10 February 2009

Group Selection, Social Systems, and the Common Weal

Ernst Mayr defined hard group selection as
"if, owing to the interaction of the composing individuals or owing to a division of labor or other social actions, the fitness of a group is higher or lower than the arithmetic mean of the fitness values of composing individuals, then the group as a whole can serve as an object of selection."[1]

In strictly biological contexts, this is almost always kin selection, because the social group involved is a kin group, a group of closely related organisms. In order for it to be actually distinctly group selection, there has to be a social group composed of relatively unrelated individuals.

Which becomes relevant in humans; we're specialized for co-operating in groups to an extent that it works well in non-kin groups. (Unlike anything else that's even close to being as good at co-operating in groups; ants and bees are mostly haploid siblings, after all.)

The second piece is that fitness values are, always, without exception, completely contextual; they're set by the environment and present need. Being a mighty mammoth hunter has no fitness value if what you need to do is set a bone or start a fire. (Being a top notch accountant has no fitness value if what you need is to change the washer in the kitchen tap; being a skilled plumber has no fitness value if what you need to do is file small business taxes, and so on.) This is a different way of putting the Adam Smith insight about the division of labour; getting really good at something and trading for your other needs is more productive than trying to know how to do everything. (Try imagining knowing how to do to a professional standard a tenth of the things listed in the yellow pages as services...)

So; humans co-operate in groups to increase their fitness values. Individual humans will generally co-operate in multiple such groups for different purposes if they have that option. (You don't want to have to rely on a single group to increase your fitness with respect to finance, health, and social standing if you can possibly help it, for instance, even if this was the traditional role of the extended family.)

The utility of the group being used as a co-operation mechanism can be evaluated by looking at that arithmetic mean of the individuals, and comparing that to the fitness results for the group, but this is not a useful evaluation. It produces a choice between the social group and functioning in isolation, and the group needs to be actively pathological before some group stops being better than no group.

It is much more useful to be able to compare the function of social groups in increasing individual fitness.

Unfortunately, that would require being able to compare the results of membership in different groups, which immediately gets into abstractions if we try to do it predictively. It should be possible to do it with a little more rigor retrospectively; if you belonged to this group or group of groups during that set of past events, how well would you do?

But before we can do that useful, we have to be able to define fitness, and wealth is not a good proxy for fitness.

This is because wealth is also entirely contextual (is that 1,000 1890 dollars or 1,000 1990 dollars? Are you in New York or Moose Factory or Tashkent? Do you need dinner or dental work?) and relating the two contexts introduces unnecessary difficulty; because wealth is not a good measure of social fitness, which necessarily involves getting copies of a social organization into the future; and because wealth does not reference the type of social organization that produces it, which does not matter in consideration of wealth but matters considerably in terms of fitness; land and trade and post-industrial means of production may produce the same degree of relative wealth for an individual, but they are not at all the same mechanism of co-operation.

Remember that we are defining fitness for social group selection; the core criteria is that the arithmetic mean of the fitness of the individuals belong to the group becomes greater, through group membership, than the arithmetic mean of their fitness considered as individuals; that in humans, a highly social species, near-completely dependent on social group selection mechanisms for survival, the default case is some social group, not no social group; that this means humans compete for position in their social group more than for membership in any social group; and that the social groups themselves compete for membership and for relative success.

Fitness has to consider four things:

  1. competition with other forms of social organization

    • efficiency in use of resources

    • withstanding direct attack

  2. how readily it permits members to increase their scope of choice

  3. how readily it permits members to have their scope of choice decrease

  4. stability of the system over generational time

More or less in order:

Other forms of social organization can use resources more efficiently; if the current group an individual belongs to turns x resources into 0.1 y increase in fitness, and some other group can turn x resources into 0.5 y increase in fitness, indiviudals will defect from their current group to the other group. (But not entirely, and not all at once, because individuals will be considering the other fitness criteria as well. This is why shifts from agrarian to industrial economies, or from from human handwork agrarian to animal traction agrarian economies, are generational and messy.)

Other forms of social organization can concentrate force better; agrarian societies displaced hunter-gather societies not by producing better living conditions for their inhabitants but by allowing an order of magnitude greater population. Whether or not it was a healthy, happy, population didn't matter; the hunter-gathers were militarily displaced. (A slower version of this happened with agrarian societies and nomadic herding cultures.)

So it does no good to produce a culture that's a lovely place to live in; it also has to be able to withstand cultures that don't care if they're lovely to live in, they want you to obey. (Even the elf-lords singing the woods have to be able to deal with Mordor's expansion somehow; the historical case is much more about dealing with high mobility barbarians.)

"Scope of choice" is being used as a convenient shorthand. Once basic survival needs are met—air, clothing, water, food, shelter, food storage...—specific priorities will shift due to individual preference and environmental factors (hardly anyone in Havana wants a parka; hardly anyone in Ottawa in January is much worried about insect repellent), the point is that individuals have a strong interest in increasing their perceived scope of choice, even if they do not necessarily understand how to go about achieving this. If you are a neolithic subsistence farmer, your scope of choice comes down to "hoe turnips or die"; it's really "be decently lucky about the behaviour of your environment and hoe turnips, or die" and as a culture we're still getting over the management strategies for being aware of the "luck" part.

