Diffusion: The problem of tracking ideas

Let’s start with the notion of a sign. A sign is something with a meaning and a form of expression, say, a word. Signs don’t have to be linguistic; they may be images, pieces of music, bodily ornamentations, or articles of clothing. In these latter cases, it may be difficult to articulate what is “meant” by the sign, but let us take meaning in a very broad sense, to include especially things like signaling membership in a particular group. Assuming the idea of a sign, I will draw a distinction between ideas and memes, although in the discussion below I will often have occasion to refer to shared properties. By an idea I mean something that can be meant by a sign, although, rightly or wrongly, I will speak of ideas as if they can be divorced from signs. By a meme I mean something that is very like a sign; that is, it has a form (a sequence of words, a particular image), and an associated content. For my purposes, the chief difference between a meme and a sign is that memes can be quite extended. A 300-word Facebook post can be a meme. That meme expresses one idea, although the idea may be complex. A very different Facebook post with very different words might be used to express the same idea. Thus, we would have two memes and one idea. A key premise of social science is that both ideas and memes have beginnings. Until very recently, this has been an assumption made out out of necessity, but observational confirmation has been lacking. More importantly, observation of the process by which social innovations unfold, spread, and take hold has been lacking. For example, the theory that the spread of a new idea might be epidemic-like, beginning with a long period of relatively slow development among a small community of early adopters, or even stasis, followed by a sudden period of explosive growth, had relatively few case studies for support. Now, with the web and social media we have the opportunity for massively documented studies tracking every stage in the establishment of an innovation. Thus, Yee et al. (2007) and Friedman et al. (2007) study the emergence of conventions governing avatars in the online community Second Life. Kooti et al. (2012) meticulously documents the spread of the retweeting convention of prefixing “RT” to a retweet. Because we have a complete record of all tweets since the dawn of Twitter, we know the identity of the first tweeter to use “RT” (@TDavid); we