Here's the problem with electronic voting systems: no matter how much funding they receive and no matter how brilliant the programmers and cryptographers are who design them, they will always have a large attack surface and they will always provide worse verifiability guarantees than paper. Even in a perfect world.
Paper ballots, and especially vote-by-mail paper ballots, have an extremely small attack surface. To commit fraud, an attacker must create a large number of bogus paper ballots and get them into ballot boxes, or they must alter the votes on a large number of existing paper ballots, or they must somehow engage every single person responsible for counting and verifying votes in a conspiracy.
Because of how tightly paper ballots can be controlled at in-person voting places, and because of how widely distributed vote-by-mail ballots are, it's virtually impossible to alter votes on existing ballots. It's also very hard to introduce fake ballots, because faking a ballot typically requires getting a fake voter on the roll. You may be able to do that for a handful of ballots, but a handful of votes isn't going to win an election.
If there are questions about the legitimacy of a vote, paper ballots can be recounted and verified manually, because they're physical objects that actually exist.
But with e-voting, if you can gain access to any of the multiple electronic systems involved anywhere in the chain, you may be able to alter votes en masse without needing to create fake voters. And there are any number of ways you could do it.
If you can inject code into the voting software, you may be able to alter the UI so that when the voter taps the name of Candidate A the system actually logs a vote for Candidate B. Or you may be able to execute a database query that flips a random sampling of previously cast votes for Candidate A to votes for Candidate B. The possibilities are endless.
And when electronic voting systems are tampered with, recounting and verifying votes can be difficult or impossible, because there may not be any record of votes except the electronic one that you can't trust.
Software security is incredibly hard to get right, especially at the level required for voting systems. Even when done right, clever attackers can still find vulnerabilities that nobody previously predicted. And sadly, electronic voting systems in the US have an absolutely horrible track record. Many vendors don't even seem to make the most basic efforts to write secure software. It's shocking.
And what's the benefit? Being able to vote from your phone is only a tiny bit more convenient than voting by mail, but it's fraught with so much more peril that it's just not worth it. Plus, not everyone has a phone, so you still have to provide an alternative.
Voting by mail is demonstrably secure. It's simple. It's cheap. It's convenient. It's better than e-voting can ever hope to be.