I think that schemes like this are doomed to failure.
Right now, a bitcoin transaction has very simple rules. There are really only two:
A transaction is valid if it satisfies the conditions specified in the inputs it redeems.
A transaction output can only be redeemed once.
That's really all there is to it. A powerful and useful system defined by just two simple rules. The pinnacle of engineering and beauty, nothing left to take away.
Now, our computers are really dumb, so things that we take for granted, we have to explain to them in great detail. So, the software isn't as beautiful as the idea it represents. We have to tell the computer that you can't have a negative output to balance out a too-large output, and things like that.
Assume that we devise some way to allow the network to take action against double spends. There are horrifying technical challenges in that task, but we are going to assume that condition, so we don't have to worry about solving them. Any such scheme is going to destroy the beauty of the system. There won't be two simple rules that you can explain to literally anyone in 5 minutes, there are going to be three. And the third one is going to be ugly. It is going to start out with "Except when..."
The first rule is math. The second rule is time. Both of these are necessary, in that you can't have a system like this without math and time. The third rule will be the mob. The third rule will be the one that we added to make some particular group of people happy. The third rule will be the foot in the door for the fourth rule, to make some other group happy. And then the fifth. Next thing you know, we'll have a group of people meeting on an island to add a rule that will solve the world's problems.
Fortunately, I think that people see this. Even if we could distill the problem down to "just a bit of code", it would destroy the beauty of the system. The software could take the hit, lot of it is already ugly. But it would be fatal to the idea that the software represents.
I like your romantic view of Bitcoin and aesthetics sure are important. Simple solutions that work well is where brilliance of mind shines. If we can improve the system by making some small changes, though, I think we should still to do it. Maybe it will even bring us to an improved simple solution.
I'm not convinced my exploit has been resolved, but it may be a misunderstanding of the proposal here. Either way, you haven't convinced me this proposal is sound, and the burden of proof is on the ones claiming they have found a way to make zero-conf Tx safe. If you are really serious about this idea, program it into an altcoin or testnet and get people to run attacks against it. If you are successful, you might create a valuable altcoin or be responsible for a critical fork to bitcoin. At this point however I'm done examining this proposal which is difficult to discern without code.
Thank you for input. I think the concept in the basic version is simple enough to make thought experiments with it. I invite you and others to make more of these about it.
That's how I see it, too.
The bitcoin community is extremely opposed to hard-forks, so don't be discouraged by the lack of enthusiasm. Major changes are only practical in the altcoin community, so consider promoting radical new ideas there.
Yup, by now there is so much money and politics involved with Bitcoin that a change like this can only be implemented once there is a successful altcoin testing the change. Maybe it even needs to rise to a level of popularity that makes Bitcoin devs uncomfortable before it will be seriously considered.