Just livestream the whole process.
However as long as tge machine is just a scanner (still paper trail) much of the same argument spplies.
People commonly do rely upon encryption schemes in precisely this way to protect their finances, healthcare, privacy, and so on (and in spite of constant implementation failures!). I would revise the premise to something like, "it will several decades for users to build trust such a system".
> In my opinion, e-voting is a solution to a problem that never existed.
I believe using it to re-implement existing voting schemes is misguided; it would be like using the internet to replace in-person banking with video calls to a bank teller. New paradigms are more exciting to me than recreations of existing paradigms, even if some features are sacrificed (e.g. cash deposits, in the banking example).
The geographically-hierarchical systems we rely on currently evolved based on their useful properties of being communication- and computation-efficient. That is, holding elections which satisfy the Smith criterion [0] or offer Byzantine tolerance [1] would be difficult to achieve if all you have is paper, pencil, and Pony Express. It also supports why elections are held relatively infrequently -- sending politicians overland to gather together to perform some kind of in-person consensus mechanism (e.g. quorum establishment) isn't something you want to do all the time.
We have back-justified this system with the rationalization that local leaders best reflect the concerns of their community, but this is an enormous simplification -- essentially, it takes the preference space of the electorate (all the issues they care about), embeds it down to one dimension (the list of candidates), and then re-projects it back to the higher space (the candidate's stances). Preserving the feature space by holding more frequent elections [2], or allowing users to choose which features they care about through issue-based vote delegation [3], would be more interesting to me.
To go back to the previous point, as complexity increases, trust becomes more difficult to grow, and existing mechanisms also have thousands of years as a head-start. It will take time, but I hope that people will be open-minded enough to try it at some point in the future. Maybe it will be as low-stakes as how to stock the office fridge, or which equipment to purchase for the local park. But these are the kinds of baby steps that will eventually enable better governance systems.
[0] https://en.wikipedia.org/wiki/Smith_criterion
[1] https://en.wikipedia.org/wiki/Byzantine_fault
In my opinion, e-voting is a solution to a problem that never existed.