So, I read the OP again, and have a few thoughts.
1) is this e-dem for the election of representatives, or for deciding on the bills and resolutions that representatives usually vote on in congress? The two ideas get conflated about 10 posts in.
It's for deciding resolutions on issues. One neat feature of proxy-voting is that it gives you a built-in structure of authority, should you wish to structure your leadership tree that way. I also think that proxy-voting is a neat way to deal with the information overload that an e-dem system would create. So I can see how that got conflated, sorry about that.
2) proxy voting and approval voting have very little to do with the medium in which they're employed. it is just as easy to do proxy and approval voting with paper ballots as it is with e-dem. So e-dem needs to offer something paper balloting doesn't. So far, reversal of vote seems to be a main difference.
It allows for smaller issues to be addressed more quickly. Health Care Reform - a massive percentage of that was agreeable to both sides, but the small percentage of partisan catnip prevented anything happening for months. Also, this would seem to marginalise highly partisan groups, since they'd no longer be able to hold up the whole show. Voter education, apathy, participation. Things actually getting done is better than nothing getting done, even if there are a few fuckups.
I'm summarising here, partly because I've gone into more detail ITT, but also because it'd be a good thing to chew over and add to the OP.
3) e-dem seems to offer the immedite proposal of ideas from citizens. But is there a vetting process, or do people just vote, change their vote, and see what happens, like some massive Jennifer's Government/Nation States game, where there are a dozen or so new proposals every day, from "more money to schools" to "bring back segregation"?
In the OP I talked about the need for a new method of group communication, for the reasons you raise above.. now I think that the solution will involve grouping people into their areas of self-proclaimed expertise, and layering concepts from social networks on top of that, particularly getting interesting content from friends-of-friends-of-friends-of-etc. E.g. I might not know much about a particular subject, but if a bunch of people in my network who are knowledgeable it are really excited about a new issue, it can bubble up and appear on my radar that way.
So you'd essentially crowd-source filtering the "schools" issues to one group of people who care about it, and "civil rights" issues to another. At the end of the day, an issue would just be a link, so anyone could bring attention to any issue in a manner of different ways.
I haven't seen much/any research on this front from metagovernment -- most of the projects in testing average around a dozen or so participants. But it is a vital issue, and it's still an open question whether it can be addressed.
4) if people have direct access to issues, who decides which issues they can vote on? Something as large as Iran could get horrifically messy if people keep changing their votes on sanctions, invasions, nuke attack, etc.
I'm going to stay away from national security for now, unless anyone really wants to get into it. I'm going to address the switching problem in my response to RWHN.