Tuesday, January 3, 2017

Learning from animals

 On the TED Radio Hour about a year ago, Deborah Gordon talked about the self-organization of ants (and other biological phenomena) and what we can learn from that.

She contrasted the elegant, decentralized solutions found in biology to the clumsy, top-down structures that humans seem to design.

One of her interesting observations was a connection between ants determining how to forage and packets of information deciding how to send themselves out along the internet.

Ants in the desert have to balance the need to find food against the loss of water that comes from being out in the desert, away from their nest. According to Gordon, these ants only leave the nest if they encounter enough foragers returning with food, thus focusing their foraging their efforts on times when the likelihood of finding food makes it worth the loss of water.

On the internet, if you're a data packet, the thing you need isn't food, but bandwidth. And rather than have a central control telling you when to go, the decision is similarly decentralized: a data packet can't go out until it gets an acknowledgment from the router that the previous data packet had the bandwidth to go through.

As Gordon observes, "We invented for the internet a very similar algorithm to one that has evolved in desert ants many, many millions of years, and we just invented the internet yesterday."

"For situations where we don't have the right algorithm yet, we could look to see, how has nature solved that problem, and maybe we could use that solution."

That's an excellent idea, but she takes it too far. Not only are there ideas in nature we can learn from, she seems to cast hierarchies in general as undesirable.
The brain, like an ant colony and so many other systems in nature, has no central control, and yet we humans have set up so many of our systems in exactly the opposite way, from feudal system of kings and noblemen, to modern government, corporations, the military. Someone at the top is always in control.
But in a blade of grass there's no central control, in a human body there's no central control. Nobody's telling one of your blood cells, 'OK, you go over here and then when you get there you're going to meet this immune cell, and this is what you should do when that happens.'  
But if you think about the feudal system, it's not only that the king has this power. But it's also - the king constantly has to persuade everybody that he has this power, and everybody has to keep reinforcing the idea that he has this power. It doesn't just sit there. It has to constantly be maintained and reinforced and controlled and fought over. So it takes a lot of work to maintain a hierarchy.
Maybe in a way it's more effective and efficient to have a system without any central control where the whole thing can keep working without having to do all that extra effort of maintaining a hierarchy.
There are three things wrong with this generalized deprecation of hierarchical systems and the complaint that humans don't make enough use of decentralized systems with nobody in charge.

First, in praising the blind wisdom of evolution that produced the decentralized wonders of ant colonies, Gordon overlooks the idea that our human institutions are themselves products of an evolutionary process, and one with a certain amount of blindness as well.

Certainly there have been efforts at consciously crafting human institutions, such as the discussions that resulted in the U.S. Constitution, or the more thoroughgoing attempt by idealists in the French Revolution to throw out much of the old and replace it with new ways of organizing society that seemed rational to them.

But the final form of our constitution wasn't anybody's idea of a perfect plan; it was what could be agreed upon to get big states and small states alike to sign on.

And whether human institutions arise unintentionally or by a high degree of conscious choice, they are all subject to ruthless selection pressure. The Bolsheviks in Russia followed the French revolutionaries in trying to rebuild society from the ground up. They held onto power a lot longer, but in the end their system simply wasn't viable, and it died.

On top of that, systems that do survive tend to evolve. The original words of the U.S. Constitution haven't changed, but we have formally modified it through amendment, and we have modified it in practice by shifting our interpretation of it over time.

It's not just governments that change. All sorts of human institutions - from families to companies to churches - evolve over time, in response to new technologies, new ideas, and changes in other institutions.

Second, humans already have a social institution that embodies decentralized coordination, and we make pretty wide use of it. It's called "markets."

Gordon marvels at how nobody has to tell a blood cell, "OK, you go over here and then when you get there you're going to meet this immune cell, and this is what you should do when that happens." And it is marvelous. But it's exactly the same thing that happens in markets.

Nobody has to tell a carpenter, "Build a three-bedroom house on this piece of land, and then have that family over there live in it." Everybody follows the scent of money, and the result is some pretty impressive decentralized coordination of human activity.

Third, we have hierarchies (for some of our coordination tasks) because there are things that we want done that markets do poorly, or that markets actually work against.

Social Security is not merely a system for providing retirement income. It is one that is insulated from the vagaries of the stock market, and for 80 years it's been insulated from the vagaries of elections (though we may soon get to find out if that continues). It also embodies an element of redistribution. It recognizes that our earnings over our lives are some combination of merit and luck, and so it partially compensates the unlucky. If your job earned less money than mine did, you will get fewer dollars back from Social Security than I will, but if we compare how our Social Security checks compare to our earnings, you'll be getting a larger percentage of your earnings than I'll be getting of mine.

Social Security gets a lot of criticism, but I think it's a pretty brilliant solution to an important problem.

And it's only possible within the hierarchical framework of a government that can compel participation.

Others may prefer other examples, but there are plenty to be had.

Even that massive decentralized tool we have - the market - depends on and has co-evolved with the hierarchical system of the state. Markets work best when it's clear who owns what, and when there are relatively neutral, non-market forums for adjudicating disputes. Governments are the best tool we've found so far to provide those services.

One last pair of questions: If we do have to rely on hierarchical systems to address some of our problems, why haven't we evolved more natural, decentralized, automatic, robust solutions? Why can't we get to good results through behaviors with the elegant automaticity of ants in a colony, or cells in a body? Because look at what the ants are capable of:
There are more than 12,000 species of ants in every conceivable environment that operate without central control. Using only simple interactions, ant colonies have been performing amazing feats for more than 130 million years. We have a lot to learn from them.
I think the answer is that we face different problems than those encountered by ants or blood cells, and we face them on a different time scale.

Humans are a new sort of superorganism, engaged in activities as diverse as all the functions performed by the myriad species in an ecosystem, handling a range of challenges that dwarfs the problems facing any single species of ant or other animal.

On top of that, the environment changes faster for us than it does for ants. My grandmother was born in an impoverished Jewish farming community in the western provinces of the Russian Empire, a place devoid of electricity or running water, elementally tied to the seasons, spending Easter time fearing the appearance of a Cossack face at the window looking for vengeance against the Jews for Christ's death.

I sit thousands of miles away, in a comfortably heated house, able to eat strawberries in winter, typing out my thoughts on a computer small enough to sit on my lap, able to share them worldwide through the internet, about to go to my office to work on my courses for the spring semester. Here on the other side of the Atlantic and the other side of the Holocaust, I don't fear Cossacks at the window, at Easter or any other time. Even Nazis don't seem like a credible threat (though I'm a little less sure of that than I was a year ago).

And that's just two generations. Genetic evolution can't move fast enough to govern a set of behaviors that are useful in a Lithuanian farming village, and two generations later govern the different set of behaviors that are "fit" in American academia of the early 21st century.

In response to our unique situation - the diversity of our activities and the speed of change - we've moved beyond genetically driven behaviors to rely more on social institutions, and as described above, we've co-evolved a set of decentralized and hierarchical institutions to organize our work and make our species more successful.

In the 21st century we may be pushing our luck, but so far we've been up to the challenge.

1 comment:

  1. I love the comparisons. It is possible to think of a strongly centralized system in Nature (e.g. a harem of elephant seals dominated by a single bull who spends all of his time and energy defending it), but it is difficult to think of an example of a centralized, hierarchical system in Nature. One of the definitions of life is that it is autopoietic (self-maintaining and self-organizing).