Value Competition

Robert Wright's Next Book

The Evolution of God is very good as a critical but sympathetic guide to the Abrahamic religions and their scriptures.  I would like to see Wright's next book be something that the current one isn't, namely an argument for God as a plausible scientific hypothesis.  That hypothetical book would challenge those of us who are secular in a way that his current book and related NYT op-ed don't, as Daniel Dennett points out in today's letters.  

How might that work?   In the book and the op-ed, Wright refers to the possibility that a divine presence arranged the algorithms of natural selection, presumably including the "non-zero-sumness" of reciprocal cooperation that Wright rightly regards as a central feature of the way history and evolution work.  My skeptical response is, "How could it be any different?"  That is, if the algorithms of natural selection, including the logic of reciprocal cooperation, have to be what they are regardless of the universe you're in, I don't see a case based for God based on those algorithms.  But if I'm persuaded that non-zero-sumness is like physical constants that could be anything but have to be just so for matter to coagulate as it actually does, then an advocate for God as a plausible hypothesis has an "isn't it amazing that things are so neatly arranged morally as well as practically in our universe" argument that a good scientific Bayesian has to engage.

Personally, I'm a skeptic about the "isn't the moral order wondrous" argument; my prior is that the algorithms of natural selection and non-zero-sumness--and the humbler algorithms of value competition that I try to tease out--are universe-invariant like math, not universe-variant like physics.  But I'm less than confident about that prior, and would love to read a book by Wright that tries to persuade me that I'm wrong.

Live Long and Prosper?

Nicholas Wade's NYT article on the prospects for life-extending drugs made me wonder about a passing claim in the article that evolutionary biologists are skeptical on theoretical grounds about life extension.  My outsider's sense of natural selection leads me the other way; life extension seems like an area in which there could indeed be twenty-dollar bills on the ground.  A mutation that produced long life without a proportional increase in reproduction would be selected against, and a mutation that produced shorter life with more children per year would be selected for, all else equal.  If in the days of old there were indeed elves living for thousands of years or Methuselahs living for 969 years, a newly-rising race of shorter-lived men would outcompete the long-lived group, which is one way to interpret Tolkien, the Bible, and other myths of a lost golden age.   

Of course, it's possible that the individual boon of life extension would be a curse for society, with stagnant oldsters dominating everything and little new getting done.  I'm fairly optimistic on that score.  Excessive leverage seems to me a bigger problem with human civilization than insufficient leverage, and a society of longer-lived people seems like a better bet to deal with risk in a way that will sustain the society than a society of shorter-lived people.   If that's right, the results of biological evolution can be improved upon at the social as well as the individual level.

If biological evolution leaves plenty of twenty-dollar bills lying around for individuals and perfectly competitive economic markets leave none, I'd say value competition is in between.   Cultural watchmakers aren't as blind as Darwinian ones, but they aren't as sharp-eyed as Rolex.

Value Competition and the Dichotomy between the Good and the True

My "Building Better Agents" paper positions value competition as a scientific/scientistic movement; so do a lot of my posts on this blog.  On reflection, though, I wonder whether VC is not better seen as a potential form of self-assertion by normative disciplines like management, law, and applied psychology and ethics.  These disciplines suffer from a persistent sense of inferiority to scientific disciplines, based on the blurriness and arguable incoherence that attends their normative, do-gooding mission.  VC may be understood as a way to stand up for these disciplines--"incoherence is good!"--as well as an effort to give them a more scientifically plausible way to talk about their mission.

Falsification and Confirmation

For a human science to work well, it should be able to negate as well as to affirm the strongly held moral preconceptions of the scientist.   I strongly agree with Marvin Harris's criticism in Cultural Materialism: The Struggle for a Science of Culture of the Hegelian dialectic in its original form and in its modified forms in Marx and Levi-Strauss as failing under this criterion.  But it also seems to me that Harris's own "cultural materialism" fails under the same criterion.  The absence of an economic analysis of whether particular forms of social stratification are or are not conducive to social welfare allows an anthropologist using cultural materialism to see the world as an endless demonstration of exploitation in action if he is so inclined, just as an anthropologist using Marx's dialectical materialism can.  By comparison, economic analysis with its constant variations on the themes of market success and market failure usefully challenges strong believers in both regulation and markets, in both equality and inequality.   At the same time, economic analysis helpfully offers the possibility of confirmation to believers.   Economists with strong value commitments--such as Milton Friedman with his libertarianism or John Maynard Keynes with his statism--can do fine work that supports their predilections. 

