6 thoughts on “Phillip W. Magness on the Leonard book review”

1. “What troubles me more about the JEL piece though is its coupling of an aggressive attack on Leonard’s scholarship with very little substance to justify its derision.”

   Based on my own reading, I’d say this perfectly summarizes Marshall Steinbaum’s style.

2. “I wonder how the editorial decision of the Journal of Economic Literature can be explained.”

   I think the term cognitive dissonance is an explanation.

   “Progressives” see themselves as noble souls standing up to “white supremacy.” When confronted with solid evidence that progressives just 100 years ago were racist, segregationists, advocated eugenics, and had a Darwinian view of racial competition, they don’t know what to do except attack the person presenting the evidence. And on today’s campuses, “attack” is not a metaphor.

   • This made me think of all the people (so the polls say) who don’t want to get rid of statues of Confederates. You want to feel that you have a heritage that you can feel part of and be proud of–at least somewhat. You don’t want to feel that your ancestors were jerks. And you especially don’t want to hear that from “outsiders.”

     I wonder if the reception would have been different if the theme of the book had been “bad right wingers forced good left wingers into bad ideas (racism, eugenics, etc.)” instead of “bad ideas (racism, eugenics, etc.) were an integral part of turn of the century Progressivism.”

3. Time is like money, it’s scarce and has to be allocated among countless choices. One can’t read everything, so one relies on reviews to either help one decide whether to read a book or article based on the reported content and quality, or in the alternative at least to provide a fair summary of those features to accompany any praise or criticism. And deciding what to believe about some piece of writing is a lot like deciding what to believe about any market product. If the systems by which we arrive at those beliefs become untrustworthy, then entire institutions can collapse down to a worse, low-trust equilibrium.

   So unfair reviews like this relate to a much larger issue and provide yet another fundamental difficulty with trying to implement Libertarian ideas in the modern era. I call the general problem, “Reputation Assurance”. (I suppose one could call it “meta-reputation assurance”, since we are talking about the reputation of the reputation systems themselves, and borrowing the term ‘assurance’ from the field of information security is no coincidence.)

   Human social psychology is hard-wired to respond strongly to environmental cues that inform inferences about what everyone else in their social group is doing, thinking, enjoying, and so forth. Modern and sophisticated forms of mass influence (to include marketing and political propaganda) in the era of broadcasting (and now social media) are engineered specifically to exploit this tendency, and the manipulation of reputation by those with control over reputation systems is a particularly powerful way to shape social outcomes, that is, it is a source of a huge amount of political power. If we should be wary of, and on guard against, the state’s powers, then we should probably be equally wary of that power of influence being concentrated into the hands of just a few entities.

   A typical progressive semi-rhetorical question to libertarians is, “Without government agents investigating to ensure quality and compliance, how could ordinary consumers be confident that they will receive suitable goods and services, especially in small, infrequent, one-off exchanges with strangers where even libertarians will concede that the tort system is not adequate to provide the appropriate incentives? Government is what enables a high-trust market equilibrium in which, if people are not actually entitled to be completely ‘carefree’, they can lower their guard enough to the point where it enables all kind of market transactions with a kind of background confidence that they can rely on ‘the system’ to make sure everything has a minimum level of safety and that advertised claims are accurate. A caveat emptor world would create a low-trust equilibrium and raise everyone’s (wasteful) security costs to the point that most potential gains from trade are not realized, because if merchants can’t rely on consumer trust, they can’t charge higher prices for honesty and quality, and so have an incentive to provide junk, and so low consumer expectations become self-validating. Only something like the state can break this vicious cycle to everyone’s net benefit.”

   This is similar to Akerlof’s “quality uncertainty / asymmetric information” argument in his infamous Market for Lemons article. Rod Dreher recently wrote about his father’s role as a food safety health inspector at the 1971 “Celebration of Life” festival in which he had to stand up against a near-riot to prevent a bunch of young people from eating spoiled chicken. They weren’t happy about “The Man” stopping them from making that choice. Another example could be those parents who were upset when the magnetic buckyballs were banned but not thinking about the risk of the balls pinching and tearing intestines in ways requiring surgery if a child were to swallow them, and as any parent known, kids swallow everything. (They are back on the market now, though with more adequate warnings about this particular risk.)

   Anyway, a typical libertarian response, especially in the age of the internet, has been, “Reputation!” There has always been word of mouth, and yes the effectiveness of that system was being undermined by steadily increasing social atomization, but by the compensatory saving grace of technological development we now we have the internet and everybody can leave comments online for everyone else to read, something everyone has gotten used to doing before making decisions, because now there’s a Yelp-equivalent for just about everything (though I wonder what Libertarian faculty think of their own RateMyProf comments?) The theory goes that under such conditions every company has an incentive to avoid dissatisfied customers who would leave critical comments that would discourage future customers.

