Overcoming Bias

Tropes Are Treasures

I've blogged before on theories of the functions of fiction in our lives, and celebrated this seminal analysis of the personality and motives of Victorian novel characters.  After browsing the TV Tropes website, it occurs to me that these tropes might be a great data source for studying fiction's functions. 

A possible research plan:

1. Identify tropes that describe common patterns of fiction which seem to deviate from patterns of reality.  Code these tropes by their degree of deviation, and by how confident we feel that this deviation is real.
2. Code these tropes according to a wide range of other possibly relevant parameters. 
3. Look for patterns among the tropes as so coded, and when possible check those patterns via formal statistical tests.
4. Compare theories of fiction's functions to these trope patterns, seeing which theories best account for the set of observed patterns.

Any student in search of a research project, take note! :)  HT to Doug.

Prefer Peace

As fiction authors know, compelling stories need conflict; readers love to root for good guys against bad guys.  As college professors know, students perk up when academic topics are posed as conflicts.  Sophomores love to hear each subject posed as a conflict between several possible isms, especially a long bitter conflict.  To them, intellectual maturity consists largely of looking over a long menu and ordering one from column A, one from column B, and so on.  But while I'd like to be a popular teacher, I'd rather be honest, and most subjects are just not well described as a conflict of isms. 

When asked to evaluate a proposed economic policy, most students identify some winners and losers, and then favor or oppose the policy based on which group they like best.  It takes a long time for students to learn to think in terms of economic efficiency, weighing the costs and benefits for all effected parties, and even then students usually find an even-handed approach much less inspiring.  Some econ profs engage students by inviting them to join the few knowing insiders against the ignorant multitudes outside, but even that rings wrong to me.

Yesterday I discussed the tension between the ideals we often verbalize and the goals our usual choices seem designed to achieve.  I tried to argue for compromise, for seeking "variations on common ideals which one can more easily admit serve ordinary non-ideal ends."  But, most commenters did not want compromise; they instead wanted to take sides and seek better ways for their side to win the war.  Generation after generation, the [added: some] old tell the young to seek internal peace; no internal side has the strength to win a clean victory, so all out war risks all out destruction.  But the young will not hear.

It seems that one of humanity's strongest ideals is actually war, i.e., uncompromising conflict.  In our culture we are supposed to oppose ordinary bloody war, preferring peace when possible there. But we do not generalize this lesson much to other sorts of  conflicts.  We celebrate those who take sides and win far more than we do peacemakers and compromisers.  But the principle is the same; every side can expect to get more of what it wants from compromise deals than from all out conflict.

Added: Byran Caplan asks:

What makes Robin think that "every side can expect to get more" from compromise than conflict?  Doesn't anyone have a comparative advantage in conflict?  And all it takes to get a conflict is one willing combatant, no?

Deals are not always enforceable, admitting interest in a deal might send the wrong signal, and one may need to threaten conflict to get the best deal.  Even so, there is some deal that beats each conflict for each party.

Meta Institutions

Institutions are stable social contexts which make and coordinate actions.  Examples include elections, agencies, courts, clubs, debates, peer review, malls, games, media, etc.  It is by now an economic truism that institutions matter a lot.  Good institutions can induce good choices and info sharing, while bad institutions do the opposite.  

Rather than advise particular choices, economists prefer to advise on general policies, that apply to many choices.  We prefer even more to advise on choice of institutions, since an institutional choice can influence a great many policies.  Following this meta line of reasoning, we should prefer even more to advise on meta-institutions, i.e., institutions that structure our choices of institutions. 

We allow most of our familiar institutions to at least influence our institutional choices.  But no doubt we use some more often in that role, and some are better suited to that role.  While I'm excited that decision markets can help advise organization decisions, I'm most excited about their potential as meta-institutions, advising us on key policy and institutional choices.  Of course we'll have to demonstrate their effectiveness more on small issues before folks will rely on them for big issues.

Some basic questions:

1. What institutions are especially good as meta-institutions?
2. What institutions should we use to evaluate meta-institutions?
3. What institutions are biased to prefer other institutions like themselves?
4. How often do different institutions agree on particular institutional choices?
5. What institutions can sensibly say if to rely on them as meta-institutions?

Efficient Economist's Pledge

Tuesday's debate with Bryan Caplan was great fun; thanks Bryan for being such a gentlemanly and compatible discussion partner!  I don't know when an official vid will be posted, but my amateur audio is here.  In before and after audience polls, Liberty vs. Efficiency got 42-10 before, and 25-20 after.  My argument and Bryan's many responses inspired me to compose this pledge:

The Efficient Economist's Pledge

I pledge to be an efficient economist, who helps clients find win-win deals to resolve social conflicts.

