He walked straight into this one, huh? His logic is simple: Even if the White House’s voter-fraud commission can cook up an estimate of how many votes were fraudulently cast last fall, there’s no way to tell which of those votes were cast for Clinton or Trump. As such, there’ll be no way to know for sure that Clinton really did win the popular vote. But, er…

Clinton popular vote margin:
2,868,692 votes (2.1%)
Trump margin in:
PA—44,292 votes (0.7%)
WI—22,748 votes (0.8%)
MI—10,704 votes (0.2%)

— Nate Silver (@NateSilver538) July 19, 2017

If you want to call Hillary’s popular-vote victory into question, you need to cast doubt on at least 2.9 million ballots — actually considerably more, since Kobach is right that some fraudulent votes inevitably will have been cast for Trump. Even if it could be proved that fraudulent voters broke three to one for Clinton, you’d need upwards of six million ballots to have been cast illegally to produce a margin of 2.9 million. If, however, all you want to do is tilt the electoral college so that Hillary wins, you’re dealing with a much smaller pool since you only need to flip Michigan, Wisconsin, and Pennsylvania to make it happen. Assume Trump got 60 percent of illegal ballots in those three states versus 40 percent for Hillary; you’d need 375,000 total fraudulent ballots to have been cast in order to produce his combined three-state margin of 75,000 or so votes.

And all of that assumes fraud at the individual level, committed independently by voters, rather than orchestrated en masse by a campaign. If you’re all-in for the most feverish conspiracy theories involving ballot-box stuffing, the difference is still obviously stark. Assuming a more or less 50/50 split in legitimate votes cast, Clinton would need to cook up nearly three million fake ballots. Trump would need far less than 100,000.

Kobach’s stuck here because, as head of a supposedly neutral voter-fraud investigation, he can’t say what he almost certainly wants to say and which nearly all right-wing believers in widespread voter fraud suspect: It’s illegal immigrants who are casting these ballots and illegals are overwhelmingly likely to vote Democratic, especially when the alternative is a border hawk like Trump. Even if it were true, though, that massive numbers of illegals were voting, you’d need an enormous share of the overall illegal population to turn out and break heavily for Clinton to produce a margin of 2.9 million. Assuming, per estimates, that the total number of illegals in the U.S. is around 11.3 million, more than one in four illegals nationwide would need to have turned out and voted unanimously for Clinton to produce her overall PV margin — even though the penalty for getting caught trying to cast a meaningless vote would be deportation at a minimum and prison time potentially. If you tweak the equation so that Clinton wins “only” 80 percent of fraudulent votes cast by illegals in this hypothetical, it would take more than 4.8 million illegal voters — more than 42 percent of the country’s total illegal population — to have been at the polls to produce Hillary’s national popular vote margin. Illegals do vote sometimes in elections, of course, and can even be decisive in verrrrry close elections (like, er, Michigan?), but even the researchers who published that finding don’t think they’re voting anywhere near the volume it would take to produce 2.9 million votes.