Exotic Agentic Development Strategies: Everything old is new again

We all know you can vibe out some slop on a grammatically incorrect, partially coherent, stream of consciousness. The question is: how do we use AI to develop code at a higher quality level? With the advent of cheap code, we should take a look back at previous failed methods of doing software engineering. Many past practices did not become popular because the cost to write was much too high. Now with agentic development being cheap, we have a treasure trove of ideas and techniques to reclaim the level of quality. In this talk, we explore multiple techniques and how to apply them to an agent/harness. We will investigate design, review and verification ideas that implement not only a coherent design, but one that meets all non-functional constraints. We will explore techniques used by JPL and NASA for maximal system reliability along with ideas promoted by Knuth and Meyer in the 1980's and 1990's, along with techniques like model driven development and property based testing that are fantastic from a theoretical perspective, and are now trivial with AI. The goal here isn't nostalgia. It's a working answer to which idea now pays for itself in an agent loop. Everything old is new again, but now affordable.