Would you like your War on Terror with Type I or Type II errors?

A recent column in National Review credited President Trump with the defeat of ISIS' stronghold in Iraq. The writer was also critical of a New York Times piece that was critical of the loss of civilian life in Iraq, seemingly at the hands of Trump.

I think it's important to synthesize these two views because they essentially tell the same story. Unlike Obama, Trump decided to give more control to his generals to do what was necessary to win the war. Obama was reluctant to do so because he wanted to minimize civilian casualties. Trump was okay with that, and thus, ISIS is now very close to being wiped out.

This brings up my new fascination with viewing the world as a continuum of Type I or Type II errors. Civilian casualties are Type I errors, false positives. They were thought to be terrorists, killed, and later identified as innocent civilians. It's a pretty impersonal way to talk about murder but it helps illustrate a larger point.

By using more caution, Obama was more comfortable with Type II errors. However, withholding a drone strike due to uncertainty about the target's innocence can lead to false negatives, identifying someone as innocent who is, in fact, a terrorist.

It's a really tough decision to make and I don't think there is an easy answer. The more you drag out the war, the more American soldiers you lose. However, the more civilians you kill, the more unpopular you become abroad and the harder nation-building becomes.

I think any praise heaped upon either leader's choice should come with an admission of the flaws that come with Type I or Type II errors in the war on terror.

Rethinking Our Criminal Justice System

People have strong feelings about the flaws in our criminal justice system, which are best described as Type I and Type II errors.

Type I errors, false positives, draw the ire of progressives. This happens when a person is falsely accused of a crime he did not commit. In the worst case, he is given the death sentence. As this is most likely to happen to minorities, members of the protected victim class, progressives view this as a form of oppression and stand on their side (the falsely convicted) and against what they perceive as "systemic racism."

Type II errors, false negatives, are the bane of conservatives' existence. This happens when a criminal is set free due to something like police tampered evidence or some other fluky event. (This phenomenon is what gave rise to the TV series Dexter.)

This can also happen if someone is sent to prison and is let out early for good behavior or is sentenced to counseling/community service in lieu of prison and proceeds to commit another heinous crime. The last part is the important one. In conservatives' view, criminals are bad and should be removed from society. Police protect us from bad people so we should stand on their side and against the lawyers and activist judges who want to put bad people back on the streets.

An Imperfect System

Our criminal justice system, like the humans who designed and run it, is imperfect. There will always be Type I and Type II errors. All we can do is tweak things so there is less of one and more of another. Which begs the question, which way should we lean?

My hope is that when people think of it in terms of Type I or Type II errors, they will see that there is no easy answer—only what is less wrong.

Got Spam?

Think of it like an email spam system. You can either tighten restrictions so you don't get spam, but will occasionally miss an email you really need. Or you can loosen restrictions, never miss an important email, but have to deal with some spam.

After much deliberation, I think Type I errors are worse, making a system that leans toward Type II errors less wrong. Not properly convicting a dangerous criminal is bad. But falsely convicting an innocent person is not only harmful for that truly innocent man, it means the real criminal gets to go free.

I look forward to changing my mind several times about this; that's the great part about being pragmatic.