   But how do we know whether those reviews are reliable?

   That’s exactly the problem, we don’t. And increasingly, it’s starting to become clear that we shouldn’t.

   There are five parts to the problem:

   1. First not everyone leaves reviews, since they are voluntary there is a selection effect for idiosyncratic motivations, and so there is no guarantee that those that do leave comments are representative of consumers at large.

   2. Second, even if the reputation system is completely “honest” and passive and accurately displays everyone’s comments without any filtering or censorship, one doesn’t know if one can believe the comments or the crowd.

   For example, with Yelp, even among the sincere, fair reviewers, my impression is that it has gone through something like the “Eternal September” transition for internet discourse communities. Before truly mass adoption, the early reviewers tended to be more mature and leave higher quality reviews that were less “bipolar” or emotionally charged and more likely to match my own evaluations and tastes. Gradually, the number of bad (really, childish) reviewers exploded to far outnumber the good ones, and even if they are honest, they tend to go way over the top with their applause or criticism, so that today I find the Yelp reviews to be highly uncorrelated with quality. Indeed, one eventually develops a kind of instinctive algorithm for how to “translate” today’s dumb Yelp reviews into one’s own personal ratings dialect, but it takes a lot more effort to intelligently search through the reviews to perform the analysis, and that’s just not worth the trouble. There’s no Yelp for Yelps, but even if there were …

   3. The third part of the problem is that of course a lot of people (and who knows what proportion?) are not diligent, honest or fair at all with their comments and reviews, and there are a lot of people who are just out for vengeance for petty issues specifically to try and abuse the reputation system to manipulate public perception of reputation.

   4. The fourth part of the problem is that we don’t even know if the reputation systems themselves are presenting accurate, complete information. How do we know if the number of “likes” or “dislikes” or purported review comments are even “real” and left by normal, real consumers about their actual experiences? How can we tell in a virtual world if the things we can see are fake, and the things that were real are invisible to us because censored? How do we know if the fear or reality of defamation or fraud tort cases have suppressed the importantly valuable and accurate comments, or even if at the same time they prove so burdensome to pursue that they can only be leveraged by wealthy institutions and thus fail to provide adequate remedies to small entities who have been unfairly maligned? Not fake news, but fake reviews.

   And since the future of censorship will use adaptive / learning algorithms (because it’s too slow, laborious, and expensive to do with humans), even a reputation system’s owners and managers may not have a good understanding about what is and is not getting through.

   5. And fifth, there is the problem that reputation systems are social focal points with positive network effects and so tend to produce centralized quasi-monopolies with concomitant market power and insulation from competition (and sometimes even criticism.) For example, Google search is just such a focal point (where people might look for criticism of Google practices), and has gotten into trouble for inconspicuously manipulating the prioritization of search rankings for commercial vendors based on both payments and their own competing commercial interests. A “rank in a search” is certainly an important aspect of reputation, as we know from studying strong consumer responses to such rankings. There are some proposals for how to deal with this problem, from trust-busting the big companies (not a popular answer with many libertarians at the moment, though I can argue it’s consistent with the principles articulated by libertarian anti-discriminationists) to blockchain-based attempts to completely decentralize crowdsourced systems to prevent censorship. But there is probably no good answer and anyway those solutions do not seem to be on the table at the moment.

   Now, in the alternative to crowdsourced reputation systems, there are curated, professional expert, or institutional evaluators. There are things like Angie’s List and Consumer Reports, but how do I know that companies aren’t just paying for recommendations and reputations uncorrelated with quality and value? There’s Tyler Cowen’s ethnic food guide, but how do people know whether there’s an “inside joke” / “constrained economic environment of the future” aspect to his recommendation? There are things like Underwriters Laboratories, but except in cases they are liable for inaccuracies likely to be discovered, how can I know whether I can trust their seal?

   There are several reputation system issues that have been in the news lately. Amazon was deleting 1-star reviews of Hillary Clinton’s book. But how do we know whether that’s because the 1-star reviews were all bogus or whether Amazon was protecting Clinton and/or her publisher, or a mix of both? Warren Meyer was recently complaining that VRBO misleads people regarding whether rental houses were able to be reviewed at all with a “no reviews yet” message.

