Lovely, but What Does it Mean?

Likewise, I’m a fan of Rob Cross’ approach to social network mapping — focusing on the assessed quality of relationships rather than just their quantity or frequency. And I think the most important value of social network maps lies in self-assessment — what a map of others’ perceptions of your social networks tells you about yourself and the nature and quality of your own relationships..

So I was intrigued to discover that one of the most popular social network maps on the web (reproduced above) looks utterly different from most of the maps I’ve seen before, which tend to be hub-and-spoke type drawings with a few overly-busy people with too many links and a few loners with too few. This particular map was developed by a professor in the US based on a survey of every student in a typical American high school, and depicts all their ‘sexual or romantic’ relationships over the past half-year.

Far from hub (the sexual athletes) and spoke (their conquests) with disconnected outliers (the school nerds et al), the picture is one of remarkable fidelity and many tiny, almost nuclear clusters. Sixty-three couples (126 students) had sex only with each other, and 63 other students were involved in closed triads — one or two partners with no overlap with anyone else in the school. Another 94 students were involved in other small clusters, and of them 79 had only one sexual partner in that cluster. Then there was a long, thin network connecting 288 students with up to 37 degrees of separation, the vast majority of whom also had only sexual partner. Most of the students in this giant cluster would have been flabbergasted to realize they were part of such a network, and (unless shown otherwise) would justifiably perceive their relationship map no differently from those in the nuclear clusters with two or three participants.

What does this mean? The link above is illuminating, but it is quite narrowly focused on preventing STDs (i.e. forget looking for hubs to teach/treat; a much broader approach is needed). I think this map raises more questions than it answers, and also has some important implications for the value of network maps in the first place:

• It suggests we don’t know much about networks. Networks, like all human and ecological systems are, after all, complex systems. By definition they can never be even close to fully known, nor can causality be inferred from them, nor reliable predictions be made from what we do know of them. We can look for patterns, but we can only hazard intelligent guesses about what those patterns might mean. 
• If we’re lucky, and reasonably perceptive in our guesses, we might be able to succeed in bringing about some modest changes in these patterns or the behaviour that gives rise to them through interventions: attractors (incentives) or barriers (disincentives). But that’s a huge ‘if’.
• Suppose we had done the survey again six months later. Would there still be one large and a hundred very small, isolated clusters? Two large clusters? None at all? Would the people in the large cluster(s) be very similar or utterly different from those who had been in the large cluster in the previous six months? If the survey was a year long, or three months long, instead of six, how different would the patterns be? If it was done at the high school a mile away, how different would the patterns be?

As intriguing as the map is, to me it poses the same huge risk as any other analysis of complex information: the risk that people will draw simple conclusions and propose complicated ‘solutions’ by misinterpreting or oversimplifying or placing far too much importance on this tiny, flawed, partial picture of a profoundly complex phenomenon, in this case the phenomenon of human networks and relationships.

It reminds me a bit of the old John Saxe poem about the six blind men trying to describe an elephant. We cannot hope to fathom human relationships, so we try to simplify them down in some way that will allow us to see the patterns and therefore come up with a course of action that, if imperfect, is better than doing nothing. The problem is that it isn’t necessarily better than doing nothing. AsJames Cascio said (perhaps quoting someone else — thanks to Martin Cleaver for the citation) the map is not the terrain.