Whence 'Spam'?

The names we give everyday objects can emerge from almost nothing, a new experiment shows.
Author:
Publish date:
(Photo: Chris Dlugosz/Flickr)

(Photo: Chris Dlugosz/Flickr)

We refer to unsolicited email as "spam," just as we call coffee, well, coffee. But why do we do that? A new experiment suggests it's actually kind of random—as a group, we generate lots of possibilities, but eventually settle on just one or a few conventional names for a given object without any external interference.

Social norms are central to our lives—just imagine the chaos that would ensue if we stopped driving on the same side of the road—and linguistic conventions, such as our tacitly agreed-upon vocabulary, are no exception. Yet how we picked the conventions we did—how we chose "spam" to mean "unsolicited email"—isn't obvious. Perhaps, some argue, media leaders like the New York Times use a new word and the rest of us follow. Another possibility is that societies somehow select widely used terms through person-to-person interactions alone, without any top-down influence from media or other authorities. 

So which theory is correct? That remains an open question, but network scientists Damon Centola and Andrea Baronchelli have now shown that settling on conventions bottom-up is at least possible. In their experiment, 510 people played several dozen rounds of the "Name Game." In each round, participants were paired off, and each had to come up with a name for an object that appeared on their computer screens—for example, they might have to give a name for a person depicted in a photograph. Each player saw what name their partner chose, and if both chose the same name, each one got a 50-cent reward. Otherwise, each was docked a quarter. The round complete, players got new partners and played the game again. Crucially, the object a person was to name stayed the same in every round he or she played.

Dominant names for the objects always emerged, often fairly quickly. For example, when participants were paired off completely at random, players universally settled on one name for the object within just two dozen rounds. 

Though the details of the results depended on the particular algorithm Centola and Baronchelli used to pair off players, dominant names for the objects always emerged, often fairly quickly. For example, when participants were paired off completely at random, players universally settled on one name for the object within just two dozen rounds. 

In another version of the experiment, the researchers created a kind of social network. Each person was linked to a small number of peers and played each round with one of them. Those players had their own, overlapping groups, similar to groups of friends on Instagram or in real life, with whom they played. In that case, a dominant name emerged alongside other, less commonly adopted names, akin to regional variations for naming, say, milkshakes—while most people order a milkshake, Bostonians often ask for a frappe.

"Our findings demonstrate that social conventions can spontaneously evolve" without institutions guiding the process, Centola and Baronchelli write, though exactly how depends on the structure of the social network. In particular, randomly connecting many different players together led to one name being universally adopted. That, they argue, suggests that future research should usefully explore whether increasing social connectedness might unintentionally homogenize our behavior.

Related