This guy thinks he cracked the algorithm used in Georgia (something similar in Pennsylvania, too, and probably other states too, but he hasn't looked through their data):
Smoking Gun, Part 2: Ratio Transfers Proved; Entire Algorithm Reversed, net 200,353 votes for Biden in seized precincts in Atlanta. Affidavits signed. (rumble, bitchute, twitter, bitchute) Watch the video. You will know. Summary: Final hijacked...
www.sott.net
I can't follow all his steps (the math is above my head), but I can follow the gist of it. If he's right, something like the following took place: A certain number or percentage of votes were reserved for 'hijacking', i.e. making sure Trump didn't get above a certain percentage of that number of votes. To cover their tracks, they didn't simply give Trump a certain ratio in any given precinct. It would be too obvious if a single precinct showed a ratio of 1:20 for all its updates, for example. Similarly, you can't just give a collection of precincts the ratio you want. It would be too obvious if 300 precincts all show the same ratio.
So what they did is to hijack a number of precincts (around 1 out of every 3, with precincts in Fulton county greatly overrepresented). A ratio would be applied to one precinct for one or more updates, then that ratio would switch to another precinct. This happened with multiple ratios in multiple precincts until the target number of votes was reached, then the precincts were allowed to continue on using their natural vote totals. If they really wanted to cover their tracks, this would require making sure election works injected enough fake ballots, and destroyed enough Trump ones, in order to match the numbers produced by the algorithm. He doesn't explicitly say so, but it's possible that there was additional fraud on the paper-ballot level. That is, the algorithm just looks at the numbers coming in and determines what to switch - it doesn't take into account the validity of those original numbers. So Trump could conceivably have won by an even greater margin on top of what the algorithm achieved.
He argues that he has mathematically PROVED that an algorithm was used. There's no way what he found can be an accident. This is where the math comes into play. The gist: Many ratios were used, but in such a way that they eventually balanced or averaged out to their target number (something around 15% for Trump). So one precinct might give Trump 25% of 1000 votes in a single update, but that would mean another precinct (or combination of precincts) would have to give him 5% of 1000 votes, to achieve the target of 15%.
Because so many ratios were used, over multiple precincts, it is almost impossible to notice without digging into the data, which this guy did. He couldn't have found it without noticing that certain ratios 'transferred' to another precinct in the next timestamp update. The reason he says it is proof (and not just theory) goes something like this: If you take all the hijacked precincts in their final state (when the ratios stop being implemented), they average to the target percentage. Here's a simplified example according my understanding. Say there were only 5 hijacked precincts. Here they are, with hypothetical vote totals, Trump totals, and ratios for Trump:
P1 - 3451 total votes - 300 Trump votes - 8.7%
P2 - 647 total votes - 325 Trump votes - 50.2%
P3 - 2416 total votes - 220 Trump votes - 9.1%
P4 - 1235 total votes - 200 Trump votes - 16.2%
P5 - 954 total votes - 260 Trump votes - 27.3%
8703 total votes - Trump 1305 votes - overall ratio 15%
Notice how none of the individual ratios is 15% on its own, but overall that's what they come to. Trump even wins P2 by 50.2%.
The 'proof' is that the set of numbers found in the actual data is what I would call impossibly fine-tuned. Each number depends on every other number - you can't change a single one. It would be impossible to randomly get such a set of numbers. He demonstrates this by showing that when you replace all the individual precincts' total numbers with any random number, e.g. 1000, the overall ratio stays the same. This ONLY works if an algorithm is applied to ALL of these precinct totals. It doesn't work, for example, with the numbers I gave above, since I just picked them randomly:
P1 - 1000 total votes - 87 Trump votes - 8.7%
P2 - 1000 total votes - 502 Trump votes - 50.2%
P3 - 1000 total votes - 91 Trump votes - 9.1%
P4 - 1000 total votes - 162 Trump votes - 16.2%
P5 - 1000 total votes - 273 Trump votes - 27.3%
5000 total votes - Trump 1115 votes - overall ratio
22.3%
In order to get a set of percentages for the 5 precincts that will equal 15% overall (and not 22.3%), I would apparently have to use complex formulas using linear algebra to calculate
each number, which is what the algorithm did and why the actual results came out the way they did. (See the 2:00:00 and 2:20:00 marks of the video in the article for his demonstration). And it's not just 5 numbers, as in my simple example. It is
in the hundreds. (371 precincts, to be exact.)
Again, that's just my understanding. I can't follow or verify the actual math. But it was fun to try! Any mathematicians might want to take a look and see if the various steps make sense.