One way to think about value competition is as a subset of economics that deals with collective decision, especially political decision backed by compulsion.  For all their very considerable value, public choice analyses of collective decision that assume rational, self-interested actors are subject to the criticism that they fail to take into account the ambivalence and complexity of human motivation.  By trying to do so, a value competition analysis has something to offer to those with more hopeful beliefs about collective action than appear warranted based on an assumption of rational self-interest.  Because a value competition analysis can also come up with the conclusion that collective decisions and agents vested with discretion will diverge from optimality, the method also has something to offer those skeptical about collective action.  Given that duality, the method has the ability to usefully discomfit both believers in politics and skeptics about politics.

A Master Theory of Human Nature?

There won't be one.  Why not evolution/evolutionary psychology?  It's the best for explaining how we came to be what we are as a species, but its reliance on natural selection, nicely as it works for biology, jibes poorly with the normative analysis of human culture and individual choice.   Why not economics?  It's the best for modeling the alignment of incentives, but its reliance on rationality, nicely as it works for some major aspects of human behavior, jibes poorly with a normative analysis of human nature.  Furthermore, economicswill not explain how we arose as a species in the way that evolution can.  Finally, why not a future science of value competition with tentacles in all of the social sciences?   It could be the best at modeling and describing irrationality and rationality together, but it will not do as well as economics does in explaining economic markets nor as well as evolution does in explaining how we came to be the creatures that we are. 

Why the US and Canadian Crime Decline in the 1990s?

I recently finished The Great American Crime Decline by criminologist Franklin Zimring.  It's a judicious discussion of possible reasons for the 1990s decline in crime that pays considerable attention to the famous Levitt/Donohue idea that the availability of abortion post Roe v. Wade led to lower crime rates years later as a generation of wanted children came of age.  Zimring is skeptical, partly because examination of pre and post Roe data shows increases rather than declines in categories, such as the proportion of babies born to unwed teenage mothers, that would logically be associated with unwanted children.  He's also skeptical because in Canada, which had a crime decline in the 1990s similar to the US (despite not increasing its incarceration or policing levels), there wasn't the US tilt toward crime rates going down especially fast among younger offenders that Levitt and Donohue use to support their claim.

I'm sympathetic to Zimring's conclusion that both purely demographic explanations (a lower proportion of young males) and rational choice explanations like incapacitation, general deterrence, policing quality and quantity, and Levitt and Donohue's leave a great deal of the 1990s crime decline unexplained.   I'm less than sure, though, about his claim that the 1990s decline occurred without a substantial change in the social fabric of American big cities, in which the decline was especially striking. 

My candidate, unmentioned by Zimring, for a significant demographic plus cultural explanation for 1990s US and Canadian crime declines, especially in certain big cities, is large-scale, ongoing immigration to both the US and Canada over the past several decades.  Contrary to a rational choice analysis, which would all else equal expect high immigration to create more crime among low-SES native-born residents of US and Canadian cities who compete with new immigrants for jobs, I think there is a value competition story to tell about how the arrival of waves of immigrants to the US from the 1970s on has enabled the growth over time of a working-class culture of pride and disdain among low-SES native-born Americans and Canadians ("we play by the rules instead of coming illegally; we know the language; we don't wait on the street for work; we're real Americans/Canadians" ) that was not possible when there was no large-scale new immigration to US and Canadian cities from Latin America, the Caribbean, Asia, and Africa.   To put it in VC-speak: It's worth developing and trying to test the hypothesis that egoistic values encouraging deferral of immediate gratification for long-term gain and altruistic values encouraging adherence to the rules of criminal law have gained some ground among low-SES, non-immigrant Americans and Canadians.  A poor young black man in New York, Chicago, or Compton, California whose family has been in the U.S. for centuries and who has come of age from around 1990 to the present has grown up in a group that lacks the clear bottom-of-the-totem-pole position that applied to his father's generation of poor black Americans, and that has plausibly made a positive difference in fostering working-class values over crime.  For new immigrants themselves, there is a very plausible rational choice story about a desire to avoid criminal problems that could result in deportation; for them, too, though, one could tell a value competition story about their having their own combination of pride and disdain relative to low-SES native-born Americans and Canadians ("we work hard; our families stay together; our women are honored") that is also conducive to the avoidance of criminal behavior.

