Here’s a thought-provoking email from CS professor Jay McCarthy to myself, Robin Hanson, and Kevin Simler (the latter two being the authors of Elephant in the Brain).  Reprinted with Jay’s permission:


Bryan, Robin, Kevin,

I’ve been reading your blogs for a very long time. I recently read both of your books and found them very insightful.

I’m a professor of computer science and much of the signaling case is extremely believable to me based on my experiences with students..

I’m curious if you’ve thought about how to “put it into practice”.

Both books primarily talk about the signaling theory influencing educational policy, but I’m thinking about the micro level.

It’s obvious by their actions that most students want to minimize work and maximize the chance to get a job and any learning along the way is incidental. Occasionally they believe learning something would be useful, so they will try, but mostly they think it is pointless and has no application. Perhaps, signaling theory says “Yes, it is pointless, but you have to prove that you can learn THIS, because people expect it.” So maybe it’s a tool in convincing students to stop complaining and study what we tell them to.

However, what could I do with this non-rhetorically?

I thought it about it as how a particular department/professor could approach their various audiences:

Department vs Alumni – They want more resources from their alumni, so they want better alumni that attribute their success to the department. Signaling says that their actual success will be based not on what the department teaches them but on how well the department selects them among available options, how fondly they remember the experience, and how much employers (for instance) think of them as a being a product of the department. This suggests that the right strategy is to have a really high standard to get in, make it a cake-walk once you are in, then try to get the students to talk about the department a lot. (I see Ivy League schools as operating like this mostly.)

Department vs University – Many universities give resources in relation to enrollment numbers rather than student quality, which pushes for not having the high standard. On the other hand, one way to do “admission” is to let students in to an early class to get high student credit hour numbers, then demoralize a huge number of them so they don’t continue. (If you fail them, word gets around and they stop coming.) So, you want really good outcomes for your students and to be seen as fun or sexy so that many people take the intro class, and then be really hard. Computer science presently has this going for it, although there’s a huge push to make intro CS courses (and entire degrees) easier so that more people get through (in the interest of diversity), signaling suggests this is counter-productive because it dilutes the brand in the same way college has generally.

Professor vs Students – Signaling suggests that the content of the courses doesn’t matter, because students don’t remember anything anyways. So that means cut anything that makes the professor’s life painful (which includes dealing with students whining.) On the other hand, there’s a trend in computer science for engineering companies to have interviews where applicants solve problems from a few of the most popular textbooks—which is a consciously admitted outcome of companies not being able to trust degrees (a failure of the CS degree signal.) It’s like they are trying to test for the combination of “good CS program” and “student that remembers things”.

In all three of these cases, signaling suggests that you ignore the “curriculum” and instead focus on the statistics and information about the students. Have you thought about these problems at all? I’ve heard in quite a few interviews that you’ll say stuff like “Well, in really technical fields, like the sciences, this might not apply as much.” I cannot tell if you are saying that backed up by the data; or if the science brand is just really good that we got you too; or what.

FWIW, in my experience a huge amount of CS curricula has very little with what software engineers really do and the thing that predicts the most success for a student is whether our classes give them enough of a taste to fall in love with it and decide that they want to practice a lot and learn on their own by actually “doing”.

Sorry for the long ramble, the books are just so thought-provoking it’s hard to not think about what to do with it,

Jay