Steve Cortes, claims that if you subtract the number of votes cast for senator from the number of votes cast for president, then this gives you votes that were ONLY cast for that presidential candidate. He then claims that in the Georgia race, this number is suspiciously high for Biden, and this is "far too wide to be believable".
He does the math with Georgia, and comes up with 818 votes for Trump (claiming these are people who voted for Trump only), and +95,801 votes for Biden (claiming these are people who voted for Biden only.)
The simple math checks out, although now with updated numbers, it's -789 for Trump and +99,868 for Biden
But wait, LESS people voted for Trump than for Perdue(R)? And Jorgensen(L) is at -52,794 compared to Hazel. So what's going on? To understand this, it's useful to also look at the 2012 and 2016 elections for context. Firstly here from the official GA state website:
And those same numbers converted into a spreadsheet, with the difference between the presidential votes and the votes for the senator of the same party calculated, at the bottom.
People don't always vote for Senators of the same party as who they voted for Presidents. Libertarians probably felt the Presidential election was more important than Senator, so those 52,794 voters that voted for Hazel but not Jorgensen largely would have voted for Biden or Trump.
And since Trump didn't get any excess votes, you might think they mostly went for Biden.
But what about the other 48,000? For this to be suspicious means you'd expect the number to be really low, or negative, like Trump's number. But in previous years the number has been high for democratic candidates. It was 235,905 for Obama in 2012, and 278,237 for Clinton in 2016, so really 99,868 is hardly surprisingly high. If anything it's surprisingly low.
Perhaps what is surprising here is that Trump got fewer votes than the Republican Senatorial candidate, after all, Mitt Romney had an excess of 220,025 in 2012. The -789 number is actually an improvement over 2016, when Trump received 46,702 votes fewer than Republican Senator Johnny Isakson.
The reality is there are several factors that go into this difference, and it varies wildly from year to year. A big factor is the voters who split the ticket. I.e. they vote one way for President and another for Senate (or House). A nationwide survey of 2020 candidates had 11% doing some kind of split: