Digifesto

Category: economics

State regulation and/or corporate self-regulation

The dust from the recent debates about whether regulation or industrial self-regulation in the data/tech/AI industry appears to be settling. The smart money is on regulation and self-regulation being complementary for attaining the goal of an industry dominated by responsible actors. This trajectory leads to centralized corporate power that is lead from the top; it is a Hamiltonian not Jeffersonian solution, in Pasquale’s terms.

I am personally not inclined towards this solution. But I have been convinced to see it differently after a conversation today about environmentally sustainable supply chains in food manufacturing. Nestle, for example, has been internally changing its sourcing practices to more sustainable chocolate. It’s able to finance this change from its profits, and when it does change its internal policy, it operates on a scale that’s meaningful. It is able to make this transition in part because non-profits, NGO’s, and farmers cooperatives lay through groundwork for sustainable sourcing external to the company. This lowers the barriers to having Nestle switch over to new sources–they have already been subsidized through philanthropy and international aid investments.

Supply chain decisions, ‘make-or-buy’ decisions, are the heart of transaction cost economics (TCE) and critical to the constitution of institutions in general. What this story about sustainable sourcing tells us is that the configuration of private, public, and civil society institutions is complex, and that there are prospects for agency and change in the reconfiguration of those relationships. This is no different in the ‘tech sector’.

However, this theory of economic and political change is not popular; it does not have broad intellectual or media appeal. Why?

One reason may be because while it is a critical part of social structure, much of the supply chain is in the private sector, and hence is opaque. This is not a matter of transparency or interpretability of algorithms. This is about the fact that private institutions, by virtue of being ‘private’, do not have to report everything that they do and, probably, shouldn’t. But since so much of what is done by the massive private sector is of public import, there’s a danger of the privatization of public functions.

Another reason why this view of political change through the internal policy-making of enormous private corporations is unpopular is because it leaves decision-making up to a very small number of people–the elite managers of those corporations. The real disparity of power involved in private corporate governance means that the popular attitude towards that governance is, more often than not, irrelevant. Even less so that political elites, corporate elites are not accountable to a constituency. They are accountable, I suppose, to their shareholders, which have material interests disconnected from political will.

This disconnected shareholder will is one of the main reasons why I’m skeptical about the idea that large corporations and their internal policies are where we should place our hopes for moral leadership. But perhaps what I’m missing is the appropriate intellectual framework for how this will is shaped and what drives these kinds of corporate decisions. I still think TCE might provide insights that I’ve been missing. But I am on the lookout for other sources.

Advertisements

Ordoliberalism and industrial organization

There’s a nice op-ed by Wolfgang Münchau in FT, “The crisis of modern liberalism is down to market forces”.

Among other things, it reintroduces the term “ordoliberalism“, a particular Germanic kind of enlightened liberalism designed to prevent the kind of political collapse that had precipitated the war.

In Münchau’s account, the key insight of ordoliberalism is its attention to questions of social equality, but not through the mechanism of redistribution. Rather, ordoliberal interventions primarily effect industrial organization, favoring small to mid- sized companies.

As Germany’s economy remains robust and so far relatively politically stable, it’s interesting that ordoliberalism isn’t discussed more.

Another question that must be asked is to what extent the rise of computational institutions challenges the kind of industrial organization recommended by ordoliberalism. If computation induces corporate concentration, and there are not good policies for addressing that, then that’s due to a deficiency in our understanding of what ‘market forces’ are.

computational institutions

As the “AI ethics” debate metastasizes in my newsfeed and scholarly circles, I’m struck by the frustrations of technologists and ethicists who seem to be speaking past each other.

While these tensions play out along disciplinary fault-lines, for example, between technologists and science and technology studies (STS), the economic motivations are more often than not below the surface.

I believe this is to some extent a problem of the nomenclature, which is again the function of the disciplinary rifts involved.

Computer scientists work, generally speaking, on the design and analysis of computational systems. Many see their work as bounded by the demands of the portability and formalizability of technology (see Selbst et al., 2019). That’s their job.

This is endlessly unsatisfying to critics of the social impact of technology. STS scholars will insist on changing the subject to “sociotechnical systems”, a term that means something very general: the assemblage of people and artifacts that are not people. This, fairly, removes focus from the computational system and embeds it in a social environment.

A goal of this kind of work seems to be to hold computational systems, as they are deployed and used socially, accountable. It must be said that once this happens, we are no longer talking about the specialized domain of computer science per se. It is a wonder why STS scholars are so often picking fights with computer scientists, when their true beef seems to be with businesses that use and deploy technology.

The AI Now Institute has attempted to rebrand the problem by discussing “AI Systems” as, roughly, those sociotechnical systems that use AI. This is one the one hand more specific–AI is a particular kind of technology, and perhaps it has particular political consequences. But their analysis of AI systems quickly overflows into sweeping claims about “the technology industry”, and it’s clear that most of their recommendations have little to do with AI, and indeed are trying, once again, to change the subject from discussion of AI as a technology (a computer science research domain) to a broader set of social and political issues that do, in fact, have their own disciplines where they have been researched for years.

