So, some 25 years after the advent of the "AI Revolution", in-the-trenches programmers are sitting around debugging J2EE stack traces that look like this insanely over-architected slop.

With programmer productivity advancements of such magnitude, can the Singularity be far behind?

Look, the reality is that CS theory (including AI) has been fairly stagnant for 20-30 years. Applied programming technology is higher-level than it was a few decades ago, but what about the theory?

Dumping 43 million layers of J2EE bloat (which is based on a mediocre implementation of decades-old OO techniques) on the programmer might give software architects stiffies, but it doesn't get us any closer to the Singularity.