I mostly agree with your proposals although:

1. I am doubtful we will ever get substantively better curation without more downvotes. The feedback mechanism that makes early upvotes on crap a bad idea is downvotes, especially in a linear-rewards system which doesn't rely strongly on consensus to build significant rewards (by being superlinear). That is, if you vote (including a self-vote) on garbage and no one ever downvotes, the worst thing that happens is you get some sort of 'fair share' of rewards based on SP. If someone downvotes, then you get less than fair share (or nothing), in which case your vote power was wasted, and you would have been better off voting for something else (non-garbage) or selling your votes to someone who will do so. Or more to the point if the problem we are seeing is self-votes and other votes on content that is garbage, the system (indeed any conceivable system) needs some sort of input that identifies said garbage i.e. downvotes. Without such input we will likely continue to run in circles on the issue.
2. I would try simply removing the 30 minute rule altogether. The nature of the system is quite different now than when that rule was implemented, in at least two ways: a) far more posts total, so it isn't possible to vote on anywhere near the same fraction of posts nor even all of the posts from reliable authors (with full power); and b) linear rewards mean that voting on most popular authors' posts is far less profitable (because they don't receive approximately all of the rewards). Yes, bots will vote early on some highly-reliable authors posts, and I think that is okay. Identifying and rewarding the most reliable authors posts is an uncreateive task well suited to bots, but that doesn't mean those authors don't deserve their rewards (see above on how to deal with it if they don't). Human curators can then focus on the other posts and avoid competing with the bots (which from the perspective of good curation is a waste of vote power). Once we see how things work without the rule, if there are still problems, we can consider a new rule better suited to the current (and future) state of the system.