I need not accept all clients, but for the clients I do accept, my suggested deals should, if accepted straight or as a starting point for negotiation, get them more of what they want, relative to no deal.  I resist temptations to slant my advice to give hidden benefits to myself or my associates.  I accept my client's situation and wants as they are, and unless asked do not preach to them on what they should want.  When actions conflict with words, I mostly infer wants from actions, and avoid needlessly exposing contradictions.

For individual clients, I suggest actions that try to best get them what they want. When I advise groups of clients together, I seek win-win deals for them all together.  That is, I seek ways to suggest deals with a high chance of putting each client in a situation where, given their limited info, they should expect to benefit from the resulting deals. And I seek the best deals; it should not be easy to find other deals which all should expect to get them even more.  When trading client expected gains against the chance each expects to gain, I consider how many are needed to clinch a deal. 

On Liberty vs. Efficiency

To "win" a debate you aren't supposed to tip your opponent to the arguments you'll use.  But to promote a more productive conversation, that is exactly what you might do.  So in this post I'll lay out my basic (rather technical) argument for tonight's debate.  I've said:

The topic, as I see it, is the relative value/importance for economists of pushing "liberty," i.e., a policy of minimal government interference, and "efficiency," a standard policy evaluation metric that attempts to neutrally weigh policy consequences for different people.

Humans often find themselves in conflicts where they might make (and enforce) "deals" instead of "fighting" (or doing "nothing").  Such conflicts are often complex enough for many parties to be uncertain how they would fare, relative to fighting, under various possible deals.  In such situations, I see a noble and important role for expert arbiters who are "neutral," i.e., who develop deserved reputations for suggesting "win-win" deals where most or all parties should expect to benefit, relative to fighting.  Given access to such neutral expert advisers, conflicting parties can make better deals, to their mutual benefit. 

One reason I'm proud to be an economist is that we often fill this neutral expert arbiter role to varying degrees, and could do so even more if we tried.  And "efficiency," also known as "cost-benefit analysis," helps make this possible.  To estimate the efficiency of a deal, relative to a status quo, one adds up estimates for each person of the dollar value that person would place on this deal. 

College Prestige Lies

Over the next two weeks my eldest son will be rejected by some colleges, accepted by others.  And then we'll likely have to make a hard choice, between cheap state schools and expensive prestigious ones.  A colleague told me the best econ paper on this found it doesn't matter.  From its 1999 abstract:

We matched students who applied to, and were accepted by, similar colleges to try to eliminate this bias. Using the ... High School Class of 1972, we find that students who attended more selective colleges earned about the same [20 years later] as students of seemingly comparable ability who attended less selective schools. Children from low-income families, however, earned more if they attended selective colleges.

A 2006 NYT article confirms this:

Higher education experts have this message ... Pay less attention to prestige and more to “fit” — the marriage of interests and comfort level with factors like campus size, access to professors, instruction philosophy. ... A 1999 study by Alan B. Krueger ... and Stacy Dale ... found that students who were admitted to both selective and moderately selective colleges earned the same no matter which they attended.

as does a 2004 Atlantic Monthly article:

Lying Stimuli

In economic downturns, most economic activity arguably has a positive externality; when A buys services from B, not only do A and B benefit but the downturn is reduced a bit as well.  (In an upturn, extra activity arguably exasperates the boom-bust cycle.)  This fact can justify subsidizing economic activity a bit more (or taxing it a bit less) during a downturn.  But it turns out this isn't the main rationale for "economic stimuli" now being debated; those plans are largely based on the idea that people can be fooled because they are biased.  Tyler Cowen: 

Let's say government can spend $100 billion today or spend the present expected value of $100 billion, stretched out over time so it is a commitment in perpetuity.  Both spending programs are financed by bonds. ... The Keynesian boost to aggregate demand arises because people consider the resulting bonds to be "net wealth" even when they are not. ...  People are tricked by the government's fiscal policy, but of course the extent, timing, and nature of the trickery is hard to predict.  Is it easier to trick people "a lot all at once" or "a little bit by bit over time"?  It depends.  If you try to trick them slowly over time, temporal learning and adaptive expectations may work against the policymaker.  But if you try to trick people a lot all at once, the trick may rise over their threshold of attention, perhaps because of media coverage.

Wise taxpayers who get stimuli tax rebate checks should mostly save them, realizing that future taxes must rise to pay for those checks.  For similar reasons, wise taxpayers should also spend less upon hearing about government spending increases.  So with wise taxpayers it is not obvious that tax rebates or government spending increases would help much with the downturn. 

