The authors attempt a quantitative explanation of human population growth, resource depletion, and population decline on Easter Island from initial settlement around 400 AD to the arrival of slave-taking smallpox-infected outsiders around 1860. They recognized the need to test their hypotheses using comparisons with other Pacific islands. They developed a mathematical model of the interactions between humans and their environment that resembles a standard predator-prey model, with humans playing the role of "predators" and natural resources being the "prey." That is, they assumed that human population growth increases with the availability of natural resources (particularly forests, which include trees that can be used to make canoes to catch fish), whose ability to recover from harvest is mathematically similar to the population growth of a prey species.

This relatively simple model matched archaeological and historical estimates of population growth and decline on Easter Island reasonably well. The human population grew enough that resources were harvested faster than nature could replenish them, leading to faster degradation of the resource base and a population collapse. The same model, with appropriate changes in parameters, also apparently worked fairly well in correctly predicting sustainability on other islands, where populations leveled off rather than crashing. One major difference is that the wine palm that grows on Easter Island can take 60 years or more to produce fruit, whereas the coconut palms that grow elsewhere in Polynesia take as little as 7 years. This was modeled as faster "reproduction" of the resource base on other islands, but you could think of it as the difference between planting a tree for your own use versus planting a tree that your grandchildren might use, assuming that your family controls the land and manages to hold onto it long enough.

One might ask "what were they thinking when they cut down the last canoe tree?", but the authors point out that the destructive trends on Easter Island would not have been obvious. Population growth never exceeded 1% per year (less than the current world average) and loss of forest cover never exceeded 5% per human lifetime (also less than the current rate of deforestation). If carbon dioxide in the atmosphere were increasing 5% in a lifetime, rather than 50%, would we be worrying about it?

The paper has lots of interesting discussion (with citations) on other societies, from Mesopotamia to Rwanda, that suffered breakdowns linked partly to environmental degradation. Could we face similar collapses on a larger scale, today? The relationship between population growth and resource supply is more complicated today than assumed in their model. Global trade and migration might help us survive problems that only affect a small area, such as the 50-year drought that led to the collapse of Chaco Canyon. But are current world grain stocks, equal to only two months of consumption, sufficient reserves to buffer the possible effects of drought, war, or crop disease epidemics? The paper also discusses social factors that affect willingness to make the changes needed to prevent collapse. In particular, "institutional change is more likely to occur when the individuals that must make the change are confident that they will be among the beneficiaries." This principle seems to be broadly applicable, even to challenges like healthcare reform.

This paper would make a great centerpiece for a seminar class, with supplemental readings from later papers that have cited it, including those with alternative points of view.

The problem: even if average retirement savings were adequate, a few long-lived individuals would outlive their savings.

Insurance seems a logical solution, at first, because only a few people will live much longer than average. Insurance works well to spread current risks, such as home fires, where it’s easy to audit reserves, payouts, etc. But it’s hard to audit a promise that benefits will be available decades in the future, especially if governments are more concerned with the next election than with oversight. So pension programs (essentially old-age insurance) turn into pyramid schemes, which may fall apart as more people start collecting benefits. Government pension programs can keep making payments by raising taxes or by printing money. However, either “solution” can hurt the overall economy, e.g., by increasing inflation.

Mutual funds with stocks and bonds have the potential to keep ahead of inflation. But to supply enough income indefinitely, you need to invest a lot more money than if you were able to use it up during your lifetime. The problem, of course, is that you don’t know how long you’ll live.

The reason shares would only be distributed among those older than the deceased is to allow people of any age to join, without diluting the wealth of older members. Want to wait until you’re 90 to join? No problem. You’ll probably collect from some younger members, but your shares will probably be recycled soon. Undisclosed health problems? Welcome!

Although there may be subtle differences, "tragedy of the commons", "collective action problem", "prisoners' dilemma", "coordination problem", and the "free-rider problem" all seem similar enough that a solution to any of them would be at least a partial solution to all of them. It's a shame that researchers working on similar problems can't all agree to use the same terminology, to facilitate communication among disciplines. But the first brave researchers to switch to the common terminology would find their papers dropping into obscurity within their own disciplines, where decisions about grants and promotions are made...

I particularly liked "Evolutionary Explanations for Cooperation" by Stuart West, Ashleigh Griffin, and Andy Gardner. Their review reprints figures from several recent papers, so you can see some of the data upon which their generalizations are based. I won't try to summarize the whole thing, just some points that may have been neglected in other reviews of this topic.

