Tuesday, October 20, 2015

Science is hard

This is a great story of how the cure for scurvy was forgotten. Basically, the British realized early on that lemon juice would prevent it very effectively, but they didn't really know why (and perhaps more importantly, they seem to have been fairly unaware of this ignorance). So when they switched to other solutions that various theories held would be equally effective (eating preserved limes with the vitamin C destroyed, avoiding tainted meat but failing to eat fresh meat, etc), the scurvy came back. It wasn't until the mid 20th century that the true explanation was finally verified. Turns out, science is hard: it's really easy to come up with explanations for facts and really hard to be sure which one, if any (of the ones yet thought of), is right.

Serendipitously, even though that story was published a few years ago, I read it the same day this perfectly appropriate xkcd cartoon was published:

I like to make fun of engineers and physical scientists for how easy they have it* since rocks kinda just obey a few laws and are easily controlled and predictable. Experiments are easy to control and replicate and there aren't the plethora of confounding factors that come with humans being human and exercising their infinitely faceted free will. I do think this makes economists very good at thinking about alternative explanations and being very harsh judges of any inference from data; the success of this approach is why economics is invading the other social sciences and even medicine. On the other hand, introspection is useful guide when trying to think of new hypotheses about human behavior that the physical scientists don't have at their disposal, and the obvious difficulty that created for scurvy makes me(even more) amazed at how far science came in such a short time. Good job guys.

*Yes I'm kidding.

[Link stolen from SlateStarCodex]

Thursday, October 15, 2015

re-SMBC part 6

More accurate redo of SMBC number 3823:

This is a bit inaccurate, because of course an economist wouldn't even ask why...

Sunday, October 11, 2015

Thanks UT Austin!

Daniel Hamermesh, one of my very favorite economics bloggers, is heading to Australia, at least in part due to concerns over campus carry legislation at his current institution in Texas. Thanks Texas!

I wonder whether his UT colleague Max Stinchcombe, who visited UQ most of last year, told him how nice Australia is. Word of mouth may also being playing a role at Berkeley: four Berkeley Econ graduates in three years (and counting, I hope) have crossed the pond.

(I mean, frankly I'm surprised it's not higher; doesn't Australia seem like a better option than being forced out of California after 6 years of learning to appreciate a good climate and the great outdoors?)

Also: I really love to see voting-with-your-feet in action.

Thursday, October 1, 2015


I was traveling for most of August and September but now that I'm back in my office, I've gotten caught up on most of the blogs I follow. A few weeks back Alex Tabarrok had a cute mathy post on the Frechet probability bounds with the warning "super wonky!" in the title. My reaction was first "duh" and then "wait, this actually has a name?" (This probably goes a long way to explaining my teaching evaluations :)

I love Alex's explanation and love that he posted it; I in no way mean that it's too trivial to be interesting. But it's such a perfect example of something that looks scary if you write it down in formal notation and give it a high-falutin name, and it scares people off who don't read math all the time so that you have to put "super wonky!" warnings on things, but if you think about a simple example for a few seconds it's incredibly simple. Much simpler than the concepts involved in most of Alex's non-wonky posts.

Anytime I teach anything involving math (and I try not to teach anything that doesn't involve math, math is too fun!) I spend a significant amount of time trying to convince students that math is not actually scary and that they definitely shouldn't memorize equations for everything. If you understand that a linear relationship is just something that starts at a particular spot and then follows a particular slope, you don't need to memorize the equation for a line, or the point-point formula, or the point-slope formula, and I loathe that the ubiquitous approach presents those three separate equations that students then copy down and memorize.

Here's an example of the Frechet probability bounds. If 80% of people are Caucasian and 70% are Christian, there has to be some overlap. In fact, at least 50% of people have to be both Caucasian and Christian. On the other hand, there might be more overlap; it could be that 70% of people are Caucasian Christians, 10% are other Caucasians, and the other 20% are Buddhist Asians. Take a second to picture the possibilities and those numbers will be obvious to you; if you get stuck, see Alex's post for a nice way to diagram it.

That's it. One tiny concept (your groups of people have to overlap somehow to fit within the population you have) with some fancy notation and a fancy name that scares people off.

I should add, I'm not actually opposed to fancy names, in fact I love jargon of all kinds when used to improve precision and economy of language rather than to obfuscate or sound smart. When I went to mathcamp in high school and learned the name for the "pigeonhole principle" I also thought it was hilarious that this warranted a name - all it says is that if you have more pigeons than holes, some pigeons are gonna have to share holes (in fact this is the fundamental concept at work in the Frechet bounds as well) . But it's actually great that it has this name because it can be applied in creative ways in much more complicated proofs that don't immediately seem to have anything to do with counting holes, and all you have to do is say "by the pigeonhole principle" for the experienced reader to quickly deduce what are the pigeons and what are the holes and how their relative number is relevant, without you having to spell it out.