Friday, September 16, 2016

A rough consensus

When I tried to begin my CS121 lecture Thursday, the projection system was not working. As the lectures are live streamed to a distance audience taking a parallel course through the Harvard Extension School, I couldn't proceed. Stuck with nothing to do until the tech arrived to diagnose the problem, and a captive audience of about 150 undergrads, I decided to learn something and teach something.

I was curious what they thought of the new College policy on single-gender social organizations. Now this is not an unbiased group--I had told them on Tuesday to read my op-ed, but I didn't suggest any reading to give the arguments in favor of the policy. (What would I have suggested?) And of course they know me, and I had just fed them donut holes. And they are computer science students, majority male, freshmen under-represented because of the level of the course, not a random sample of the student body.

Even acknowledging all the biases likely present in the group, I didn't want to ask them for yeas and nays in a way that would make them reluctant to acknowledge publicly what they thought. So I used the protocol of the Internet Engineering Task Force: Rough consensus, determined by humming.  No one can tell whether you are humming or not; and if there is a rough consensus, it's apparent. Of course you can't decide a closely contested question this way, but the genius of the IETF is to avoid votes entirely. All engineering design decisions, after the very first ones, were made by rough consensus--without a consensus, you don't change anything. (The memorable expression is "rough consensus and running code," but the "running code" part is not in the by-laws. It's just a signal that you get a leg up if you can show that your dandy idea can be developed quickly to a state where it sort of almost works.)

So we tested, by humming, whether CS121 students were in favor of the new policy or not. The consensus was there and it was overwhelmingly negative.

Now maybe if you ran the same test in a Gov course you'd get an entirely different result. (Hey, Prof. Porter, how about trying it in The American Presidency?) But still. It was astonishing. I am sure I read somewhere (the Crimson?) that students are in favor of the policy. It never went through the student government or any of the official student-faculty committees, so there is no evidence there.

If anything, I would have expected CS students to be less sympathetic to off-campus social clubs. I think, and the class seemed to agree, that they are underrepresented in those organizations, probably because they are working too hard. And yet the favorable hums were barely audible in a large lecture hall.

By the way, doesn't that suggest a solution to the Final Club problem? Just ramp up the workload in Ec and Gov to be more like it is in CS.

Bonus link: a story about my senior year. Recommended reading for students who think their senior year is stressful! Professors' Professors

1 comment:

  1. The linked Professors' Professor piece made me both nostalgic (turns out its somehow been more than a year since I graduated!) and relieved. I still cringe when I think back to my dreadful errors in CS 20 my freshman spring!