That's a decent understanding. I can second other poster's comments that while he understands this stuff very well himself, Solomon isn't a natural teacher (but he's trying).
The key seems to be that they would set some target percentage of votes to switch (say they predicted Trump would win the state 52% to 48% so they would target to switch 3% to get Biden winning at 51% to 49%). So to minimize the chance of being detected they would instead target the ballot stuffing to a smaller number of precincts and move those to a more extreme degree (in theory because they have physical control of the ballot boxes at a few of these precincts so it will be easier to insert the fake ballots to match up with the digital totals being manipulated by the program). So if they want to swing the whole state 3% it would be easier to seize a small number of the data uploads and swing them like 80/20 in Biden's favor.
Rather than just sending a bunch of uploads at exactly the ratio of votes they want (80/20) they are instead using a wheel to average many different uploads of many different sizes to that target ratio of 80% Biden to 20% Trump votes. The bit you mention about ratios "transferring" is what happens when the program decides it needs a certain amount of votes to go in at a specific ratio. So say for example its in the middle of balancing out some uploads and decides it wants 1,000 votes reported with a ratio of 90% Biden to 10% Trump. To avoid detection, let us say it spreads those 1,000 votes across 3 precincts. It might upload 350 votes at one precinct, 400 at another and 250 at another. Now there's a lot of ways of looking at data, but what Solomon discovered was that he could find ratios transferring across precincts when he sorted them with respect to time. So in this simplified example, he would find that 3 separate precincts uploaded votes of different amounts (but crucially, of with the same ratio of Trump-to-Biden) either at the same time or back-to-back.
Why this is proof of fraud, is that when a vote total is reported it has thousands of possibilities of what ratio the votes might come in. So say you have that first upload of 350 votes. It has, let's say, a 1-in-7,000 chance that it will happen to have an exact ratio of 90/10 Biden/Trump votes. Then you have this other upload that is reported immediately after or 400 votes. That one has a 1-in-8000 chance of happening to have an exactly 90/10 Biden/Trump split.... but then you have a third! And they all happen within minutes. The chances of that happening naturally even one time would be like winning the lottery. But it happened hundreds of times in the Georgia election he examined. Together, the chances of hundreds of 1-in-a-million chance events occurring is essentially impossible.
Essentially what discovering the Wheel Number is, is exposing the underlying mathematical rules for how the splitting and assigning of these fake vote uploads happened. The algorithm had rules designed to decide when to upload batches of votes and what ratios to assign them at, and that was determined by the Wheel Number. Solomon's method of reverse engineering the output to find that wheel number is just astounding.
I applaud his efforts at trying to explain how this all works to the public but it's not an easy task. This whole things feels a lot like the tangle of mortgage-backed securities that caused the 2008 recession: that too was a very complicated problem that most people couldn't understand and most of the people that did understand it weren't able to make other people understand it as well and it caused a bit of a crisis because it was hard to get the public to push for the changes needed to prevent that from happening again (because 99% of them couldn't even understand what had happened or why).
Please tell me if I'm doing any better of a job explaining this. It's hard to breakdown into something digestible without specific nuance being lost and I think that's a large part of why Solomon's struggling to make people understand what happened--he can't help but be mathematically precise, even when doing so makes his explanations inaccessible.
The wheel is entirely unnecessary but the wheel essentially explains how he was able to figure it out. I believe he is trying to explain his thought process or methodology in hopes that others can have a similar "a-ha" moment.
Something that is missing from your simplified explanation, is that the fraud was essentially pre-meditated. Using the algorithm, they can pretty much figure out how many ballots they need ahead of time and figure out which precincts they need to control. This allows them to only cheat by as much as they need and allows them to cheat by only controlling a few key areas.
Where things really start to fall apart is when there's larger than anticipated voter turnout. There are only so many votes in any given precinct/county that you can "balance" before you end up with things like 90%, 100%, 150% voter turnout.
I believe this is what Sidney Powell is talking about when she says there were so many votes that the algorithm broke. The algorithm did not break, it was used in advance. What happened was the Democrats produced hundreds of thousands of ballots BEFORE the counting even started.
We've mostly caught all this activity in areas where voter turnout was much higher than anticipated (possibly even from both sides) because this will manifest itself in one of two ways:
a) unusually high or impossible voter turnout (particularly in lower population areas)
b) Democrats needing to produce MORE ballots on election day to catch up (in higher population areas). These are the B grade/reject/back-of-the-truck ballots. They were only used as a last resort because the head-start that they got from early ballots was not enough. The democrats didn't want to use those under the table ballots, but they HAD to, even if it meant you're caught on CCTV doing it. They had no choice, they were going to lose even after stuffing the ballots ahead of time.
Important take away is that this algorithm was used AHEAD OF TIME. This allows the democrats to mass harvest/produce absentee/mail-in ballots using a golden list of fake voters that the democrats have been working on FOR YEARS, then figure out which counties they need to slip these ballots into without people asking too many questions and let the "extremely close but ultimately Democrats pulled ahead by 2%" race begin.
Yes, this is essentially a feed-forward algorithm, based on initial inputs which are derived from polling data beforehand. It would be too obvious if it were based entirely on feedback - information coming in from the polling centers on voting day. They had a fair idea of how many manufactured ballots they'd need. There initial polling was off, however, and they did have to make corrections based on voting day feedback, and our eyes were drawn to these areas first.