Kin selection can favor altruism ("behaviour that is costly to the actor and beneficial to the recipient") towards close relatives. W.D. Hamilton predicted that a gene that leads to altruistic activity will spread when the cost (decrease in reproduction of the actor) is less than the benefit (increased reproduction) to the recipient times their relatedness.

West and coathors point out that research to test Hamilton's Rule has emphasized relatedness, but benefit and cost are also important. For example, species in which help raising young is most important to the survival of those young are more likely to help closer relatives (kin discrimination) as previously shown by Griffin and West (Science 302:634).

Indiscriminate helping may not be the result of kin selection, but rather some individual benefit to the helper, such as being allowed to remain in another's territory.

The authors reiterate an important point they have made previously, related to the effects of migration on kin selection. Animals that don't move around much may end up surrounded by relatives. This may be even more true of plants. This increased relatedness favors altruism, except that it may also increase the cost of altruism or reduce the benefits. If a bunch of relatives are all competing for the same resources, help that lets one sister reproduce more may come at the expense of the altruist's own reproduction (increasing cost) or that of another sister (reducing total benefits).

Cooperation between unrelated individuals, often of different species, can also be favored by enforcement mechanisms that tie individual benefit to cooperative behavior. For example, cleaner fish that bite their hosts get chased and other potential hosts avoid them. Humans (at least Swiss college students) tend to punish noncooperation in "experimental economics" games, even at some expense to themselves.

Enforcement mechanisms need not require conscious intelligence, however. The review mentions research by Toby Kiers in my lab showing that soybean plants punish rhizobium bacteria that fail to provide them with nitrogen. We assume that the soybean plants, in contrast to Swiss college students, obtain some individual benefit, such as saving scarce photosynthate, from doing so. Further work in this area was discussed last week.

The review points out that the mechanisms that prevent "cheaters" from undermining cooperation may be different from the mechanisms by which cooperation arose in the first place.

They make the important point that

we do not need to keep reinventing the wheel with more theoretical models that incorrectly claim to provide a new mechanism for the evolution of cooperation [12,97,98]. This has especially been a problem with models that examine limited dispersal or group structures [99–103] and which are, therefore, just reinventing kin selection.
Second, we do not need redefinitions of terms that already have specific and useful meanings.

If even one of a hypothesis's predictions turns out to be unambiguously wrong, the hypothesis must be discarded or revised. On the other hand, multiple correct predictions do not prove that a hypothesis is true -- there might be other hypotheses that make the same predictions. Either way, it's useful to consider several hypotheses, if you can. Tom Kinraide and I discussed these points in an article in American Biology Teacher in 2003. His boss wouldn't let him include his USDA affiliation, because someone at the lab complained that testing hypotheses would undermine his religion. I don't know; maybe it would.

In interpreting the data in this week's paper, we need to remember that conflicting hypotheses can sometimes make some of the same predictions, a point which is also reinforced in a recent review article. Both papers consider the benefits of biological diversity.

Mattila and Seeley impregnated honey bee queens with the same total amount of sperm from either one or 15 male bees. The latter led to more genetic diversity among worker bees, as in most wild colonies. More diverse colonies usually foraged more actively and stored 39% more food. They also produced more female worker bees and more male drones.

What hypotheses can explain these results? Bees apparently vary genetically in the amount of stimulus they require to start foraging or other tasks. The authors suggested that more diverse colonies might "respond to a broad[er] range of task-specific stimuli", but noted that the treatments differed even when food reserves were so low that "stimuli reflecting these needs could not have been greater."

An alternative hypothesis is that genetic diversity would affect within-colony cooperation. The authors suggested that any such effect should be in the wrong direction, because greater genetic diversity "erodes high levels of relatedness among female offspring, thereby hindering the evolution of altruistic behavior toward kin." (See discussion of Hamilton's Rule.)

This seems to contradict an earlier analysis by Ratnieks (American Naturalist 132:217), who wrote that:

As the number of males that mate with a queen increases, the relatedness (G) between a worker and the male offspring of the queen remains constant (G = 0.25), whereas the mean relatedness between workers and the male offspring of other workers declines from 0.375 to 0.125... Therefore, it is hypothesized that an allele, referred to as a "police allele," that causes workers to favor the production of queen-produced males over worker-produced males (i.e., to police other workers) should spread in eusocial hymenopteran populations in which queens mate with many males. Increased reproductive harmony, in which worker reproduction is reduced, is, therefore (and counterintuitively), a possible outcome of lowered relatedness between workers.

