I don't know what the US does, but in Germany all ballots are poured onto a big table and then sorted into staples for each candidate / party. Especially since the votes / crosses are always at the same position for each staple, it is trivial to keep an eye on 5, 6 staples at once, and the remaining parties get almost no votes anyway. After that, the staples are split into 10s and counted by two people, independently, after each other. All results are called out loud. The results are also given to the city hall via phone, so everyone in the room can hear it. City hall publishes the official results per party per polling station as nice images, easily digestible. I don't see how this is next to impossible to supervise, even for a single individual.
If you want, you can even stay the whole day and keep an eye on the whole voting process except when the voters make their choice behind the privacy screen, you can see everything which enters the ballot box. I've seen myself someone regaining trust in the democratic process because of this - a guy who openly accused us, the polling station workers, of voting manipulation, being openly hostile, agreeing that everything was done correctly in the end. This would've been impossible while using electronic or online voting.
> it's about a million times easier.
If you, as a random citizen, know SUM() and even think about downloading the data for Excel, you are the top-n% in computer literacy. You are aware of that, right? ^^'
For most people, verifying their own vote on a website with no understanding of the underlying process is the absolute maximum you can expect, IMO. In this case, it is "trusting a magic box with a display". You compared it yourself with HTTPS, for which the same is true for the general public.
> What's important to rely on is the fact that people are able to verify that their own votes are consistent with what's in the public votes
But will people do this at scale and do people trust that they do so? The latter is the most important. It doesn't have to make statistical sense, it's about feelings in this case.
Because if most people (of a certain demographic like the elderly) don't check their own vote or a significant amount of people don't believe that they do so, you cannot automatically assume that all votes in the polling station have been counted correctly. It may have been e.g. only the votes of certain demographics (who are unlikely to check their own votes), which have been tampered, even if this believe is statistically unjustified.
If you supervise a whole analog polling station, you see for yourself this is not the case.