Memes Arent Digital

MemesArentDigital. For an opposing view see MemesAreDigital.

For a start, memetics doesn't really bear resemblance to genetics. Genetics relies on the digital propagation of genes; when we identify a gene, we can taxonomically compare its form with other genes and make predictions about its effects. We can't do those things with memes.

RichardDawkins makes a pretty convincing case in RiverOutOfEden? that genes need to be digital in the 'low-level' sense - the underlying genetic code has a finite number of states, and that is what makes it possible for genes to make exact copies of themselves. Genes work for the same reason that digital recordings work - they make high-fidelity copying possible, analog encodings are copy-lossy and don't have that property.

[In the introduction to "The Meme Machine", Dawkins makes a pretty convincing case that memes don't need to be digital, just "high fidelity". He gives an example of built in error correction in the Chinese Junk Origami Making meme. He says that the meme is the instructions for folding a boat out of paper, not the boat itself. These instructions can be passed from person to person by demonstration even if the demonstration contains flaws.]

I think you overstate the digitality of genes. We often can't predict whether a mutation will leave an animal better off without knowing something about that animal's environment. A gene for a stronger heart may be a good mutation if you're a giraffe that has just evolved a longer neck and need a strong heart to pump blood to the head. Under other circumstances, it would waste resources. This dependency on environment (especially on other genes in that environment) I see as a point of similarity with memes, not a difference.

Genes are not discrete, sure. They're a cookbook, not a blueprint. But their means of replication and expression is entirely digital. It is tempting to think that a specific text is the digital representation of a meme; certainly texts are replicated digitally. But they are not expressed digitally; they are metabolized within the context of the reader's mindset, circumstances and presumptions.

PeterMerel objects to the "science" of memetics because they are not digital, and they are not predictive.

Not quite. I'm saying that minds are not digital replicators, where ribosomes are. I'm saying that therefore the analogy between genes and memes is fundamentally flawed. Memetics, to me, is equivalent to Lamarckianism.

Obviously - and I think that is a much more grievous blow against memetics than their not being digital. You have inheritance of acquired characteristics - an idea which is improved is passed on improved. In one sense, reproduction of memes is distantly similar to Darwinian mutation - a meme which is passed on by 'word of mouth' or other similarly unreliable channels can 'mutate' at that time, and a form of Darwinian selection might account for some evolution.

However, it is blindingly obvious that the main thrust of change in memes is not 'evolutionary' - or rather, it is 'evolutionary' in the proper sense of the term, which Darwin balked at applying to his theory because the term implied "directed progress". That is, the main manner in which memes change is in being "improved"; made more palatable to a larger audience and hence more likely to be spread.

isn't this Darwinian evolution at its best? Those memes that are most popular are most fit, and therefore survive and thrive. Those memes that are not popular, but can attach themselves to the popular ones can piggyback on that fitness. -- Pete Hardie

Thus we have a mechanism of 'evolutionary' change for our putative memes that is precisely the opposite of that which Darwinian evolution rests upon, and we know that genetics is the mechanism which underpins Darwinian evolution. Thus and IMHO, taking memetics seriously as analogous to genetics is a Grand Mistake.

'Memetics' is a lovely and expressive metaphor for the one and only similarity between ideas and genes - the fact that both genes and ideas may under certain circumstances contrive to "make copies of themselves" (in an extremely loose interpretation of that phrase).

-- LaurentBossavit (really AnonymousDonor, but I don't want the above opinions attributed to PeterMerel (who may or may not share them(, which could happen as I've taken some liberties in refactoring the discussion)))

The fundamental premise of Lamarckian evolution is that animals are trying to develop towards some sort of good. This is distinctly unlike memetics. The whole point of memes is that they exist solely because of their ability to self-replicate. This is very much like Darwinian evolution, and hence the analogy. But as memes are ideas and genes are chemicals, one shouldn't expect them to work in exactly the same way - and in particular, memes shouldn't be expected to copy themselves more precisely than ideas are defined.

Why are memes not quantifiable? It's pretty easy to write down an encoding of a meme in a digital format - what you are reading right now is a good example. Sure, the encoding, transmission and expression of memes is a pretty complicated business, but it is similarly complicated for genetics. There was a beautiful article in Scientific American a few years ago that described the mechanisms involved in the expression of a particular gene, and it's nowhere near as simple as just looking at a sequence of base pairs. For instance, the physical geometries of the DNA strand and expression protein played a considerable role (and by the way that geometry is analog, not digital).

This page, or any portion of it, isn't a meme. If it is can be likened to anything in genetic theories, it is to the phenotypic expression of a gene - in this view, texts are phenotypic expression of memes. Asserting that some texts can be digital isn't the same as asserting that all memes are digital.

Digital transmission isn't essential to genetics. I've often simply defined a meme as "a gene of culture" to people who know some genetics; they are then often able to come up with examples of memes, the idea of parasitic memes, etc. Clearly, the are getting some substance out of the analogy; you may regard this substance as insubstantial, but that's another argument. [Also, I can't help but point out that some people (JerryFodor? comes to mind) seem to believe that ideas really are stored digitially. I don't believe this, but I also don't believe that any current evidence rules it out, either.]
- If you can propose any theory of genetics that doesn't involve digital encoding of genes, and the consequent permutation of their associated alleles, I'd be amazed and delighted. As to substance, I'm not troubled that many people derive substance from memetics, the bible, the funny papers, and the racing guide, so long as they don't pretend to regard these entertainments as science. As to your burden of proof, I think I'll leave it where I found it ;-)
  - See my paragraphs in MemesShmemes regarding substance and burden of proof. With respect to digital encoding, I can easily imagine a mother passing on certain quantities of two chemicals to their children's eggs and then having the ratio of these chemicals both determine some aspect of their children's appearance and how much of the two chemicals to pass on to their children, say. I'm not saying that such a thing happpens, but I wouldn't be surprised if it did (prion folding comes close) and I think geneticists would regard its study as part of genetics.
- a good example of such a mechanism is the alligator, whose eggs will develop as female if they are at one temperature, and male if at another. Same genes, but different phenotype depending on environment. Kinda dents the "digital genes" idea

I would say that digital genes are just one possible specialization of a more general requirement: genetic operators must be non-linear. If your parents provide 2 blobs of genetic information, and they are simply mixed together linearly, then the result will always be an average of the two sources. Ater a few generations, all life would become uniform. In constrast, non-linear genetic operators allow interesting combinations. The great strength of digital encodings comes from their extreme non-linearity. Thus they are ideal for genetics. -- DaveWhipp

I think we should make the analogy between genes and memes more clear. Then the differences - and the similarities - become more clear:

phenotype - idea
chromosome - meme (complexity: sentenceor explanation)
gene - word
protein - phoneme/letter

I wonder about these:

allel (location of a gene) - location plays a significant role
polymerase

I think from this analogy it becomes clear that memes (in this sense) are digital in the same sense that chromosomes are: They can be transmitted (copied) with sufficient low error rate because their constituents (words/letters) can.

It is also clear that the ideas resulting from these memes in the processing human brain may differ significatly and are much less digital - like the phenotypes are.

But maybe you disagree with this setup of the analogy. Fell free to provide your own.

CategoryMemes