A reading list for anyone interested in tech policy

What follows is a syllabus. These are the key texts that will help boost your power of analysis. It is easy to make your way through college without having to think. These are the texts that make me think.

For a full list of datasets, resources, technical manuals, and the like, check out my commonplace.

Do the work

Always read the article. Always read the law. Always read the footnotes. Always read about the assumptions of the model. Do the work.

Start with The Butler Did It. It is short, less than a page, and it cuts right to the point:

Everyone knows the Butler did it. But in graduate school, you don’t have time to enjoy the whole mystery. You have to be able to very quickly take in who did it, how they did it, why they did it, when and where they did it, and who else knew about it. This method will help you get the answers that you need quickly and confidently and allow you to survive graduate school reading without undue stress.

The more you do this, the more you will see patterns in writing. It will help you tear through material. The more you read, the quicker you’ll become.

The best public policy work fits into one genre or another. For example,

Read good prose

Unless you had incredible teachers, most of the written material what you have encountered in your life is just bad prose. Seek out great writing. Here are some of my favorites:

Opposing arguments

The Dissoi Logoi, or Opposing Arguments, draws its title from its first few words. It is an untitled work that concerns more than just opposing arguments. The treatise was written by an anonymous author around 403 to 395 B.C.E. Most scholars agree that the author was a Sophist (and therefore probably male) who was strongly influenced by Protagoras and was even possibly his student, and was also influenced by Hippias, Gorgias, and Socrates.

I see the Dissoi Logoi as an exercise. It forces you to think of contrasts:

For these things are bad for those who are sick, but good for the person who is healthy and needs them. Or again, lack of restraint in these matters is bad for those who lack restraint, but good for and make money out of them. And illness is bad for the sick but good for the doctors. And death is bad for those who die, but good for the undertakers and the grave-diggers. Farming also, when it makes a handsome success of producing crops, is good for the farmers, but bad for the merchants. And is bad for the ship-owner if his merchantships are involved in collision or get smashed up, but good for the shipbuilders. Furthermore, it is bad for everyone else, but good for the blacksmiths if a tool corrodes or loses its sharp edge or gets broken to pieces…

The Dissoi Logoi, or something like it, animated Plato to search for truth. In other words, Plato began the search for transcendent forms of knowledge to anchor moral judgment in direct response to Protagoras.

Anti-logos and the Protagorean Spirit

Rhetoric is a lost practice. What we think of now as rhetoric is actually ornamental rhetoric, the proper way of speaking. Before the Renaissance, rhetoric also included another type of rhetoric, pragmatic rhetoric, which is the study of argumentation.

My favorite modern treatment of the subject is Michael Billig’s Arguing and Thinking: A Rhetorical Approach to Social Psychology. He has this to say about Protagoras:

If Protagoras is interpreted in this way, he can be seen as making a statement about the constant possibility that any speech can be opposed by a counter-speech. At this point it might be helpful to take liberties with classical Greek, at the risk of offending the purists. Rather than keep with the term antilogikos, the concept of ‘anti-logos’ (or ‘anti-logoi’ in the plural) can be introduced for the sake of convenience. Literally ‘logos’ is Greek for ‘word’, but, in both its singular and plural forms, it has a much wider significance, which makes the English ‘word’ or ‘words’ an impoverished translation. ‘Logos’ denotes word-making in general, and so can be used as a synonym for discourse, speech or talk.

Billig calls this anti-logos the Protagorean spirit of contradiction. The power of this kind of thinking comes into focus when Billig drills down into the “all the world’s a stage” metaphor:

What the dramaturgical metaphor does not do is to take the whole theatrical world as a model for social life. It narrows the world of the theatre down to the staged performance. If all the world is a stage, then what goes on backstage is being excluded. Thus, a complete sub-world, that of the theatre, is not being considered as the model for social life, but only one element of that sub-world: the public performance. The problem is that this is the one part of the theatrical world which demands the suppression of arguments. During a performance, all members of the cast must leave their disagreements in the wings, and must work together to produce the drama. On stage, a scene would be destroyed if an actor were to voice a backstage complaint, declaring suddenly that a fellow player suffered from bad breath or from a complete lack of talent."

All logos contains within it an anti-logos. Every argument demands the suppression of other arguments.

Do read the whole book.

Wicked problems

Public policy is defined by its complexity, or what Horst W. J. Rittel & Melvin M. Webber call wicked problems in Dilemmas in a general theory of planning. Public policy is plagued by three interconnected and intractable problems:

  1. defining problems (of knowing what distinguishes an observed condition from a desired condition)
  2. of locating problems (finding where in the complex causal networks the trouble really lies).
  3. identifying the actions that might effectively narrow the gap between what-is and what-ought-to-be.”

