Monday, April 9, 2012

On Ironing Out the Difference Between Art and Science

Two short stories. I made them up, but they're almost certainly not fictional.

Auto engineer goes to his boss, explains that the reason the new car prototype didn't reach new peaks of efficiency is not his fault. His materials weren't good enough, the technology isn't there yet, whatever. The boss buys it.

A doctor goes to his boss, explains that the reason his patient died was that the patient was special. It was a harder case than the one the other doctor had, that, while superficially similar, lived. The boss buys it.

The problem, of course, is that the engineer next door isn't working on a special car. If they get their efficiency up, it becomes hard or impossible for engineer #1 to continue to sell their excuses.

We often say engineering is objective and a science, while medicine is much softer and more of an art.

I think I now understand the actual causal difference - it is in unmistakable facts. Human brain, even untrained brains, have a certain baseline of epistemic reliability. Certain facts and relations it cannot be fooled about. In medicine, there is no direct connection between these facts and the situations on the ground.

You can verify the unmistakable facts with the engineering example. The boss buying it is implausible. Similarly, when you go to make coffee, your coffee grounds are where you think you left them. If someone tries to argue that you don't know where your coffee is while the stuff's in your hand, it just makes you think - and viscerally feel - that they're crazy. The car engineer isn't quite as clear-cut, but it is close enough.

As for the direct connection, that is the idea I'm suggesting should be verified. By understanding the cause of a field being arty, it should be possible to fix the problem and make it sciency.

A list of unmistakable facts - anti-biases or anti-fallacies, if you will - would be helpful, but probably not necessary. The extant categories of 'objective' and 'subjective' fields seem reliable enough to me.

Thinking about it in specific, concrete terms should also bestow a better sense of how and when it is easy to be fooled.


Aretae said...

Level of complexity matters as well.

In building a building, there are maybe a dozen variables in play.

In building a car, there are probably a few dozen.

In fixing a person, there are hundreds, at least.

Very simply, humans have yet to get to the point of handling hundreds of variables as science. Though...computers can.

Alrenous said...

I wonder if training to deal with variables is as good as it can be?

I wonder if the number of necessary variables can be hacked?