Schneier on Security

Martin Budden • February 26, 2011 6:09 AM

I am always wary when I hear an assertion that a particular characteristic is unique to humans. Partly because I think it is arrogance, but mainly because it just does not make evolutionary sense. There are many assertions of unique human qualities, without really considering what this means in evolutionary terms: if these assertions are taken as true it implies that, in the last million years or so, a large number of characteristics have simultaneously evolved in a single species. I just don’t think this is very likely. I think it is a much better null hypothesis that incremental improvements in a number of areas have evolved in humans, and, because of the non-linear nature of ecosystems, these have given humans a disproportionate evolutionary advantage. I accept that humans may have evolved unique characteristics, but believe that such claims should be challenged until there is good evidence.

So back to your claim:

“Humans have a natural propensity to trust non-kin, even strangers … except for a few simplistic counterexamples, it’s unique among life on this planet.”

I disagree. I think trust of non-kin is extremely common.

Trust involves three things: an implicit or explicit deal or bargain, a belief in the competence of the other party, and a belief that the other party won’t cheat. In the absence a communication mechanism or common language (as is the case in most of the natural world) the bargain is, of course, implicit. A deal can be made without trust but it means both parties to the deal have to spend significant amounts of energy checking up on each other and enforcing the terms of the deal. Bringing trust into the equation means that both parties can spend less energy on checking and therefore more energy on more productive things. Continuous checking that the bargain is being upheld is replaced by taking some form of punitive action when the bargain is breached. The punitive action may be as simple as not dealing with the incompetent or cheating party again, or it may involve inflicting some form of harm on the party that breached the agreement.

Note also that trusting someone does not mean that you believe that they do not cheat, it just means you believe that they do not cheat above a level implicit in the deal. Some cheating is tolerated, since a deal with some cheating can be more beneficial to both parties than no deal at all.

If I had to boil down the essential manifestation of trust, I would say it is this: the replacement of continuous verification by intermittent checking and a punishment/reward mechanism.

So let’s get back to biology.

I’ll begin with a simple statement: symbiotic relationships are extremely common in biology. By my definition of trust, symbiosis is a trusting relationship between organism that are not members of the same species. Of course parasitic relationships are also extremely common in biology, but the existence of parasitic relationships does not deny the existence of symbiotic ones. If non-species trust is common, then non-kin trust within a species is also likely to be common.

I imagine that there are two scenarios where trust relationships are particularly prone to evolve:
i) groups of prey cooperating in predation defence
ii) groups of predators cooperating in the hunt

Whenever a group of animals larger than a kinship group forms, there is some kind of implicit deal made. It’s almost inevitable that trust evolves, since if the deal is not upheld a member of the group can sanction the group simply by leaving.

Here’s an example of monkeys cooperating (sharing food supply) because of predation defence benefits (http://beheco.oxfordjournals.org/content/15/3/400.full). A single monkey could cheat – imagine it saw a bunch of fruit, but another monkey was closer – the cheating monkey could give the danger call and then get the fruit for itself. I imagine detection and sanction mechanisms have evolved to keep this behaviour to a tolerable level.

Although cooperation and trust within a group of prey is interesting, my view is that the cooperation and trust mechanisms within groups of predators are more complex and more interesting. There is just so much more going on. Assessment of competence is more important (is that young pup ready to join the hunt? is that injured group-member too much of a burden?). There is the issue of sharing out the kill – note that the criteria for fairness is not “are the shares equal”, but is “are the shares what was implicitly agreed before the hunt” – everyone expects the “top dog” to get more than an equal share, but even the “bottom dog” expects a small share. Predator groups are often large than familial units, so there must be non-kin trust within these groups.

As well and predator and prey groups, trust can be beneficial to a group with a common goal. An example that is particularly interesting to me is that of a flock of geese in a V-shaped skein. It’s particularly interesting to me because I am a cyclist, and cyclists often ride in “pacelines” where, like a skein, the effort of being on the front is shared. A paceline makes an interesting societal security case study. Pacelines form because it is mutually advantageous to all the cyclists involved – the energy expenditure in a paceline is (for riders not on the front) is about 30% lower than the energy expenditure of a rider on their own. Pacelines assume trust – both in competence and in willingness to do a fair share of time at the front. Mechanisms exist for ejecting riders who are incompetent (and/or dangerous) or who don’t pull their fair share. Mechanisms also exist for protecting club members or team members who are struggling with pace. It would be an interesting study to see if similar mechanisms exist in flocks of geese.

In your post you say “Security isn’t just a tax on the honest; it’s a very expensive tax on the honest”, ” Security — especially societal security — used to be cheap” and “we routinely trust non-kin (an evolutionary puzzle, to be sure)”. You first two statements answer the evolutionary puzzle – trust of non-kin evolved because societal security is cheaper than individual security. There is no evolutionary puzzle: trust (by which I mean the replacement of continuous monitoring by detect and punish) is the cheapest way to improve security. An animal that trusts others, even if it is occasionally cheated, is likely to be more successful than an animal that does not trust others. (Of course that trust cannot be simple naive trust – the animal also needs to be able to evaluate if another animal is trustworthy and if it is being cheated.)

Which leads we to a final observation about societal security. You say it used to be cheap, implying it is getting more expensive. Could this be because society is moving away from trust based security (detect and punish) to prevention based security (where we try and prevent wrongdoing by, for example, continual monitoring and surveillance)?