Swarm Marketing: More Tyranny of the Power Law?

A while ago I read a couple of articles on swarm intelligence, social network theory, and marketing. One of the articles was in The Economist, Swarming the shelves (Nov. 9, 2006), and the other was a Harvard Business Review piece co-written by Columbia mathematical sociologist Duncan J. Watts and McKinsey consultant Steve Hasker, Marketing in an Unpredictable World (Sep. 2006). I wanted to very briefly analyze these two articles because I think they: (a) are interesting bits of research on their own, and (b) they become even more fascinating in light of the long tail and a previous blog piece I wrote, Tyranny of the Power Law (and Why We Should Become Eclectic).

The Economist piece discusses a marketing strategy developed by two Florida Institute of Technology computer scientists for use in supermarkets and other retailers. Taking advantage of radio frequency identification tags that are increasingly getting embedded into consumer products, the researchers propose to use that technology, along with wi-fi screen/scanner included with every shopping cart, to inform customers about how many other consumers choose that particular product.

This scheme is basically a take off on an old marketing trick, 'bandwagon effect,' updated with the latest in social network analysis and mobile technology. The idea is that consumers will tend to 'herd' toward particular products because it was popular with other consumers; with up-to-date technology, consumers can now be more rapidly informed of what choices other consumers made. Think of the 'cool kids' in high school becoming more popular because everyone was informed they were popular.

The Economist piece also makes a passing reference to the research done at Columbia University that is written up in a popular format in the Harvard Business Review article. The research, published in Science as Experimental Study of Inequality and Unpredictability in an Artificial Cultural Market (311:854-856, Feb. 10, 2006), by Salganik, Dodds, and Watts involved a web-based experiment where the researchers created a market for downloadable music. They used this experimental market to see the influence of social influence on consumer choices and the predictability of a product's success. They found that "increasing the strength of social influence increased both inequality and unpredictability of success." In other words, increasing the influence that people have on each other -- e.g., by informing consumers of what products are popular -- will tend to make popular products more popular and widen the gap between them and less popular products as well as make it harder to predict beforehand which products will turn out to become popular.

All of this should sound familiar to those of us familiar with things like smart mobs, the business/pop-science book The Wisdom of Crowds, and the power law; the crowd tends to know better so we should go along with the crowd. This stands in contrast to the ideas put forward in the book The Long Tail and in my blog post Tyranny of the Power Law (and Why We Should Become Eclectic). In fact, Watts and Hasker directly challenge and contradict the long tail theory -- that technological innovations should lead to a wider and eclectic assortment of choices for consumers -- in their HBR piece.

As I made it known in the "Tyranny of the Power Law" blog post, I have serious misgivings about the type of world that Messrs Watts, Hasker, et al. envision. Yes, it's nice and helpful to know what other people are buying. As I made clear in my previous blog post, I don't disagree with the idea that crowds are sometimes wise. But are they always wise? And, as I tried to philosophically dissect, even if people tend to go along with the crowd and make popular things or people more popular, does that mean that such a mentality is an unalloyed good or is something we should encourage as a normative policy prescription?

Lemmings also follow the herd ... down ravines and to their doom. Crowds can be daft and mad and following them has often led to some of the most foolish and catastrophic incidents in history. As the Science piece points out, the network effects of social influence -- which can be fitted with a power law distribution -- can cause a serious disconnect between quality and ultimate success.

In that experiment, the 'best' (a quality which was determined by comparing the social influence conditions to the placebo/independent condition) songs, while doing well on average, did not always succeed as one might have expected and, in terms of market share, had the most unpredictable outcomes. In other words, at least for the music download market experiment, there was at least some substantive detachment between quality and success.

In real life, as opposed to artificial experiments, we can -- and have seen -- even greater disassociation between quality and success. The best products and people don't always get the success they deserve. That is, at least in part, due to the 'tyranny of the power law': Some products, services, people, and organizations get a leg up (sometimes, unfairly) and their pre-ordained success gets reinforced by social structures that are rigged to keep it that way ... meritocracy be damned. It's like high school, the cool kids weren't always the smartest or the most talented kids. The cool kids were cool because ... well, they were cool to begin with. That might be okay for teenagers, but is that any way we want to run a country, encourage culture, promote an educated populace, and run a business?