What you keep denying is that there are solutions (all solutions, and provably so in the case of BGP) that solve the problem within a specified range. Generally up to 33%-of-generals in the case of BGP and maybe 50%-of-hash rate for Bitcoin.

There is no solution that solves the problem up to <100% of participants. If you introduce identities then you have made it worse in a sense because now a failure of just one component -- the certificate authority -- breaks the entire system, instead of many (33%/50%/etc.) failures.

On the topic of the thread, I consider Bitcoin and BGP as distinct, but related, problems. The setup is quite different, and I haven't seen anything close to a method (including Satoshi's email) to reduce one to the other.

It is possible you can define another related problem that is in turn more useful than both Bitcoin or BGP solutions for some practical application. You still have to overcome Bitcoin's network effect even if your approach is somewhat more useful.