Finding the Needle in the Haystack

The two instances on the news recently where they have caught kids with arsenals ready to do harm BEFORE they did anything are applauded for very good reason.  Astute people observed their actions and what they were posting on social media.

However, the challenge no one seems to want to talk about is the more general question of how you find these individuals and help them before they go off the deep end or intercede before they actually commit such horrible acts.  Yes, we have way too many of these events, but how willing are we to truly address them.

My premise is that one measure, such as better checks on the sanity and wellness of the gun buyer or owner will do little to stop mentally unstable people from doing things like this.  They will get what they want and need somehow.  They will learn to not post their evil ideas online.  Bad people are not stupid.  Speed limits and police do not stop speeders and the horrible consequence of their actions.

My wife Susan and I were going to work and were harassed by someone wanting to speed, slaloming around all the cars.  Just as we were shaking our heads in disbelief a police car pulled them over.  Oh, that was a sweet moment and I will cherish it… but it is way too rare.

Things like this should cause us to think deeply about the realities of avoiding rare and high consequence events.  Yes, they are rare as percentages go, but they are way too common in aggregate.  The problem is that they are rare so detection and avoidance pose a very real opposite problem: the invasion of privacy and intrusion plus false accusations of those who MIGHT do something this bad.  There are thousands of these for everyone that does them.  So, are we willing to embrace a police state?

Now, let me take you into my world of artificial intelligence, which is precisely the software they would use to cull out these prospects.  The errors in this process are called alpha and beta errors: false detection and false nondetection.  That is even with the best of these computer algorithms you run the risk of thinking something is true when it is not, and then even worse is thinking something is not true when it is.

As you tighten down the criteria for either one of these, you make the other one worse.  That is what we face as a society.  But, the best we can all do is be on the alert for signs of trouble: if we see something, say something.  That at least can help in the short run.