Wow, that is crazy.......
It is, but if you break it down - not so much!
Lets assume that your typical 30 minute show has 3 commercial breaks, around 3 minutes each (roughly). You need the start and end time for each break. Luckily, they don't usually mess with the commerical timings across airtimes for most content.
So, even on 30 minute blocks, from 8-11PM, 6 30 minute blocks by 3 breaks, thats 18 "commercial start" and 18 "commerical end" times. You could pay 4 people min. wage to watch a show, and press a button when a commercial starts, and press a button when a commercial stops. Much cheaper than dealing with automation, and contending with networks.
Edit: Your automation can come in upon those "button presses" that will record the time, and after cross checking (against other users?) assumedly transfer it to the hopper at the alloted time. The Hopper would then que up the list of "when to skip", and just auto-skip if the show and airing matches against it's updated database.
For all we know, Dish could be crowd-sourcing the data, say the Hopper reports back after each PTAT recorded show playback. Users do the work by auto-skipping the commercials, the largest skipped time areas get overviewed, and pushed back to other users > 1AM.