Need a little help working out the math...

Well, as a GM, I would not set jumping a chasm as a set "you need to get a Superb result to succeed" because as you pointed out, there are varying degrees of success. I would instead say something like "Superb to get across unharmed and vertical; Great to get across and stumbling, now make a Dex check to keep from losing your footing and falling over backwards; Good to get most of the way across but now you're clinging like Indiana Jones, make Str & Dex checks to pull yourself up." Also, I probably wouldn't even tell the PC what he needs to roll. Just have him roll and let me interpret the result (based on the chart and circumstances). (I hope that didn't come across as challenging another GMs style, wasn't the intent)

On the surface, keeping the PCs skill level static and changing the difficulty seems to be the same as keeping the difficulty level static and changing the PCs effective skill. I think you're a lot better at working with odds/probabilities so could point out where and why I'm wrong.

Definitely something to think about.

Personally, I think that skill level is static. The knowledge &/or muscle memory is there, it's just a matter of tapping it. How successfully you tap it will be based on all sorts of other factors.

I have a 20' horizontal long jump. It's 20' whether there's a chasm below me or just a sand trap. The chasm might effect me psychologically, but that's still an external factor. There might be a wind, same thing. Neither of these change the fact that I can jump 20', they're just making it more difficult (higher "target number") for me to do so.