The review article focuses on the relationship between species diversity and the stability of ecosystems, but most of their insights would also apply to other ecosystem-level properties, including productivity. They point out that there are many ways to measure stability, and different measures can lead to opposite conclusions. Consistent with my comments above about hypotheses, they say:

it is not sufficient for theory to predict correctly whether the diversity-stability relationship is positive or negative; models could give the right prediction for the wrong reasons. Instead, theoretical models must be judged by their ability to capture the entire dynamics of the empirical system.

if the mechanisms underlying diversity-stability relationships are not identified, it is unclear whether
an observed diversity-stability relationship can be generalized to any other system.

For example, research by colleagues here at the University of Minnesota found that, although the number of plant species had no statistically significant effect on soil nitrate levels, nitrate was significantly lower when there were more distinct functional categories of plants (Science 277:1300). Can we generalize from these results to other ecosystems, possibly including agriculture? Ives and Carpenter might say that it would depend on the whether the mechanism linking functional diversity to lower nitrate applies to other ecosystems. For example, if less diversity meant that roots were inactive (not taking up nitrate) more months of the year, then you might expect a similar pattern in other ecosystems with a similar pattern. But what I noticed, in visiting the research plots, was that at least some of the single-species plots had lots of bare soil, not shaded by any plants. Other ecosystems dominated by a single species (redwood forests, say, or corn fields) often have fairly complete plant cover. Of course, some differences between ecosystems might be irrelevant to the diversity-nitrate relationship. But I agree with Ives and Carpenter: more rigorous testing of hypotheses that explicitly include mechanisms will lead to faster scientific progress.

He contrasts the Internet's current inability to make idiots shut up -- he phrases this more politely -- with three older institutions: markets, science, and democracy. "In markets, a really bad product eventually dies. In science that happens to the worst theories. In democracy, bad policies are supposed to fade away." Monopolies can force bad products on consumers, though, and bad policies -- he mentions segregation -- may not stay dead forever. Science mostly seems to move forward, but increasing specialization means those outside the relevant subdiscipline may cling to appealing but disproved ideas (e.g., ecosystem-as-organism) almost indefinitely.

I think my "social escrow" scheme (and related ideas such as Pledge Bank) could help to implement ideas requiring collective contributions (especially from groups larger than families but smaller than nations), but bad ideas might be implemented as efficiently as good ones. How could such groups (or larger ones) pick the best ideas?

Here are a couple of articles on the challenges of collective decision making that I thought were interesting and relevant. (The issue of what decisions should be made by some version of democracy and which are best left to individuals, markets, etc. is also important, of course.)

1) MacKenzie D. (2000) May the best man lose. Discovery November 2000: 85-91. There are several different ways for a group to choose from among more than two candidates (or options). They give different results and it's not clear whether any of them is best.

2) Friedman EJ, Resnick P. (2001) The social cost of cheap pseudonyms. Journal of Economics and Management Strategy 10: 173-199. There are a number of good reasons to allow anonymity in interactions among people, but letting people change pseudonyms encourages bad behavior. "We also discuss the use of free but unreplaceable pseudonyms,and describe a mechanism that implements them using standard encryption techniques, which could be practically implemented in electronic transactions."

Social escrow could change this. Rather than making a donation directly, you deposit the money with a social escrow agency. If enough other people do the same, thereby convincing your preferred candidate to run, great! Otherwise, you get your money back. People will pledge more than they would otherwise donate, because they know they won't be out any money unless their candidate has a realistic chance of winning. For example, 10 million people, each pledging $100, would raise $1 billion. Tough luck, special interests!

But social escrow could do more than just raise money for candidates. It could also influence their behavior once elected.

The simplest way to do this would be to put large sums of money (a few dollars each from a lot of people) in escrow, tied to the actual performance of people in office. Consider the congressman's dilemma: Big Tobacco is offering $1 million (campaign contribution, "consulting fees", whatever) to support their legislation, but there's $10 million in escrow ($10 each from a million tobacco victims) that will be released only if he votes no.

At some point, special interests might change their tune, and decide that making campaign contributions to candidate A dependent on candidate A's votes should be illegal after all. This would be an improvement over the present situation, and we should support it.

But it wouldn't kill social escrow. What if campaign contributions to one party were made conditional on the other party doing something bad? For example, 10 million people could each put $100 in escrow ($1 billion total) that would be released to the Democratic Party if any of the following happen:
1) the Supreme Court reverses Roe vs. Wade,
2) the US invades Iran,
3) polar bears go extinct in the wild, or
4) the budget deficit increases.