Rittel and Webber call them Wicked problems.

As distinguished from problems in the natural sciences, which are definable and separable and may have solutions that are findable, the problems of governmental planning and especially those of social or policy planning are ill-defined; and they rely upon elusive political judgment for resolution. (Not “solution.” Social problems are never solved. At best they are only resolved-over and over again.)

“As we seek to improve the effectiveness of actions in pursuit of valued outcomes, as system boundaries get stretched, and as we become more sophisticated about the complex workings of open societal systems, it becomes ever more difficult to make the planning idea operational.”

See also: Oliver Williamson on remediability & John List “The Voltage Effect”

What are the myths underlying your research?

Hans Noel’s “Ten Things Political Scientists Know that You Don’t” is the perfect mythbusting article. As he opens,

Many political scientists would like journalists and political practitioners to take political science more seriously, and many are beginning to pay attention. This paper outlines ten things that political science scholarship has learned that are at odds with much of the conventional wisdom of American politics.

While mythbusting can get old as a form of analysis, it can be an important way to enter a topic. What are the conventional truths that everyone seems to agree upon?

The seen and the unseen, or the limits to information

Frederic Bastiat’s classic What Is Seen and What Is Not Seen laid out the goalposts for economists long ago. “There is only one difference between a bad economist and a good one,” he said. “The bad economist confines himself to the visible effect; the good economist takes into account both the effect that can be seen and those effects that must be foreseen.”

The seen and the unseen have echoes of the Dissoi Logoi.

The theme of the seen and the unseen has a family resemblence to the theme of withholding attention from other parts. Pedro Domingos' (2012) A Few Useful Things to Know about Machine Learning" offers a variation on the theme as well.

Nevertheless, you should be skeptical of claims that a particular technique “solves” the overfitting problem. It’s easy to avoid overfitting (variance) by falling into the opposite error of underfitting (bias). Simultaneously avoiding both requires learning a perfect classifier, and short of knowing it in advance there is no single technique that will always do best (no free lunch).

See also “Data mining fool’s gold” by Gary Smith

How does attention get economized?

Herbert Simon’s Designing organizations for an information-rich world remains one of my top reads because it is generative. Simon games games out the impacts of an information-rich world. And it is driven by a simple yet powerful question: In a surfeit of information, what gets economized?

Simon’s reasoning is always worth quoting in full:

[I]n an information-rich world, the wealth of information means a dearth of something else: a scarcity of whatever it is that information consumes. What information consumes is rather obvious: it consumes the attention of its recipients. Hence a wealth of information creates a poverty of attention and a need to allocate that attention efficiently among the overabundance of information sources that might consume it.

In applying this logic to individuals, Simon was the first to name and describe the attention economy. And in applying this logic to organizations, Simon brought attention to the processes that condense information for groups and organizations. The key question for understanding information in systems is to understand what is being “withheld from the attention of other parts of the system.”

For a current review of the literature on the attention economy, see:

Companies aren’t seeking profits, they’re seeking alpha

There is this pernicious belief that companies are seeking to maximize profits. But this isn’t exactly right. Firms are looking for positive returns. They are seeking alpha.

Armen Alchian’s “Uncertainty, Evolution, and Economic Theory” (1950) plumbs this topic to its depths. As a result, Alchian ends up making a deeply rhetorical argument about the language we use to describe economic phenomena:

Under uncertainty, by definition, each action that may be chosen is identified with a distribution of potential outcomes, not with a unique outcome. Implicit in uncertainty is the consequence that these distributions of potential outcomes are overlapping. It is worth emphasis that each possible action has a distribution of potential outcomes, only one of which will materialize if the action is taken, and that one outcome cannot be foreseen. Essentially, the task is converted into making a decision (selecting an action) whose potential outcome distribution is preferable, that is, choosing the action with the optimum distribution, since there is no such thing as a maximizing distribution.

In other words, companies aren’t seeking to maximize profits if there is uncertainty in the world.

Instead, they are trying to choose those actions that will give better distributional outcomes. Because these actions are always vexed with uncertainty, what ends up getting chosen is a distribution of returns, not a maximization.

For what its worth “everything said is applicable equally to utility maximization by consumers.” Consumers don’t maximize utility either.

Prices, dispersed knowledge, and the economics of information

Hayek opens his classic essay The Use of Knowledge in Society with one of the big questions in economics, “What is the problem we wish to solve when we try to construct a rational economic order?”

