OK, so I've watched Solomon's videos and haven't seen a good clean laymen's explanation, so I figured I'd try. I'm skeptical it happened but the proposed algorithm itself is brilliant so that seems worth explaining.

The key thing it all hangs on: as an election unfolds, if a precinct has a vote count at a certain ratio at given time, you can push manufactured votes into it while sustaining that ratio for a bit without looking suspicious, and, here's the key, you can define some rules about when to do that to ensure you "land" extremely close to where you want: a precise re-distribution of votes to get a desired margin that doesn't over or undershoot by the time everything is over. This is the brilliance of it: its easy to come up with algorithms that land in the right place, but will be obvious to detect, and easy to come up with algorithms that are hard to detect but unlikely to land exactly where you need to be to make it just rigged enough to win but also not screw things up to have a huge win for the winner or inadvertently lose. This algorithm does both. It's exactly what I'd suspect from a NSA/CIA designed algorithm, but I don't want to sound like a conspiracy theorist especially since this isn't yet proven to have happened (and for many reasons seems hard to believe.)

The algorithm starts out with a few inputs: # of votes to hijack, margins to target, and a "wheel" number which guides the scheduler. The wheel stuff is focused on a lot in the videos, but its just describing how the algorithm decides what to do through the night. The biggest benefit of the wheel analysis is that if you can get a wheel number that matches the data, it is strong evidence this algorithm is being used (but is not "the smoking gun" cited elsewhere.) No wheel number means this algorithm was not used, or the data was too noisy to prove it.

The "virtual precinct" concept I think is the wrong way to describe it. A better way to think about it is that the algorithm identifies opportunities to push votes through at fixed ratios. As the night goes on, and the ratios change or votes stop coming through at a precinct, it switches from one precinct to another to push its budgeted votes through at the existing ratios that precinct is at, sometimes guiding it up or down to the ratio it needs to get close enough to the desired result. The wheel model guides it, but the "wheel" isn't a real thing in the code per se, it just is a way to describe how it works: it needs to be continually 'balancing' what ratios it takes advantage of and for how long to ensure it lands where it wants. The wheel model describes how it makes its choices. The virtual precincts aren't really a thing, its just a way to describe the periods where it is pushing votes through at a fixed ratio temporarily before it 'releases' the precincts. The trick is just that it knows if it does this over and over again in the right distribution of ratios (which it picks dynamically, via the 'wheel' model) it gets what it wants.

The "smoking gun" in the video is that, at the end of all this, if you can identify which votes were pushed through by this algorithm during these periods of it sustaining ratios by finding places where ratios were fixed and then "handed off" through time, you can show that the algorithm was performing this balancing (and it wasn't a normal election) if it turns out if you hypothetically flow a fixed number of votes through each ratio it took advantage of instead (summed over all the votes it flowed through that ratio, across all precincts), you get the same overall margin, that's strong evidence that an optimizing algorithm like this was flowing votes to hit that target.

In the wheel model, this just boils down to, instead of pushing a random "balanced" flow of votes through the spokes (as the algorithm did on the fly to elude detection and take advantage of the real vote flow ratios that manifested), you test if the same wheel model works out if you push the same number through each spoke. If both the real flow and this artificial flow get you the same outcome, that is super strong evidence the real flow was guided by this model.