Saturday, April 15, 2006

dispatch from storrs

So, yesterday was the ASA Methodology conference. I did my presentation arguing that quantitative sociologists should be depositing code sufficient to reproduce analyses at the time of publication. I think it went all right. Actually, I think the basic idea went over really well, and the presentation was okay. I've been continuing to develop the argument in response to conversations I've had with people. The slides for the talk are here.

Originally I was going to drive back to Cambridge right after the conference. But then the conference ran long--instead of the last talk ending at 4:30, as was scheduled, it ended at 5:50. And, even more importantly, someone told me that taxes weren't actually due until April 17 instead of today.

Update: A commenter asks, "What kind of opposition were you expected, other than inertia?" The chief point of non-inertia opposition I have gotten from people who actually do quantitative social research--indeed, I suspect this may be a formidable obstacle--is that some people feel like the code that goes into analyses takes a long time to write, and they feel like they shouldn't have to just give this code away to others by making it publicly available. In other words, some people have basically the same stance regarding code that some people have about data, and so the various proprietary arguments against sharing data are re-presented with regard to code.

I have wondered whether this kind of argument may have greater traction in sociology rather than economics because programming skill in sociology are more scarce and might result in sociologists having more of a "hoarding" attitude toward the code they write (when, from a collective-knowledge-production standpoint, the scarcity implies that it would be even more valuable for sociologists who are good programmers to share their code than economists).


Anonymous said...

Other than inertia, what kind of opposition were you expecting?

jeremy said...

I decided this was worth a post update to answer.

sarahliz said...

Alternatively good programmers who are lousy documenters might feel some apprehension about making that known to the world. I know I get nervous when I give code to classmates because a lot of the code they want is stuff from my MA project, when I was an even worse documenter than I am now. If I had to go clean up all the early code to make it public (the project I'm currently working on builds heavily on that code) it'd take a lot of time. But I wouldn't want to release my code as is because some of it is kind of embarassing (both in its lack of documentation and its inelegance).

Brayden said...

The solution to the corde-hoarding problem is that by releasing your code widely you increase the number of citations you potentially receive (as long as people who borrow code directly properly cite the source).

jeremy said...

Actual evidence to this effect exists with respect to providing data rather than hoarding it, and yet the belief persists that you are better off hoarding data. In more ways than this, I have been regularly surprised by the ways in which social scientists damage their own interests through worries that someone else will benefit from their labors.