That's a lot of insurance for $100 each. If the social escrow company invested the money, they could even pay dividends to the "depositors." 4% interest less 2% management fee would yield 2%, better than most savings accounts.

The terms of the escrow agreement would need to include some disincentives to keep Democrats from confirming extremist judges, voting to invade Iran, etc., just to get the money!

Of course, this mechanism could be used by conservatives as well, releasing campaign contributions to the Republicans if:
1) the Supreme Court forces gay ministers to perform gay marriages,
2) Iran invades the US,
3) polar bears invade the US, or
4) the budget deficit increases (I'm assuming many conservatives still care about this, even if their leaders don't).

Mutual coercion seems to work to limit some kinds of antisocial behavior (e.g., violent crime) in many societies, but the only societies that coerce people into having fewer children (e.g., China), or more children, for that matter, -- Rumania in the 1980's is one of the more extreme examples -- can't claim that the coercion is mutually agreed upon. Furthermore, some kinds of antisocial behavior are so difficult to monitor that coercion might not be practical even if it were widely supported by the public.

Technical solutions to these sorts of problems, based on viruses that alter human behavior (e..g. a virus for altruism) or reduce human fertility have been proposed in science fiction.

Could this actually happen? This question really has two parts:

2) would a person or group with the technical capability to do such a thing be crazy enough to do it?

There are lots of real-world examples of parasites that manipulate their hosts. I recently ran into a nice summary on a site called Biology in Science Fiction. The rabies virus makes hosts more aggressive, so maybe a genetically engineered virus could make people less aggressive.

There has also been significant research on the development of viruses that control rodent pests by reducing their reproduction rather than killing them. In at least one case, however, a virus intended only to reduce mouse fertility by "immunocontraception" ended up killing the mice instead (Trends in Ecology and Evolution 16:418-420). Oops!

Our ability to manipulate genes is advancing rapidly, so things that are impossible today may soon be routine. Already, the 1918 flu virus, which killed millions, has been made from scratch based on the published DNA sequence (Nature 437: 794-795). So the deliberate development of viruses that would reduce fertility or alter behavior in humans, or perhaps in some subset of humans, may be a project that, within a decade or two, would seem like a feasible project for a single molecular biology graduate student, working in secret weekends and evenings.

Of course, the actual consequences of releasing such a virus would probably be very different than what was intended, as the mouse immunocontraception example shows. Most people with the technical ability to do such a thing would recognize this, but would everyone? Most would consider that major decisions about the future of humanity should be made democratically and with respect for individual choices, but would everyone?

I find it troubling that reviews of "Tide Turners", the story about releasing a virus to reduce human fertility, mostly seem to think it might be a good idea. One wrote, " it offers perhaps the only hope we have of staving off the more traditional methods nature has of dealing with overpopulation." I hope not!

I also asked about excessive lane-switching as a possible tragedy of the commons. He writes, "The actual benefits are marginal, both in the simulations and in tests performed on real roads. Due to a statistical effect, however, the PERCEIVED advantage may be substantial" and cites a paper in Nature (vol. 401, p.35) explaining why people incorrectly think the next lane is moving faster, even when it's not. If lane switchers are slowing everyone else down and not getting there any faster but feel like they're getting there faster, is that a tragedy of the commons?

He says his models don't support the claim (see last week) that a single motorist can erase a jam by increasing his following distance and driving at a constant speed. "First, a single motorist can create a jam (by creating a large perturbation of traffic flow), but once a jam is created, beneficial effects due to different driving behaviour grow proportional to the fraction of drivers applying the different driving style, so a single driver has a minute influence. Second, by increasing the following distance, the capacity is reduced (since the capacity is always lower than the inverse of the average time headway). So, driving with (overly) long following distances will lead to more stable traffic but also to more congestions."

OK, looks like I should leave this problem to the experts!

He's also got an impressive animation apparently showing that increasing following distance and letting people merge makes traffic flow faster when constricted by lane closures.

Is there a tragedy of the commons here? I always try to leave three or four seconds ahead of me because I think it increases my own safety and figure I'll only arrive at my destination one or two seconds later than if I left a two-second gap. (OK, two seconds times the number of people who cut in front of me, but that still never adds up to more than a 10 seconds or so.) If increasing following distance is the best strategy for individuals and also helps everyone else, then this isn't a real tragedy of the commons, just a widespread failure to recognize enlightened self-interest.

