« Workshop: Interdisciplinary Workshop on Information and Decision in Social Networks | Main | Statistical Analysis of Terrorism »

November 18, 2010

Algorithms, numbers and quantification

On the plane back from Europe the other day I was reading this month's Atlantic Monthly and happened across a little piece by Alexis Madrigal called "Take the Data Out of Dating" about OkCupid's clever use of algorithms to increase the frequency of "three-ways" (which in dating-website-speak means a person sent a note, received a reply, and fired off a follow-up; not exactly a direct measure of their success at helping people find love, but that's their proxy of choice). It's a thoughtful piece largely because the punch line resonates with much of my recent feelings about the creeping use of scientometrics in the attempts of higher eduction administrators to understand what exactly their faculty have done or not done, and how they compare to their peers. (I could list a dozen other ways numbers are increasingly invading decision-making processes that used to be done based on principles and qualities, but ack there are so many.) More generally, I think it puts in a good perspective what exactly we lose when we focus on using numbers or algorithms to automate decisions about inherently human problems. Here it is:

Algorithms are made to restrict the amount of information the user sees—that’s their raison d'etre. By drawing on data about the world we live in, they end up reinforcing whatever societal values happen to be dominant, without our even noticing. They are normativity made into code—albeit a code that we barely understand, even as it shapes our lives.

We’re not going to stop using algorithms. They’re too useful. But we need to be more aware of the algorithmic perversity that’s creeping into our lives. The short-term fit of a dating match or a Web page doesn’t measure the long-term value it may hold. Statistically likely does not mean correct, or just, or fair. Google-generated kadosh [ed: best choice] is meretricious, offering a desiccated kind of choice. It’s when people deviate from what we predict they’ll do that they prove they are individuals, set apart from all others of the human type.

posted November 18, 2010 10:41 AM in Thinking Aloud | permalink


Just wait until you're up for tenure. :( Actually, this isn't close to as bad in mathematics as it is in, say, biology, but things like h-index are used blindly for way too many things in the tenure and promotion process. (And it will not surprise you at all to know that there are horror stories.) I can't say how it is in CS. For all of its faults, one of the things (and there are many of these, too) that mathematics does well is that citation count isn't overcredited as much as it seems to be in other fields. I think this has arisen because even most of the best pure mathematics papers don't have very many citations, so the culture is quite different. When faced with my career choices, I actively decided that my strong preference was to be in a math department...

Posted by: Mason Porter at November 18, 2010 12:13 PM