This article has been incredibly influential, netting Hayek the Nobel Prize and inspiring other work like Leonard Reed’s “I, Pencil.” His answer to this question has been increbily generative.

Prices economize on dispersed knowledge.

The Economics of Information” (1961) by George J. Stigler The search for information–including the prices buyers and sellers are willing to give–is fundamental to understanding the workings of the market. Nobel Prize-winner, George J. Stigler, explains the importance of information, and information-sharing devices like advertising, in bringing the market towards equilibrium in this seminal article. Differential prices, or what economists call dispersion, exist due to ignorance of buyers and sellers. Ignorance is ubiquitous due to the costliness of searching for information. Advertising reduces dispersion by decreasing search costs to buyers.

The error cost framework

The error cost framework advocates for the use of “filters” to screen antitrust cases and minimize the costs of judicial errors when assessing the competitive effects of business practices.

As Judge Frank Easterbrook first laid out in The Limits of Antitrust,

  • Antitrust enforcement is imperfect because judges and juries often lack the economic knowledge to accurately evaluate complex business practices. Condemning beneficial practices creates significant costs.
  • The traditional Rule of Reason analysis asks courts to conduct an open-ended inquiry into competitive effects, which they are ill-equipped to do effectively. This leads to high error costs.
  • The per se rule reduces error costs by conclusively presuming certain practices are anticompetitive, but has been narrowed as more potential benefits of practices have been recognized.

The error cost framework proposes using sequential “filters” to screen cases. Each filter would aim to eliminate cases where the risk of anticompetitive harm is low. The filters focus on market power, profit incentives, widespread use of the practice, output effects, and plaintiff identity. The goal is to dismiss cases where competitive harm is unlikely. The filters intentionally err on the side of permitting questionable practices, to account for the greater costs of wrongly condemning beneficial practices versus mistakenly allowing harmful ones.

If we assume that all agencies are prone to error, then the FTC will stop transactions that are good for consumers (a false positive), while also glossing over transactions that aren’t in the consumer interest (a false negative). As antitrust expert Judge Frank Easterbrook noted nearly 30 years ago, false positives are often swept away by competitive forces in the market. However, false negatives impose unforeseen costs on consumers as transactions that are in the consumer interest are deterred. The FTC has come a long way in limiting these costly actions, aided by the injection of economic reasoning at all levels of analysis. But the FCC is not constrained by these limits and is far more expansive, which is leading to higher unforeseen costs for consumers.

Discounting and time

Standard economic models see risk taking and time discounting as two independent dimensions of decision making. However, there is a lot of evidence that risk taking and time discounting are twinned. The inherent uncertainty associated with the future generates a unifying framework for explaining a large number of puzzling behavioral regularities like delay-dependent risk tolerance, aversion to sequential resolution of uncertainty, preferences for the timing of the resolution of uncertainty, the differential discounting of risky and certain outcomes, hyperbolic discounting, subadditive discounting, and the order dependence of prospect valuation.

Platform economics

Multi-sided platforms create value by bringing different economic agents togethers. They facilitate interaction among these agents and generate welfare for individual agents by reducing transaction costs. Such platforms aren’t new. They’ve been around for decades in industries like video games, credit cards, newspapers, and radio stations. The Twin Cities’ Mall of America is as much a platform as Google is. The Internet has only facilitated the creation of such platforms by allowing agents to interact in real-time.

I’ve also written about this topic in the past:

Transactions costs economics

Coase’s “The Nature of the Firm,” aims “to discover why a firm emerges at all in a specialised exchange economy.” In other words, why do we have companies in the first place? Why don’t we just contract in the marketplace for everything?

Firms exist because they are better equipped to deal with the transaction costs inherent in production and exchange than individuals are. Coase’s work on firms, transaction costs and contracts was later advanced by economists Oliver Williamson, Douglass North, Oliver Hart, Bengt Holmström, Arman Alchian and Harold Demsetz. Economists and political scientists have used insights from Coase’s work to explain the functioning of organizations in general, not just firms.


The economics of AI

Robots as workers

Zero priced goods

The economics of privacy

Freedom of speech

Power laws

Power-law distributions occur in an extraordinarily diverse range of phenomena. In addition to city populations, the sizes of earthquakes, moon craters, solar flares, computer files and wars, the frequency of use of words in any human language, the frequency of occurrence of personal names in most cultures, the numbers of papers scientists write, the number of citations received by papers, the number of hits on web pages, the sales of books, music recordings and almost every other branded commodity, the numbers of species in biological taxa, people’s annual incomes and a host of other variables all follow power-law distributions.