To adopt you guys' Hide vs. Spot example ... If there is an important plot-clue that requires passing a Spot check, then the current skill system is borked, because there is simply too much of a chance that the Spot-skilled character will get unlucky and miss the clue, or that Belkar will get lucky and Spot the clue.
The RNG is too wide compared to the modifiers.
However, in combat (
or other situations resolved by rolling many times), since each skill check's outcome only lasts 6 seconds, this isn't such a problem. I don't mind if the Goblin has a 20% chance of being able to Hide from the Dragon when it only lasts one Round before he has to try Hiding again. That keeps combat interesting, and can be chalked up to the general chaos and unpredictability of combat as far as in-game sense goes. So in these situations, I'm not sure the RNG width vs. the characters' modifiers is actually a problem.