For a rich source of engineering failures, just pick an engineering standard or certification requirement and research its origin. A great many of today's engineering requirements were written in blood.

The Fokker Trimotor crash that killed Knute Rockne hastened the understanding of wing-aileron flutter and ultimately led to modern flutter test methods and flight clearance procedures. The Comet fatigue failures led to including pressurization cycles in today's structural lifetime tests.

Asymmetrical "g" requirements and the rolling pullout maneuver came from WWII experience with Corsair fighters. It took a while to understand why Corsairs were coming back from missions with the wings permanently deformed. It turns out that the Corsair had a heavy armor plate behind the pilot's seat, and on ground attack missions, the pilots would pull up and roll to put the plate between them and the ground fire. Engineers at the time didn't know to consider the added load of rolling while pulling hi g's.

The crash of the YF-22 called attention to the need to give greater consideration to mode changes in modern flight control systems. It also underscored the need to thoroughly test for susceptibility to pilot-in-the-loop oscillations. I'm sure you'll see changes in the way the F-22 will be flight tested and in future versions of MIL-STD-1797 as a result.

The process is continuing today: the ATR turboprop crashes have increased our understanding of aircraft icing; in particular, the way freezing rain and runoff from the deicing system can form a ridge of ice aft of the deicing boot. The ATR was designed and certified to the known requirements of the time and once the new environment was understood, the company moved out expeditiously to adapt their aircraft to it. Future icing certifications will undoubtedly include tests for freezing drizzle.

I believe the Comet was an example of the state of the art's ignorance at the time. Previous aircraft had used such conservative static strength margins that they had essentially infinite fatigue life - look at the DC-3. The Corsair experienced usage that was different from that envisioned by the designers. Likewise, I don't see how one could classify the ATR accidents as a gross engineering failure, even though engineering changes were part of the solution they adopted.

Is there value in making distinctions between engineering failures? Or are the dead just as dead, whether it's negligence or ignorance?