Since time began man has been searching for that perfect “someone”; mate, protector (alpha male), friend, ally, partner, employee, customer, retailer, vendor, expert, advisor, supporter, etc. Historically the pool from which we chose was small and geographically, socially, and culturally proximal. Our choices relied on face-to-face interaction or direct word-of-mouth. I come from the West of Ireland where the bicycle made the single greatest contribution to the gene pool :-)
Today things are very different! The pool has grown to an incomprehensible size and we can no longer rely on personal relationships. We need smarter ways of narrowing the pool and identifying that “perfect person” for whatever the purpose.
Enter “influence analysis”; the most useless, misleading, distracting, and possibly destructive term ever invented. OK, that may be a bit of an exaggeration, but you get my gist :-)
Social network analysis (SNA) is unquestionably hugely valuable, but the obsession with dumbing it down to a single number is absolutely ridiculous. It’s like saying that there is one single verb that can be used to describe every person on the planet.
So what problem are we really tring to solve here? At the most basic level we are doing social segmentation which in most cases is multidimensional. For example; I care about their network connectivity, the topics that generate the greatest response from their community (not that they most frequently talk about), the type of people that make up this community (ie. their classification; wealth, geography, interests, gender, etc.) and their associated network and interaction patterns.
We need to stop over-simplifying the problem, which doesn’t mean we need a complex end user solution. We need to leverage our entire social analytics portfolios, where social analytics to me simply means overlaying social networks (derived from interaction patterns) on top of traditional content analytics (classification, sentiment or topic detection, term extraction, …).
So what I really want is a Person Recommendation System, to which I can issue complex queries, and also provide some simple query constructs for the layperson, such as “find people who know a lot about phones and that people seem to trust”, “find me people like Tom”, or “find me women who know their sports”.
This obsession with influence scores has to stop before we all become self-obsessed, egocentric, and paranoid bores!
So my question to you is: Anyone got a “catchier” way of describing a person’s social classification? Pleassse… let’s kill Influence Numbers :-)