I've yet to meet a human being that came with an operator's manual, a maintenance manual, schematics, engineering diagrams, a diagnostic and logging system designed to survive catastrophic system failure, or a multi-billion dollar megacorp full of thousands of people whose jobs depend on being able to perform root cause analysis after catastrophe. There's a whole dimension of failure analysis that is uniquely permitted by human-designed engineering systems, thanks to our comprehensive understanding of their construction and capabilities, which allows us to narrow down root causes for failures with great precision. We can do that to some extent with people, for their less complex and more common failure mechanisms (medicine)... but I don't think "deciding to go on a shooting spree at a school" is the kind of failure mode you can pinpoint to some specific hardware or software phenomenon (at least not today), let alone how to fix it. The level of scrutiny applied to human-designed engineering systems, when instead applied to interpreting tail-end-of-distribution human behavior, does not yield an improvement in root cause specificity, and likely wouldn't yield any actionable design improvements. Instead we can really only suggest process controls, based on inferences about which particular parts of the process have the most influence on the outcomes. There are so many inputs into the black box of human behavior, and so many functionally unverifiable assumptions about the mentality of the tail-end-of-distribution people who become mass shooters, that it's extremely challenging to identify process controls that would be both effective and feasible to implement. Harder still when many suggested process controls also contain embedded political goals that have little to do with controlling the process, but which can be speciously furthered under the guise of preventing high-impact system failures.
All this to say, this is a hard problem, composed of thousands of overlapping adjacent elements, we have very little insight (or reliable insight-gathering mechanisms) applicable to the system under investigation, and so throwing money or effort at the investigation is a gamble that a feasible solution exists at all, rather than a process of concrete deduction with a deterministic endpoint. Maybe the optimal amount of scrutiny should be different.