The problem, really, is not that any particular conversation is not happening, or is being excluded, or is being shut down. The problem is that the engineering focused conversation about AI-as-a-technology has grown very large and become an awkward synecdoche for the rise of major corporations like Google, Apple, Amazon, Facebook, and Netflix. As these corporations fund and motivate a lot of research, there’s a question of who is going to get pieces of the big pie of opportunity these companies represent, either in terms of research grants or impact due to regulation, education, etc.

But there are so many aspects of these corporations that are neither addressed by the terms “sociotechnical system”, which is just so broad, and “AI System”, which is as broad and rarely means what you’d think it does (that the system uses AI is incidental if not unnecessary; what matters is that it’s a company operating in a core social domain via primarily technological user interfaces). Neither of these gets at the unit of analysis that’s really of interest.

An alternative: “computational institution”. Computational, in the sense of computational cognitive science and computational social science: it denotes the essential role of theory of computation and statistics in explaining the behavior of the phenomenon being studied. “Institution”, in the sense of institutional economics: the unit is a firm, which is comprised of people, their equipment, and their economic relations, to their suppliers and customers. An economic lens would immediately bring into focus “the data heist” and the “role of machines” that Nissenbaum is concerned are being left to the side.

The politics of AI ethics is a seductive diversion from fixing our broken capitalist system

There is a lot of heat these days in the tech policy and ethics discourse. There is an enormous amount of valuable work being done on all fronts. And yet there is also sometimes bitter disciplinary infighting and political intrigue about who has the moral high ground.

The smartest thing I’ve read on this recently is Irina Raicu’s “False Dilemmas” piece, where she argues:

  • “Tech ethics” research, including research explore the space of ethics in algorithm design, is really code for industry self-regulation
  • Industry self-regulation and state regulation are complementary
  • Any claims that “the field” is dominated by one perspective or agenda or another is overstated

All this sounds very sane but it doesn’t exactly explain why there’s all this heated discussion in the first place. I think Luke Stark gets it right:

But what does it mean to say “the problem is mostly capitalism”? And why is it impolite to say it?

To say “the problem [with technology ethics and policy] is capitalism” is to note that most if not all of the social problems we associate with today’s technology have been problems with technology ever since the industrial revolution. For example, James Beniger‘s The Control Revolution, Horkheimer‘s Eclipse of Reason, and so on all speak to the tight link that there has always been between engineering and the capitalist economy as a whole. The link has persisted through the recent iterations of recognizing first data science, then later artificial intelligence, as disruptive triumphs of engineering with a variety of problematic social effects. These are old problems.

It’s impolite to say this because it cuts down on the urgency that might drive political action. More generally, it’s an embarrassment to anybody in the business of talking as if they just discovered something, which is what journalists and many academics do. The buzz of novelty is what gets people’s attention.

It also suggests that the blame for how technology has gone wrong lies with capitalists, meaning, venture capitalists, financiers, and early stage employees with stock options. But also, since it’s the 21st century, pension funds and university endowments are just as much a part of the capitalist investing system as anybody else. In capitalism, if you are saving, you are investing. Lots of people have a diffuse interest in preserving capitalism in some form.

There’s a lot of interesting work to be done on financial regulation, but it has very little to do with, say, science and technology studies and consumer products. So to acknowledge that the problem with technology is capitalism changes the subject to something remote and far more politically awkward than to say the problem is technology or technologists.

As I’ve argued elsewhere, a lot of what’s happening with technology ethics can be thought of as an extension of what Nancy Fraser called progressive neoliberalism: the alliance of neoliberalism with progressive political movements. It is still hegemonic in the smart, critical, academic and advocacy scene. Neoliberalism, or what is today perhaps better characterized as finance capitalism or surveillance capitalism, is what is causing the money to be invested in projects that design and deploy technology in certain ways. It is a system of economic distribution that is still hegemonic.

Because it’s hegemonic, it’s impolite to say so. So instead a lot of the technology criticism gets framed in terms of the next available moral compass, which is progressivism. Progressivism is a system of distribution of recognition. It calls for patterns of recognizing people for their demographic and, because it’s correlated in a sensitive way, professional identities. Nancy Fraser’s insight is that neoliberalism and progressivism have been closely allied for many years. One way that progressivism is allied with neoliberalism is that progressivism serves as a moral smokescreen for problems that are in part caused by neoliberalism, preventing an effective, actionable critique of the root cause of many technology-related problems.

