IMHO electronic voting is easy. You have a machine that is open-source (yes, open-source so that you can verify before and after operation that is has not been tampered with and is completeld "honest" as far as elections go; the voter selects their choice(s) and the machine prints out the ballot (which then goes into the appropriate ballot box).
The electronic vote can then be stored/transferred/whatever to provide up-to-the-minute results information. It is a preliminary result and shouldn't be relied on unless the candidates in that election agree that the result is what they expected. If the candidates don't agree or the result is within some "closeness" parameter then the paper votes are used to provide the final, ratified result.
The paper record is critical because it serves two purposes. First, the voter can see straight away whether the machine as recorded the correct vote. Or at the very least that the machine has printed the right vote. Second, that paper vote is what is used for recounts and any queries at a later stage. It is no use relying on the electronic record because it is immediately suspect because of computer errors or data transfer errors or whatever other error you can think of. Of course, the paper vote would be stored for a set period of time to allow electoral challenges and so on.
That's it. Simple. Of course, you would want to implement the same system absolutely everywhere do make sure that every voter gets a consistent voting experience. And you want the voting record and data transfer to be secure. All that nice stuff. But at the end of the day, the paper vote is what counts.
So what electronic voting delivers is not convenience for voters (after all, marking a bit of paper with a pencil is not so hard) but convenience and speed in reporting first-pass results. And in the event of a result which is not contested a lot of human labour and time is saved by not having to count the paper vote at all.
The issue is that this stuff won't be cheap to produce - it requires significant development resources; lots of time and money verifying the code; custom hardware that is relatively tamperproof or tamper-evident; consultation with the electorate to ensure that voters with special needs are catered for; and obviously significant legistlative changes.
While I'd like to see electronic voting happen in this way I'm not convinced it will happen - mainly for cost reasons. Paper voting (the way it is done now in Australia) is simple and it works.
I could have a rant here about the US systems of voting (yes, deliberate plural) but that is a much longer topic and I really can't be bothered right now. But the short version is they really need to sort their stuff out if they want to be called the best democratic system in the world because it is clear to anyone who has done any small amount of research that they're system is highly broken. But, a topic for another time.