This approach leads us to an increasingly complex and awkward scripting system, as we add more and more special cases which must be maintained forever. If we can restore script as it was originally, as a general programming language, we don't have to try to "pick winners", and you don't have to ask permission to do what you want with your own UTXOs. Doing this safely and cleanly is my current area of research. My favourite thing about this approach is that following soft forks become less about "is we had this new feature we could do X” and more "this would make existing scripts more efficient and allow us to do X% more in a block" which is a much more quantifiable assessment.

That's exactly how it is! After months of Rubin aggressively trying to push his fork, one day (after much thought) he came up with another idea! Namely that his first thought was not the right one and that no change to Bitcoin was needed! Maybe it's due to his youth or lack of experience - or both - but this behaviour has been seen time and again from some developers over the last 15 years. Too fast, too impatient and too superficial. Fortunately, Bitcoin can fend off these "improvements".