Saturday, February 24, 2007


So I'm at home with a cold with the usual symptoms which need not be described here. But it got me thinking of the analogy of viruses to music. In a way music can be easily compared to viruses - people can get "infected" if they really like a song, while some are "immune" if they just don't like it. The question though, which everyone would like to answer, is what makes certain viruses spread faster than others? And when I say viruses, certainly we can generalize to music, youTube videos, the next big thing, stocks, etc.

(One quick aside: there's a lot of existing theory (as mentioned in Fooled by Randomness and Origin of Wealth) which states that we're all heavily biased by hindsight. In the case of business, we tend to see companies which perform well and associate success factors to them, in the hopes we might emulate those factors and equally succeed in some fashion. Built to Last, In Search of Excellence, and a whole slough of business texts are based on this premise.

Sounds reasonable, but it's essentially wrong. Because unless you analyze the losers as much as you analyze the winners, there's no way to really suppose those "winner traits" are wholly unique or effective. In most cases, it may simply be luck - after all, you can theoretically flip 5 heads in a row, but flipping 5 heads may have nothing to do with your ability as a coinflipper. And because losers, at least in business, are quietly wiped away by the tides of bankruptcy/acquisition, we have no way of separating both groups. We do have "some" evidence to suggest these companies were just as much lucky as they were capable, given the fall in prominence of several of the companies mentioned in all those texts.)

I mention the parenthetical as often times we ascribe the viral growth of certain phenomena (Harry Potter, American Idol, fads in general) to certain factors, but modern experiments in network theory suggest attributes may be wholly irrelevent. In models mentioned in Six Degrees, the quality of the virus had little impact on the growth of the virus - meaning it was just as likely to fail than not. Rather, it's the ability of the virus to reach a subgroup of people (called a percolating cluster) and to completely saturate that group. The next step is for some members of that percolating cluster to have connections outside the immediate group, and which can then allow transmission at a far higher rate.

What's the big conclusion of all this studying? Instead of analyzing features, product, functionality, etc. of certain viral phenomena, we have to look at the consumers who use it. Are they fairly insular (e.g. I imagine a social networking site for refrigerator repair men might be grow beyond that immediate audience)? Are they well-connected? What do they use my product/service/idea for beyond what I (as the business owner) intended?

MySpace may be a good example of this - initially starting with a focus on bands, it soon became the "defacto" social network. And in retrospect it may make sense because bands and their fans are a diverse lot, given there may not be deep similarities between you and another fan of the same band. Hence the ability to reach several distinct groups quite readily.

As a counter example which did well but did not reach huge levels of usage, Paypal initially focused on auction owners (which in itself was simply a random event, where marketing individuals found initial instances of the product being used on auction sites), who in turn were growing thanks to the growth of auction sites (primarily Ebay). Now Paypal has differentiated by offering other financial products, though their primary service is still auctioneers. (For more info on Paypal, check out this quick read: Paypal Wars.)

No comments: