and while you're at it, eat your broccoli

I adopted a new podcast this week: Radio Lab. This American Life keeps joining forces with them, so I figured, why not?

This week's episode was about testing the limits of the human body/brain/species. Yup, you read that last one right. The final ten minutes of the show was devoted to a group of roboticists at Cornell who have designed a giant computer to answer open-ended questions. What do I mean by that? They built a double-pendulum, put it under a video camera, hooked that camera up to their giant computer and asked "what's going on here?" After about a day (with no background knowledge, just the video footage), the computer came up with the following answer:

Force equals mass times acceleration.

Yes, after one day watching one (admittedly chaotic) pendulum, this computer figured out the fundamental law of modern science that humanity (acting through Aristotle, Sir Isaac Newton, and everyone in between) took about 20,000 years to discover (depending on where you reckon the start of humanity).

Let's just take a moment to appreciate the beauty of that.

Let's take another moment to freak out about the fact that we are teaching robots to do science on their own. Crap.

Let's move on. At the end of the segment, some biologists feed the computer years of data about the behavior of single-celled organisms and ask the same question. The computer ponies up, delivering relatively simple equations that not only explain the data but predict future data. Unfortunately, the biologists (or really any human, actually) have no idea what the equations mean or why they work (foolish mortals!). To quote the radio program "they have the answers, but not the insight." Anyone who's ever taken a physics class knows where I'm going with this. 400 years after the birth of modern science, researchers are looking at the back of the book.

Any text book that is actually useful will include the answers to every other problem in an appendix. Most professors assign at least some of these problems as homework. The thinking is that once you do the problem, you can peak at the answer and, if the book disagrees with you, you can go back over your work, figure out where you went wrong, and hopefully learn something about either your intuition or your ability to do arithmetic. It goes without saying that merely handing in the answers without showing your work (usually not included in the back of the book) will net you precisely zero points.

It is oft lamented that what we do as students in classrooms (listen to lectures, sit exams, do homework problems about spherical cows in frictionless vacuums) has very little to do with actually being a scientist. Apparently, that is no longer true.