What Would Memetics Look Like?

I've seen memetics described in many places as an existing science, but I've never seen anything that I'd call a science. In fact, to my mind, the present state of affairs looks like introductory psychology minus the experimental backing … a sorry state indeed.

On the other hand, just because memetics doesn't exist now doesn't mean that it won't exist in the future. So, what I'd like to do here is speculate about what a science of memetics might look like.

First of all, memetics ought to incorporate the evolutionary aspects of genetics, of course … that's the whole point of the analogy between genes and memes.

Next, although I've argued that not all memes are viral, it's still true that there are some viral memes, so memetics ought to incorporate some elements of virology, or perhaps epidemiology. And, considering what bountiful sources of viral memes they are, I figure memetics ought to include elements from the fields of advertising and marketing as well. (Or, if you like, from the public relations side of politics.)

I imagine that if we put all the above together, we might have a reasonable handle on viral memes. For example, we might be able to produce a mathematical model of how such memes spread through populations via various channels of communication. For any given meme there would be a bunch of free parameters, but once we'd measured the parameters in one context, we'd be able to apply them in others … basically, a form of test marketing.

The main problem that remains, I think, is identifying non-viral memes. We could go study the fields above as much as we wanted, and we still wouldn't have the faintest idea what we were supposed to be applying them to. In other words, before we can talk about non-viral memes as spreading or evolving, we must be able to identify them as discrete, atomic entities.

The problem of identifying non-viral memes isn't addressed by any existing field that I know of, but I've had a few thoughts about it myself, which I hope may be of some value.

• Since I've said that one of the reasons I write these essays is to propagate good non-viral memes, one might hope there would be a few such scattered here and there, and indeed there are. Consider the idea that there's nothing new under the sun. Although the name (or, rather, proverb) is perhaps a bit viral, having been cast in memorable language, I'd say the meme itself is not.

  So how do I identify the memes I write about? I'm not sure. I suppose it's a habit of observation and self-observation I've gradually acquired … in other words, some memes. (E.g., the idea that I should write about those things my thoughts keep returning to.) I suppose I ought to be able to use these memes self-referentially to identify themselves, but I'm not going to attempt that here and now.

• If science fiction really is a literature of ideas, then one ought to be able to find and identify the ideas it contains. Sometimes the authors even identify the ideas themselves, as when Heinlein came up with the memorable name “tanstaafl” for the meme “there ain't no such thing as a free lunch”.

  Unfortunately, when I've actually tried this approach, I haven't had much luck, not even with Egan's books (say, Distress), which for some reason strike me as being full of separate, recognizable ideas. The trouble, I think, is that part of what makes fiction enjoyable is that it invokes multiple ideas at the same time, making it hard to figure out where one ends and the next one begins.

• Sometimes people will identify non-viral memes without knowing that that's what they're doing. To put it another way, all works contain memes; sometimes the memes are arranged so clearly and systematically that they're effectively identified. Think of Familiar Quotations, or of The Jargon File. I know two more very interesting examples, but I'm going to save them for discussion below.
• Finally, there's a nice method of identifying memes that I discovered more or less by accident, namely, to focus on a particular domain. The domains I've had the best luck with are driving and games, but the method ought to work with other domains as well. (I'm particularly fond of grid games; I have whole sets of notes on Minesweeper and Cross Sums that I haven't even begun to write up.)

  As an exercise, I wrote down all the memes I could think of that I developed while playing Oni. That ought to give you some idea of what I think non-viral memes look like.

I have to say, when I look at the memes I collected from Oni, they strike me as a huge mess of random little ideas. However, I don't think this is a mistake, or an accident—I think it is an accurate characterization of what any set of memes will look like. In fact, taking into account what a small domain I was looking at, it's probably fair to say that what I collected is rather a small mess compared to what any complete set of memes (a memome?) would look like.

Interestingly, the same characterization applies to genes, or so it seems to me. One gene might code for a protein that catalyzes the first step in a series of reactions, while another gene might code for the protein that catalyzes the next step, but the connection isn't apparent from looking at the genes or the proteins, and the same proteins are likely to perform other functions as well.

Comparing memes to enzymes or proteins suggests another field that might contribute to memetics, namely, molecular biology. In particular, one might hope to find ideas on how to present a set of memes in an organized way, or how to study the interaction of a fixed set of memes on short time scales.

Remember what I said above, that people will sometimes identify non-viral memes unintentionally? Well, while I was writing this essay, I was pleased and excited to discover not just a single work but an entire field in which such identification occurs: heuristics. Technically, heuristics is only concerned with a particular type of meme, memes that help discover new things, but since the things that are discovered also tend to be memes, the field tends to be pretty wide-ranging. I present here for your edification an example of an actual, quantifiable, non-viral meme, from Lenat's paper EURISKO: A Program That Learns New Heuristics and Domain Concepts.

“if trying to find extreme examples of C, then extract the base step from a recursive definition of C as one such example”

By the way, the domains referred to in the title of the paper are exactly the same kind of domains as the ones I was talking about, above … for example, the design of three-dimensional VLSI devices.

In the end, Eurisko was found to be limited by its lack of common-sense knowledge; its successor, Cyc, was an attempt to collect that knowledge, and so might be thought of as a kind of human memome project.

That completes almost everything I wanted to say. All that's left is two more examples of fields that may be relevant to memetics.

First, consider the “extreme examples” meme I quoted above. I'd have a hard time acquiring the meme just by reading it, but if I imagine using it, or, better, see someone else actually use it, then it will probably stick with me—a benefit of apprenticeship, and of programming in pairs. Of course, others may find it easier to acquire the meme by reading it, or by listening to someone explain it, or who knows what. That idea, that different people have different learning styles, is an educational truism, and that brings me to my point: the field of education is concerned with the effective transfer of memes, and so is relevant to memetics.

Second, the site I just referred to, the Portland Pattern Repository, serves not only as an excellent source of identified memes, but also as an example of another whole field in which memes are identified, the field (?) of design patterns. The canonical reference, on design patterns in architecture, is A Pattern Language: Towns, Buildings, Construction; another standard reference, on design patterns in software, is Design Patterns.