Cohen and Stewart. The Collapse of Chaos. (1994)

      Jack Cohen and Ian Stewart. The Collapse of Chaos (1994) Cohen and Stewart attempt a meta-story here: that of how the chaotic, messy events on one level of reality (or perhaps merely analysis) produce regular and orderly features at a higher. An excellent book, often heavy going for anyone without at least a smattering of a variety of disciplines, but also often offering high spirits and sly irony.
I read it when it first appeared, but had forgotten almost all of it. Only a few marginal notes (typo-corrections, mostly) testify to my former reading.
     But I realise that many of its ideas have become commonplace for me. Chief of these are four. The first is that theories or models may or may not represent reality as it is. They are certainly work-alikes. That is, their observable external relations are the same as what they model, but there is no guarantee that their internal workings are the same. Nor is it ever possible to discover whether models are more than work-alikes, since attempts to get inside the black boxes merely produce more models with the same ontological deficiency.
     The second idea is that of emergent features: that it is impossible to predict, and often impossible in practice to explain, how the behaviour of one set of entities gives rise to features observable at a larger scale (or “higher level.”) Related to this is the idea that to explain how something happens is not the same as predicting what will happen. Science’s attempt to combine explicability and predictability, indeed most people’s belief that they are the same, has kept us from noting and investigating many things, or has misdirected our investigations. Ironically, it was just such a misdirected investigation (that of trying to derive a model of the weather from statistical data) that led to the discovery of chaotic systems, and prompted the development of chaos theory. Mandelbrot, also, testifies to this irony: according to Stewart, he said he had studied fractals a long time before he realised that he was looking at a new class of mathematical objects.
     The third idea is that the genome does not describe the organism, but merely the production the proteins that interact with each other and the environment to produce the organism. Understanding this puts a huge question mark over all genetic engineering. We simply cannot predict all the effects of transferring a gene from one organism to another. The fact that at present a very small minority of such transfers actually work to produce any result, let alone the desired one, shows that genetic engineering is still the crudest form of trial and error. But the genome-as-blueprint metaphor has great power, probably because of its simplicity, and because people do not understand blueprints, but think they do. Everyone has seen blueprints, for example in the weekly home-plans feature carried by many newspapers. The fact that such plans are really directions to the builder, and do not contain enough information to describe the final building, is lost on most people. That is why the metaphor misleads. People do not consider the blueprint as a recipe, which is really what it is. It might be better to make the metaphor explicit, and think of the genome as a program or recipe. A recipe for a cheese omelette does not describe the omelette, it describes how to make one. It takes ingredients and a cook and a stove to make an omelette. Just so, a genome does not describe an organism, it describes how to make one. It takes a zygote and a womb and an organism to make one.
     The description of the process of development is indirect, too, and consists mostly of instructions to make or stop making proteins. The proteins themselves react with each other and other chemicals, under the influence of temperature, pH, etc, and the result is a developing organism. What’s more, the proteins affect the genome’s functions: the products made under process A trigger instruction X, which stops process A and starts process B. B triggers instruction Y, which starts process C, which triggers instruction Z, which stops process B; and so, in all sorts of interlaced and intertwining instructions and processes.
     Finally, Stewart and Cohen have a healthy respect for the limits of scientific explanation. More than most popular science writers, they emphasise the fuzziness and tentativeness of science. This is a good thing, if only to remind us all that knowledge, even the most strongly supported, is never certain. If only religious folk understood this and accepted it, they might have more faith. **** (2002)
     Update 2013: It now appears that genetics is even more complicated than Cohen and Stewart knew. The environment (i.e, other cells, the chemical bath surrounding the cell, the organ of which it a part, the organism embedded the external environment, ...) turns genes on and off, which in turn affect the cells interaction with neighbouring cells, the chemical bath that surrounds it, and so on a wonderfully recursive dance. And just within the last year or so it's been discovered that genes can be transferred "horizontally" between species,probably via the microorganisms that inhabit it). See this National geographic article. The problem is that we don't have a language to describe the dynamic web of reactions that constitute an organism. In ordinary language, an organism is at best a gearbox. In fact it's something much more difficult to describe. we are thrown on the mercy of our metaphors. Her's one: an organism is shape created by its substrate, in the same that a fountain is a shape created by its substrate: water for the decorative fountain in your garden; plasma on the surface of the Sun.