chasm that looks wide, but maybe not too wide. You make your jump roll.... Do you succeed or do you fail? Failure stops the game, breaks the 'chase' suspense and brings everything back to the table and mechanically determining the results. Success either ends the chase arbitrarily or it becomes less than passing scenery.
That's why I prefer complications. I always know (roughly speaking) where a typical game is going to go. In one fashion or another, the gamemaster will demonstrate something cool he has in store, be it item, situation, location, et cetera. The patronizing feeling is inescapable; I'm always aware that we are just wasting time before the 'show'. If I wanted that, I'd play CRPGs to get to see the cut scenes.
Complications on the other had, may inform the upcoming. They can set the stage, prime the pump and cock the crossbow. Each one can be tailored to create a need, foreshadow an event or carry a theme. Ultimately, we still get to the 'cool' stuff, but the complications don't wind up feeling so arbitrary; they seem more like an enhancement. The best part is when thought of as complications, rather than adversity, they can obviously play out on more and different levels, rather than just violence. Typical gamemasters might have a hard time seeing a romance adverse to the direction of a game, but everyone knows what a romantic complication is.
So here's the spotlight-driven idea: how about complications that don't offer failure? I know that sounds weird, but I've had it work quite well. Let's say a chara wants into a locked building. He attempts to pick the lock; does it end like the chasm? No, a bad roll calls for more complication, not failure. So he doesn't pick the lock, but does notice an open second-story window. You still get to do what you want, it just takes longer.
...Hence the advantage of spotlight time here. If you 'fail', you aren't stymied like at the chasm, you wind up getting more spotlight time. It becomes a tradeoff. Either you succeed or you get more spotlight time. Who complains? There is no failure.