To test the VC hypothesis just suggested, it would be interesting to have longitudinal attitudinal data, including ethnographies as well as surveys; an updated version of Talley's Corner, Elliot Liebow's 1968 study of lower class black men in Washington, D.C. would be extremely interesting.

One easy thing to do: It's interesting to take Table 1.1 in Zimring's first chapter, which shows the ranking of the fifteen biggest U.S. cities in reducing crime in the 1990s, and relate it to the proportion of foreign-born residents in the city in 2000, which I calculated from the census website:

1.  New York City  .36   2.  San Diego  .26   3.  San Francisco   .37   4.  Dallas  .24    5.  Los Angeles  .41    6.  Houston  .26   7.  San Antonio  .12   8.  San Jose .37    9.  Chicago  .22   10.  Indianapolis .05   11.  Phoenix .19    12.  Detroit  .05   13. Columbus .07   14.  Philadephia  .09   15. Baltimore  .05

Like most social science relationships, it's not an air-tight one, but it's pretty strong.  For each 10% increase in its foreign born population, a big city is expected to do 2.7 ranks better in its level of crime decline; the correlation between % foreign born and excelling in crime decline is .79, so over 60% of the variance is explained.  Obviously, a more thorough analysis would need to control for other factors (e.g., whether cities with higher foreign born populations also have higher economic growth, and what proportion of greater crime declines in high foreign born cities is associated with low crime among the new immigrants themselves), but the quick and dirty analysis suggests that the effect of immigration on crime by old-timers as well as newcomers is at the very least worth looking into further.  That project can be carried out with or without a value competition focus; a value competition emphasis on a possible growth of egoistic and altruistic values among low-SES native-born Americans and Canadians in response to an influx of new immigrants is one way, but of course not the only way, to interpret the data.

Creativity in Art, Technology, and Business

Jan Lourie, an artist who combines computer-rearranged, fractalized photos with other images into "metaprints" that sometimes resemble cubist renditions of buildings and sometimes work from other traditions, (in one print that I particularly liked, a Chinese water painting), gave a talk yesterday at Dana Library on creativity.  My RBS colleague Lee Papayanopoulos moderated, and recruited me and my colleagues Faroborz Damanpour and Farrokh Langdana to comment. 

 http://lp.rutgers.edu/metaprints

She focused on a 1937 conference featuring the French mathematicians Henri Poincare and Jacques Hadamard, in which they speculated on a subconscious process by which creativity arises to the surface, juxtaposing that conference with 1930s Freudians, a Herbert Simon paper, John Searle-like questions on whether creativity is involved in certain situations (for example, if someone finds a randomly-generated series of notes that corresponds to a work by Bach and recognizes it as excellent), and questions about the nature of the creativity involved in a business discovery like just-in-time inventory (could it involve an analogy to a surgeon's just-in-time team?). 

Lots of interesting things were said by all; invoking poster's prerogatives, I focus on my own reactions.  Since Jan's presentation fit very well into value competition themes, I was very happy to riff on how one could try to analyze the subconscious creative process alluded to by Hadamard and Poincare in terms of calculative and cultural centers of reason.   I claimed that that combination could be used to analyze what was going on in the J-I-T case (analogical, pattern-oriented reasoning plus calculation) and in Jan's own art (quasi-algorithmic logic ciombined with an overall pattern sense); I closed with an effort to draw in self-other themes a la Freud and their connection to creation.  (The thought I had in mind but didn't express involved the degree to which effective as opposed to ineffective creativity, for all of its association with individuality bursting free from social constraint, also involves a close alignment with certain social norms.)