The consensus among macro-economists seems to be that people can in fact be fooled by such stimuli, but as Tyler indicates, it is not clear which policies most fool us.  In particular, the more public attention we give to the stimuli, the less they might work.  We might make people realize that they need to compensate via saving, and the more we scare folks into thinking we need huge stimuli, the more we might scare them away from normal economic activity levels.

So should we stop explaining macro-economics during this crisis, and stop saying how desperately we need stimuli?  After all, similar rationales were offered against allowing financial market short-sales.

Weak Social Theories

Much of armchair social science is pondering "puzzles," i.e., social behavior that doesn't fit well with simple theories, such as yesterday's cashless Christmas wish-list puzzle. A few standard theories can be easily invoked to explain most any social puzzle:

• Preferences - we just intrinsically want the specific outcomes produced by certain behavior; it is not instrumental toward other goals. E.g., folks pick their nose because it just feels good, not to clear out their noses. 
• Mistakes - typical mental heuristics and patterns of thought just consistently misjudge the situation. E.g., we buy mid octane gas because just we didn't think of mixing high and low octane gas.
• Norms - we do the usual thing so we won't look weird; norm violators tend to be odd and poorly informed about social conventions. E.g., you don't want to hire a guy who doesn't wear a suit to an interview.

The problem is that it is way too easy to explain most anything with such theories, at least if most possible preferences, mistakes, and norms seem nearly equally plausible. We can get more mileage out of specific and constrained theories of what are plausible preferences, mistakes, and norms. But no doubt many puzzles are in fact explained by odd preferences, mistakes, and norms. So what do we do?

It seems to me that the right approach is to seek theories that, with only a few free parameters, can explain a wide range of diverse puzzles. Trying to explain each puzzle with a new ad hoc story just doesn't get you very far; much better to lay out lots of related puzzles on the table and then ask how to explain many of them all at once. For example, instead of making up an ad hoc theory to explain US Christmas giving, collect "stylized facts" about holiday, wedding, birthday, and other gifts from across diverse cultures, and then look for theories which might explain a lot of that variation all at once.

Who Cheers The Referee?

Almost no one, that's who.  Oh folks may cheer a ruling favoring their side, but that is hardly the same.  On average referees mostly get complaints from all sides.   Who asks for their autograph, or wants to grow up to be one?

Similarly who cheers the officials who keep elections fair, or the teachers who grade fairly?  Inspiring stories are told of folks who win legal cases or music competitions, but what stories are told of fair neutral judges who make sure the right people win?  After all, competition stories are not nearly as inspiring with arbitrary or corrupt judges.  Oh judges are sometimes celebrated, but for supporting the "good" side, not for making a fair neutral evaluation.

Sure we give lip service to fairness, and we may sincerely believe that we care about it, but that mostly expresses itself as sincere outrage when our side is treated unfairly.  We usually can't be bothered to pay much attention to help settle disputes in which we have little stake.  So if you want to be celebrated and gain social support, take sides.  But if you want to instead do the most good for the world, consider pulling the rope sideways instead of joining the tug-o-war.  Consider being a neutral arbitrator, or better yet consider developing better systems of arbitration and evaluation.

Are AIs Homo Economicus?

Eliezer yesterday: 

If I had to pinpoint a single thing that strikes me as "disagree-able" about the way Robin frames his analyses, it's that there are a lot of opaque agents running around, little black boxes assumed to be similar to humans, but there are more of them and they're less expensive to build/teach/run.  ... The core of my argument has to do with what happens when you pry open the black boxes that are your economic agents, and start fiddling with their brain designs, and leave the tiny human dot in mind design space.

Lots of folks complain about economists; believers in peak oil, the gold standard, recycling, electric cars, rent control, minimum wages, tariffs, and bans on all sorts of things complain about contrary economic analyzes.  Since compared to most social scientists economists use relatively stark mathy models, the usual complaint is that our models neglect relevant factors, and make false assumptions. 

But of course we must neglect most everything, and make false assumptions, to have tractable models; the question in each context is what neglected factors and false assumptions would most mislead us. 

It is odd to hear complaints that economic models assume too much humanity; the usual complaint is the opposite.  Unless physicists have reasons to assume otherwise, they usually assume masses are at points, structures are rigid, surfaces are frictionless, and densities are uniform.  Similarly, unless economists have reasons to be more realistic in a context, they usually assume people are identical, risk-neutral, live forever, have selfish material stable desires, know everything, make no mental mistakes, and perfectly enforce every deal.  Products usually last one period or forever, are identical or infinitely varied, etc.