I'm not going to claim that "current" electronic voting machines are good, but the article makes no convincing arguments beyond an "I am a computer scientist" appeal to authority that it "couldn't" work. Honestly if you are going to talk about the problems involved in voting at least discuss some of the interesting anonymity preserving and deniable cryptographic techniques people have come up with- ring signatures, blockchain verification, etc. Maybe he's just assuming the powers that be would never let an actual trustless system get into play. If I can manage and view my money trustlessly with, e.g., bitcoin (and verify all transactions back to the original block) then there is no reason I shouldn't be able to do the same for votes - and verify all votes back to the original block.