InstaGov consists of the following components, starting with the most important and building from there:
1. A system whereby anyone can post a question for referendum and others can vote on it.
This will use range voting (rate your approval on a scale from -10 to +10, say, rather than just binary yes/no), to allow voters to show their relative approval of the various options and arrive at the best compromise.
- Possibly there will also some verification of attributes which may be relevant, such as professional qualifications and area of residency -- but that's one of those refinements that can wait; users will probably end up proposing better refinements than I could ever think of, and the system can be used to decide which ones to implement.
I'm also inclined to keep secret balloting to a minimum, to encourage discussion and taking a stand (the system should let you see how any given individual has voted on non-secret items). Yes, it causes problems (vote-buying, intimidation, etc.) -- but I have some ideas for dealing with them, and think they may be less than the vote-verification problems we've been running into with the current system.
2. A system for managing these referenda so that individuals are not flooded by issues of no interest to them: Each referendum will be assigned to one or more "topics"; each topic will have a "feed" to which interested voters can subscribe. Voters can also rate whether the referendum belongs in the assigned topic or not, in order to prevent spamming and minimize mis-filed referenda.
There will probably be a feedback mechanism to reward posting of relevant referenda in appropriate topics and discourage individuals from promoting particular referenda in exchange for material gain.
3. A system to enable structured debate on contentious issues, so that logical dependencies can be established on some items. Ultimately such arguments come down to premises which can't be debated -- so participants should be able to pick their premises and see which of their positions are inconsistent with those premises, as determined through the ongoing debate process.
A less-crucial feature (one of many possibilities):
4. A means whereby users can choose a "circle of advisors" from among all other voters on the system (and possibly from a few "virtual voters" representing the reconstructed opinions of public figures who don't have accounts). Each user could assign a "credibility" weighting to each advisor (possibly negative), and the system would present the summed results of the opinions of all their advisors on any given issue. Each user's weightings for each of their advisors could also be viewed from the other direction – statistics about the weightings given to any given advisor, reflecting that person's overall credibility. (This has evolved into the proxy system.)
It seems to me that just the first two items could seriously change the world; the third item should help keep it civilized. There's a lot more which could be done by building on this infrastructure, too.
Also, key point: the idea is not so much for this to be a centralized system but rather to create a set of standards for managing these things, and (free/open-source) software to implement them -- so that if any one implementation becomes corrupted (e.g. the way Conservapedia corrupts the wiki concept by censoring rather than discussing), voters can move to another one (or set up their own).
Part of the standard needs to be a format for data-exchange, so the various sites can pool their data and aggregate their agreements. If there are a thousand sites, and a thousand users on each can all agree on a set of principles, that would be a million people agreeing on something. What might not be accomplished?