What I'd like to see is a realistic traffic simulation to answer these questions:
1) how much does aggressive driving decrease travel time for the perps? (On an open road, driving twice as fast gets you there in half the time, but what about in traffic?)
2) how much does it increase their chance of having an accident?
3) what are the effects on the rest of us?
4) assuming that most drivers follow too closely, what happens if a few drivers increase their following distance? How does it affect travel time and accident rates for these drivers, and how does it affect overall traffic flow?

There's another possible connection to Harding's original essay. Remember his suggestion that some problems have no technical solution? This might actually be a case in which a technical solution would work pretty well. I guess there would still be an element of "mutual coercion, mutually agreed upon", in that the solution may depend on government-mandated liability insurance.

“In fact, four men such as they were - four men devoted to one another, from their purses to their lives; four men always supporting one another, never yielding, executing singly or together the resolutions formed in common; four arms threatening the four cardinal points, or turning toward a single point - must inevitably, either subterraneously, in open day, by mining, in the trench, by cunning, or by force, open themselves a way toward the object they wished to attain, however well it might be defended, or however distant it may seem. The only thing that astonished D'Artagnan was that his friends had never thought of this.”
-- The Three Musketeers

I suggest that the key words in these contrasting views of cooperation are “ten million” and “four.” One person in ten million deserting won’t change the course of a war, but if one swordsman in a group of four consistently puts his own safety ahead of the success of the group, the group is more likely to lose a fight, with severe consequences for everyone in the group.

In 1974, Levin and Kilmer (Evolution vol. 28 p. 527-545) came to similar conclusions about the power of natural selection, acting on groups (a school of fish, say), to favor cooperation within groups. In order for group selection for cooperation to overcome individual selection for selfishness, they showed that group sizes “of less than 25 and usually closer to 10 were required”; their computer models also showed that “migration could not be too much greater than five percent per generation.” In other words, if even a small percentage of selfish individuals can switch to another group when their selfishness has undermined their current group, then differences in the success of groups have little effect on the evolution of individual traits that affect group success.

For example, suppose there are ten thousand people who really want the local public radio station to buy a new $500,000 transmitter. Each is willing, in principle, to pay the $50 apiece it would cost. But no one is willing to donate unless sure that enough others will also donate for the transmitter to actually get built. What to do?

Pledge Bank claims that 75% of people do follow through on financial pledges, but there’s no enforcement mechanism. I bet most would honor a $10 pledge, but might have second thoughts about a $50 pledge or a $500 pledge, depending on income. This would be especially true if someone could plausibly say either "they'll have enough money without my contribution" or "even if I honor my pledge, enough others won't that the project won't get completed."

A wide range of problems similar to this might be solved by what I call "social escrow", by analogy with escrow agencies that help with real-estate transactions.

A social escrow agency would accept contributions from a large number of people, keeping those contributions in trust until certain conditions are met. For example, the agency could accept checks designated for a particular purpose requiring a specified amount of money. The agency would keep the checks until the financial goal was reached (e.g., $500,000 for the new transmitter), then turn the checks over to the designated organization (e.g., the public radio station). If the goal isn’t reached by a specified date, the checks would be destroyed, leaving the money in the potential donors accounts. I suggest that this “money-back-unless-success" guarantee would make people more willing to donate.

A business called Fundable is offering to handle such transactions for a 7% fee. Maybe, if the idea catches on, competition among social escrow agencies will drive down fees (possibly to zero, if pledged money can be put in an interest-earning account) and spur innovation. For example, additional conditions could be imposed on the recipient -- Fundable just hands over the money to the group leader -- such as a public radio station reducing the amount of on-air fund-raising.

Alex Tabarrok has proposed something called a "dominant assurance contract" whereby people making pledges would receive a bonus if the pledge drive fails. In effect, people pledging would bet on failure of the pledge drive. But because this would be such a tempting bet (either you get a stronger radio signal or free money), lots of people would pledge, the drive would succeed, and the social escrow company wouldn't have to pay the bonuses.

I will discuss further variations on social escrow in a later post.

Here are what I see as Hardin's three main points:

1) some problems can't be solved through better technology, but require a change in human behavior;

2) freedom for individuals to pursue their own self-interest in ways that seem harmless, individually, can collectively result in conditions that hurt everyone -- if I graze two cows rather than one on the village commons, I'll have twice as much milk to sell, and one more cow won't hurt the grass, but if everyone grazes two cows, the grass will be so badly damaged all the cows will starve;

3) freedom for individuals to choose how many children to have is an example of #2 (population will grow enough to undermine human well-being) and #1 (better technology can't solve all the problems resulting from over-population).

I plan to discuss each of these points in this blog, probably (like most previous commentators) paying most attention to #2.