   Journalism itself is of course a particularly powerful reputation system, and coming back to book reviews, see how some outlets have treated Nancy MacLean’s trainwreck of a smear job with kid gloves or The Guardian’s glorification of Cordelia Fine’s anti-science pile of steaming nonsense Testosterone Rex: Unmaking the Myths of Our Gendered Minds, which, in a further undeserved boost to its reputation, won the Royal Society prize for science book of the year.

   It seems to me that we are living in an era of not just widespread breakdown in Meta-Reputation, but a general growing awareness of that breakdown, and a growing consensus that the breakdown does not originate in baseless paranoia but is fully justified and validated by enough last straws of real abuse to break a lot of camels’ backs.

   The question to libertarians is how does one answer the commerce assurance problem after the breakdown of reputation assurance, and I suppose a retort question to progressives is why should we trust the government to be any less untrustworthy.

   And perhaps there just is no good answer to either question and we are simply fated to observe the breakdown of all meta-reputation in our lifetime and the completion of the transition to a worse and much less pleasant and prosperous low-trust social equilibrium.

   P.S. Oh, almost forgot. Have a nice day.

   • Note that the 9th* of the Ten Commandments does not say “do not lie”, but “do not bear false witness against your neighbor”. Preserving reputation systems has been an age old problem.

     At some level it seems that people have to be continually reminded not to gossip or slander, and that they really do want an honest and fair reputation system even if it benefits people they don’t like because otherwise we will all lose out.

     *Some traditions sort the commandments differently, and thus label this the 8th

     • One of the reasons libertarians like to rely on “Reputation!” was because the common law definitely recognized the value and importance of reputation and the nature of this problem. And so, to ensure that people could benefit from general trust in a reliable system, it established strong structural incentives in the form of penalties and liability to dissuade people from engaging in forms of deceit and fraud like the tortious harms of malicious defamation and manipulative misrepresentation.

       The big problem is that these legal tools don’t work equally well in different social and economic contexts, and sometimes they are not adequate at all. (A higher order problem is that it has become prohibitively burdensome for ordinary people to pursue some of these remedies in court).

       Now, this is of course what many of the old progressives argued to justify the original expansion of the health and safety regulatory, inspection, and enforcement apparatus. They said that the incentives of the tort system were inadequate to ensure that commerce among increasingly atomized and alienated “strangers in the city” would reliably deliver goods and services of a consistent, minimally acceptable quality (” … as determined by new Scientific understanding …”) So the nascent nanny state was necessary to provide a substitute and plug the gap.

       It’s important to note just how unhealthy life in concentrated cities was before “the health transition” of the modern era, which gave rise to the idea of the big cities being “population sinks”, (e.g., life expectancy in London was once only 29.)

       The problem with the old progressive arguments, and the libertarian arguments against them, was that they were occurring at a time of significant and rapid change in many areas of life – which included technological and institutional innovations co-evolving with the emergent modern state apparatus – and so it’s just impossible to adjudicate the question of who was more right about their present and future counterfactuals, at which times.

       As it happened, the emergence of modern information systems produced real game-changing possibilities that simply transcended any practical arguments relying on old assumptions, or ideological arguments operating at some abstract, context-independent level. This happened in several key steps, each of which acted as a kind of saving grace for the strength of the libertarian argument. So, for example, even before the internet or even widespread popular use of computers, it became possible for companies and governments to keep databases about credit activity and car accidents and so forth, and to cheaply and quickly produce reports to interested parties so they didn’t have to rely on suspect accounts.

       Hence we got CARFAX, and we get libertarians criticizing Akerlof’s Market for Lemons and Stiglitz’s Information Asymmetry by pointing to CARFAX.

       Of course, how do you know you can trust a CARFAX report? Sometimes people get into wrecks, don’t report it to insurance or the police, and fix things on their own. We also got Equifax, but how do we know we can trust Equifax? They just got hacked, so one certainly can’t dismiss the possibility of tampering, like Ferris Bueller’s days absent or David Lightman’s grades.

       And now that we’re getting close to a situation in which everyone has internet all the time and uses it for commerce and to research and comment, the power of the “Reputation!” argument seemed to reach a new pinnacle.

       But the incentives to fraudulently manipulate reputation are incredibly strong, and so without inventing some kind of security measure to provide reputation-system assurance, it was inevitable that the systems themselves would come under one kind of attack or another, to the point where it is starting to become reasonable instead of paranoid to suspend one’s trust and have some healthy skepticism about the results, especially as the digital reputation-system industry becomes increasingly concentrated and monopolistic.

       So it’s not unreasonable to conclude that we may be at another transition in context in which the social incentives that preserved the value of the old reputation systems are no longer up to performing this function.

Comments are closed.