That's a decent understanding. I can second other poster's comments that while he understands this stuff very well himself, Solomon isn't a natural teacher (but he's trying).
The key seems to be that they would set some target percentage of votes to switch (say they predicted Trump would win the state 52% to 48% so they would target to switch 3% to get Biden winning at 51% to 49%). So to minimize the chance of being detected they would instead target the ballot stuffing to a smaller number of precincts and move those to a more extreme degree (in theory because they have physical control of the ballot boxes at a few of these precincts so it will be easier to insert the fake ballots to match up with the digital totals being manipulated by the program). So if they want to swing the whole state 3% it would be easier to seize a small number of the data uploads and swing them like 80/20 in Biden's favor.
Rather than just sending a bunch of uploads at exactly the ratio of votes they want (80/20) they are instead using a wheel to average many different uploads of many different sizes to that target ratio of 80% Biden to 20% Trump votes. The bit you mention about ratios "transferring" is what happens when the program decides it needs a certain amount of votes to go in at a specific ratio. So say for example its in the middle of balancing out some uploads and decides it wants 1,000 votes reported with a ratio of 90% Biden to 10% Trump. To avoid detection, let us say it spreads those 1,000 votes across 3 precincts. It might upload 350 votes at one precinct, 400 at another and 250 at another. Now there's a lot of ways of looking at data, but what Solomon discovered was that he could find ratios transferring across precincts when he sorted them with respect to time. So in this simplified example, he would find that 3 separate precincts uploaded votes of different amounts (but crucially, of with the same ratio of Trump-to-Biden) either at the same time or back-to-back.
Why this is proof of fraud, is that when a vote total is reported it has thousands of possibilities of what ratio the votes might come in. So say you have that first upload of 350 votes. It has, let's say, a 1-in-7,000 chance that it will happen to have an exact ratio of 90/10 Biden/Trump votes. Then you have this other upload that is reported immediately after or 400 votes. That one has a 1-in-8000 chance of happening to have an exactly 90/10 Biden/Trump split.... but then you have a third! And they all happen within minutes. The chances of that happening naturally even one time would be like winning the lottery. But it happened hundreds of times in the Georgia election he examined. Together, the chances of hundreds of 1-in-a-million chance events occurring is essentially impossible.
Essentially what discovering the Wheel Number is, is exposing the underlying mathematical rules for how the splitting and assigning of these fake vote uploads happened. The algorithm had rules designed to decide when to upload batches of votes and what ratios to assign them at, and that was determined by the Wheel Number. Solomon's method of reverse engineering the output to find that wheel number is just astounding.
I applaud his efforts at trying to explain how this all works to the public but it's not an easy task. This whole things feels a lot like the tangle of mortgage-backed securities that caused the 2008 recession: that too was a very complicated problem that most people couldn't understand and most of the people that did understand it weren't able to make other people understand it as well and it caused a bit of a crisis because it was hard to get the public to push for the changes needed to prevent that from happening again (because 99% of them couldn't even understand what had happened or why).
Please tell me if I'm doing any better of a job explaining this. It's hard to breakdown into something digestible without specific nuance being lost and I think that's a large part of why Solomon's struggling to make people understand what happened--he can't help but be mathematically precise, even when doing so makes his explanations inaccessible.
The wheel is entirely unnecessary but the wheel essentially explains how he was able to figure it out. I believe he is trying to explain his thought process or methodology in hopes that others can have a similar "a-ha" moment.
Something that is missing from your simplified explanation, is that the fraud was essentially pre-meditated. Using the algorithm, they can pretty much figure out how many ballots they need ahead of time and figure out which precincts they need to control. This allows them to only cheat by as much as they need and allows them to cheat by only controlling a few key areas.
Where things really start to fall apart is when there's larger than anticipated voter turnout. There are only so many votes in any given precinct/county that you can "balance" before you end up with things like 90%, 100%, 150% voter turnout.
I believe this is what Sidney Powell is talking about when she says there were so many votes that the algorithm broke. The algorithm did not break, it was used in advance. What happened was the Democrats produced hundreds of thousands of ballots BEFORE the counting even started.
We've mostly caught all this activity in areas where voter turnout was much higher than anticipated (possibly even from both sides) because this will manifest itself in one of two ways:
a) unusually high or impossible voter turnout (particularly in lower population areas)
b) Democrats needing to produce MORE ballots on election day to catch up (in higher population areas). These are the B grade/reject/back-of-the-truck ballots. They were only used as a last resort because the head-start that they got from early ballots was not enough. The democrats didn't want to use those under the table ballots, but they HAD to, even if it meant you're caught on CCTV doing it. They had no choice, they were going to lose even after stuffing the ballots ahead of time.
Important take away is that this algorithm was used AHEAD OF TIME. This allows the democrats to mass harvest/produce absentee/mail-in ballots using a golden list of fake voters that the democrats have been working on FOR YEARS, then figure out which counties they need to slip these ballots into without people asking too many questions and let the "extremely close but ultimately Democrats pulled ahead by 2%" race begin.
Yes, this is essentially a feed-forward algorithm, based on initial inputs which are derived from polling data beforehand. It would be too obvious if it were based entirely on feedback - information coming in from the polling centers on voting day. They had a fair idea of how many manufactured ballots they'd need. There initial polling was off, however, and they did have to make corrections based on voting day feedback, and our eyes were drawn to these areas first.