re: ESPN: Chance to Make the College Football Playoff

I presume this is likely generated from X number of simulations (say, 10,000) where the computer saw an avenue for Alabama to be one of the top 4 teams (considering that, at the end of the day, that's pretty much the only real criteria you could give a computer based on the previous selections by the committee).

So in 97% of those simulations Georgia finishes in the top 4. The only outcomes where they don't are likely losing to both Georgia Tech and the SECCG, though there may be some quirk where they lose just one and someone leapfrogged them into the top 4 spot.

It's hard to see 15% of the scenarios where Alabama moves past enough teams to get to the top 4, though I imagine it has more to do with the following:

Alabama likely wins it's final game (Auburn)

Then:
Clemson falters against South Carolina or North Carolina (who would likely have had a somewhat decent shot before losing to Georgia Tech)

TCU loses to Iowa State (as #4, they can drop the easiest)

The Ohio State/Michigan loser adds another layer of potential, and the scenarios clearly favor OSU (but still think that the loser gets in).

USC loses to Notre Dame or in the Pac-12 CG (as the opponent in the Pac-12 CG isn't know for certain, it's probably a weighted average of Washington and Oregon and maybe Utah - would have to look up the tie-breaker scenarios for the PAC12 as Utah lost to Oregon but did not play Washington).

If you remove the human element and only consider "What path leads to moving above other teams into a top 4" then it has possibilities; it's just when you think about the fact that the committee will weigh other factors in that the computer does not that it becomes more apparent that Alabama's 15%$ "chance" is probably closer to 1-2% with those 1-2% of scenarios being just a massive dose of late-season chaos.

Is it "impossible"? No.
Is it "so unlikely that you have better odds winning the Pick 5"? Yeah, pretty much.