My day-after question for myself: My remarks implicitly assumed that the same calc-culture, self-other framework that can be used to analyze optimal incentives in business and politics can also be used to analyze optimal incentives in art (and also, though I wasn't making that point, in science and technology).  I think that's true, but I also think that a lit crit analysis of how lit works or a physics analysis of how atoms work will wind up being quite different from a management analysis of how firms work.  The production of truth and beauty may well prove to be quite comparable to the production of the good, but there will also be more than a few differences that any decent future theory will have to have some good takes on.

 

Why No Science of Value Competition?

If it is viable to analyze value competition scientifically, an intriguing question is why such a science hasn't already been developed.   That is a very big and complicated question, to which I have three favored responses:

1) Such a science is in fact developing, but in diverse ways (evolutionary moral psychology, formal political theory, divided-self modeling in economics) that makes it difficult to recognize it as a single entity rather than as disparate research streams.   

2) There isn't a clear, powerful interest driving value competition forward.  Support for rising business enterprise helped drive the classical economists' analysis of markets, and religious skepticism helped drive the Darwinian analysis of evolution.  There isn't any comparably powerful viewpoint that has a stake in the development of a science of value competition.

3) The most obvious example of value competition in the social sphere (as opposed to within individuals), that between the left and the right, is widely troubling to people, even those who identify with one side or another, and empirically has less than a stellar track record given the failures of socialism on the left and fascism on the right.

Avoiding Just-So Stories

One pitfall that's easy to fall into: Thinking of VC explanations as to why an observed pattern of responses is adaptive without thinking at the same time of what other pattern(s) would be maladaptive.  A signal virtue of optimality modeling done right is that it clarifies the conditions under which optimal outcomes will not be attained as well as those under which they will be attained.

Norms Modeling

Can those who believe people often or generally act according to social norms that are not necessarily in their self-interest create models with the intricacy and real-world payoff of models that postulate people act according to calculations of their individual interest?   In principle, yes. Just as rational choice modeling generates interesting results on how actions based on individual interest align or fail to align with social welfare, norms modeling should be able to generate interesting results on how actions based on norms accord and fail to accord with individual welfare, and/or with social welfare if the model assumes norms reflect the interests of some but not of the whole. 

If norms modeling is a logical alternative to rational choice modeling, why isn't it as prevalent?   A path dependency explanation--that rational choice modeling got off the ground first in economics and preempted the modeling field--may be true, though it lacks intellectual bite.  A more interesting possibility is that norms modeling is harder to do well than rational choice modeling.  For one thing, the idea that the welfare of some indivduals is harmed by social norms that are good for the whole lacks the normative payoff of the rational choice idea that the pursuit of individual interest may be desirable or undesirable for the whole.  For another, the idea that social norms differ from the interests of the whole too readily becomes a normative argument distinct from the logic of any particular modeling of norms.  For another, unless norms modeling can assume individual variation in behavior as RC does, it is unlikely to generate interesting results.

One way to understand value competition is as an effort to overcome the problems of pure norms modeling.  In the spirit of norms modeling, VC assumes that people may be motivated by values or norms contrary to their personal or group interests and favorable to others' personal or group interests ("altruistic values").   Also in keeping with norms modeling, though less obviously so, VC assumes that people may be motivated by values or norms in keeping with their individual or group interests that lead them to overcome inaction ("egoistic values").   VC also assumes that these norms/values vary across people (and within people at different moments), rather than being fixed and uniform.  Finally, VC allows for the possibilities that people can act either according to their interests as modified (or reversed) by norms/values, or that they can act purely on a norms/values basis.  The VC claim or hope is that the norms-related assumptions, combined with allowing for the RC assumption in some cases, can produce intricate models with interesting results more readily than pure norms modeling, and can generate certain results inaccessible to pure RC modeling.

Given that VC allows for pure norms-based action with no individual interests or calculation in some cases, I would characterize VC in those cases as a subset of norms modeling, with variance in individuals' adherence to norms/values allowing for interesting modeling results not obtainable when individuals are uniform.  Given that in some other cases VC allows for people acting calculatively according to individual interests, one could say that in those cases VC modeling amounts to a complex subset of RC modeling, with individual interests that include internalized norms/values as well as standard preferences.  Given the well-established status of RC modeling and the nascent status of both norms modeling and VC modeling, that's an understandable position; I believe it's more accurate, though, to characterize VC as a fusion of RC and  norms modeling in such cases.