On a scale of social organizations, increase in scope of choice—those things you as an individual can practically hope to achieve through work and planning, without a requirement of either exceptional luck or exceptional ability—happens through import replacement. (How and why this works are open questions; that it happens is an observation. Those who object to the idea of import replacement are reminded that you can falsify axioms but not events.)

On the scale of individuals, increase in the scope of choice is a function of your ability to accumulate a surplus. If you can't accumulate a surplus, all you can do is meet present need; you don't have an ability to plan for either environmental nor deliberate change. This set of implied constraints—the more surplus you have, the better you can cope with environmental surprises, and selection is all about coping with environmental surprises—drives competition between individuals within social groups as well as providing the motivation for individuals to join social groups.

So there's three kinds of selection pressure on systems of social organization that come out of scope of choice;

  1. how well can individuals accumulate surplus? (because if another system lets them accumulate more, individuals will defect to it)

  2. how well does the system structurally increase scope of choice? (industrial civilization has only happened once, despite there being a number of possibilities; increasing scope of choice isn't automatic.)

  3. how well does the system structurally manage parasitism by individuals seeking to divert structural increases in scope of choice to their individual surplus? (This is the basis of aristocratic systems that use peasant production to fund luxuries; it's also a serious problem in any complex society, because parasitism is obvious, easy, and tends to kill the host. Consider the Roman Senatorial classes' collective belief that the Roman state existed to keep them rich.)

Stability of the system over generational time is a requirement because selection functions on how well things get copies of themselves into the future. Having a great and good culture functioning happily does no good if it collapses as the founders die out. This has implications for education, class mobility, and generational transfer of power. (Aristocratic autocracies, for instance, generally have serious problems when anyone able enough to concentrate great power dies, because even successful automatic succession, by no means a given, does not replace the web of promises and knowledge that successful person represented. Consider Oliver Cromwell in England. Consider also any number of very long term American senators or multi-term Canadian prime ministers.)

Ok; so we have a requirement that social systems -- these things that are objects of selection because, and only because, they produce an arithmetic mean of fitness across their members that's higher than the arithmetic mean of the fitness of those members considered as individuals. Like any other entity subject to selection, chance environmental events matter; there's still the production of variation and the winnowing process of selection, as various social systems perform less well in the present environment than others, and diminish or die, being replaced by the more successful systems. Other social systems are of course part of the environment, as are individuals, some of whom will have new and clever ideas. Repeat indefinitely.

Given that model, one can reach some political conclusions.

  1. Aristocracies (functional or declared) are analogous to parasites; they're diverting the ability of the system to produce increased choice into a narrowly held set of individual surpluses. This really is bad because:

  2. All fitness is contextual; social systems have to function in an environment that includes predation (hostile social systems), parasitism (aristocracies), and random events (volcanoes, earthquake, drought, floods, locusts, wheat rust, epidemics....). In this context, it is almost impossible what positive traits will be required by easy to predict what negative traits are harmful. As such, diversity of positive traits is to be desired, which implies broadly distributed ability to accumulate surplus.

  3. Mechanisms that forbid your escape—switching to another social system—are an overt admission that this social system is being run for the benefit of someone who isn't you; you don't belong to it, you are being used by it.

  4. Fairness, perceived and actual, is important.

  5. If we can quantify fitness (or accept some proxy for it, which money isn't) we can quantify how well society is doing.

  6. Individual access to choice through accumulating surplus and system increase in access to choice through import replacement mechanisms are good candidates for fitness proxies

  7. The unknown future is a good reason to support diversity

  8. and a reason to plan; the social system has to get copies of itself into the future somehow. (Start thinking about opposition to public education in these terms; it's an attempt to ensure that most people have no ability to get their social system into the future with a level of complexity larger than a small village or largish church congregation.)

And there, I am going to stop, because this is already way long.

[1] Mayr, Ernst What Makes Biology Unique?, Cambridge University Press, 2004; the definition cited is on page 146 of the paperback edition.


jennie said...

(Aristocratic autocracies, for instance, generally have serious problems when anyone able enough to concentrate great power dies, because even successful automatic succession, by no means a given, does not replace the web of promises and knowledge that successful person represented. Consider Oliver Cromwell in England. Consider also any number of very long term American senators or multi-term Canadian prime ministers.)

Consider also some of the great early- and mid-19th-century corporations, which were more or less the personal fiefdoms of some industrial baron or other (Hershey, for example). Once their founders retired or died, they frequently tottered; many didn't survive into the third post-founder generation, because their extraordinary founders didn't adequately plan the long-term succession strategy, and a bunch of reasonably capable people couldn't replicate the web of knowledge and influence that the founder had amassed.

Graydon said...


That is also a good example of the kind of thing I'm talking about there.