Progressivism encourages political conflict to be articulated as an ‘us vs. them’ problem of populations and their attitudes, rather than as problem of institutions and their design. This “us versus them” framing is baldly stated than in the 2018 AI Now Report:

The AI accountability gap is growing: The technology scandals of 2018 have shown that the gap between those who develop and profit from AI—and those most likely to suffer the consequences of its negative effects—is growing larger, not smaller. There are several reasons for this, including a lack of government regulation, a highly concentrated AI sector, insufficient governance structures within technology companies, power asymmetries between companies and the people they serve, and a stark cultural divide between the engineering cohort responsible for technical research, and the vastly diverse populations where AI systems are deployed. (Emphasis mine)

There are several institutional reforms called for in the report, but the focus on a particular sector that it constructs as “the technology industry” composed on many “AI systems”, it cannot address broader economic issues such as unfair taxation or gerrymandering. Discussion of the overall economy is absent from the report; it is not the cause of anything. Rather, the root cause is a schism between kinds of people. The moral thrust of this claim hinges on the implied progressivism: the AI/tech people, who are developing and profiting, are a culture apart. The victims are “diverse”, and yet paradoxically unified in their culture as not the developers. This framing depends on the appeal of progressivism as a unifying culture whose moral force is due in large part because of its diversity. The AI developer culture is a threat in part because it is separate from diverse people–code for its being white and male.

This thread continues throughout the report, as various critical perspectives are cited in the report. For example:

A second problem relates to the deeper assumptions and worldviews of the designers of ethical codes in the technology industry. In response to the proliferation of corporate ethics initiatives, Greene et al. undertook a systematic critical review of high-profile “vision statements for ethical AI.” One of their findings was that these statements tend to adopt a technologically deterministic worldview, one where ethical agency and decision making was delegated to experts, “a narrow circle of who can or should adjudicate ethical concerns around AI/ML” on behalf of the rest of us. These statements often assert that AI promises both great benefits and risks to a universal humanity, without acknowledgement of more specific risks to marginalized populations. Rather than asking fundamental ethical and political questions about whether AI systems should be built, these documents implicitly frame technological progress as inevitable, calling for better building.

That systematic critical reviews of corporate policies express self-serving views that ultimately promote the legitimacy of the corporate efforts is a surprise to no one; it is no more a surprise than the fact that critical research institutes staffed by lawyers and soft social scientists write reports recommending that their expertise is vitally important for society and justice. As has been the case in every major technology and ethical scandal for years, the first thing the commentariat does is publish a lot of pieces justifying their own positions and, if they are brave, arguing that other people are getting too much attention or money. But since everybody in either business depends on capitalist finance in one way or another, the economic system is not subject to critique. In other words, once can’t argue that industrial visions of ‘ethical AI’ are favorable to building new AI products because they are written in service to capitalist investors who profit from the sale of new AI products. Rather, one must argue that they are written in this way because the authors have a weird technocratic worldview that isn’t diverse enough. One can’t argue that the commercial AI products neglect marginal populations because these populations have less purchasing power; one has to argue that the marginal populations are not represented or recognized enough.

And yet, the report paradoxically both repeatedly claims that AI developers are culturally and politically out of touch and lauds the internal protests at companies like Google that have exposed wrongdoing within those corporations. The actions of “technology industry” employees belies the idea that problem is mainly cultural; there is a managerial profit-making impulse that is, in large, stable companies in particular, distinct from that the rank-and-file engineer. This can be explained in terms of corporate incentives and so on, and indeed the report does in places call for whistleblower protections and labor organizing. But these calls for change cut against and contradict other politically loaded themes.

There are many different arguments contained in the long report; it is hard to find a reasonable position that has been completely omitted. But as a comprehensive survey of recent work on ethics and regulation in AI, its biases and blind spots are indicative of the larger debate. The report concludes with a call for a change in the intellectual basis for considering AI and its impact:

It is imperative that the balance of power shifts back in the public’s favor. This will require significant structural change that goes well beyond a focus on technical systems, including a willingness to alter the standard operational assumptions that govern the modern AI industry players. The current focus on discrete technical fixes to systems should expand to draw on socially-engaged disciplines, histories, and strategies capable of providing a deeper understanding of the various social contexts that shape the development and use of AI systems.

As more universities turn their focus to the study of AI’s social implications, computer science and engineering can no longer be the unquestioned center, but should collaborate more equally with social and humanistic disciplines, as well as with civil society organizations and affected

communities. (Emphasis mine)

The “technology ethics” field is often construed, in this report but also in the broader conversation, as one of tension between computer science on the one hand, and socially engaged and humanistic disciplines on the other. For example, Selbst et al.’s “Fairness and Abstraction in Sociotechnical Systems” presents a thorough account of pitfalls of computer science’s approach to fairness in machine learning, and proposes a Science and Technology Studies. The refrain is that by considering more social context, more nuance, and so on, STS and humanistic disciplines avoids the problems that engineers, who try to provide portable, formal solutions, don’t want to address. As the AI Now report frames it, a benefit of the humanistic approach is that it brings the diverse non-AI populations to the table, shifting the balance of power back to the public. STS and related disciplines claim the status of relevant expertise in matters of technology that is somehow not the kind of expertise that is alienating or inaccessible to the public, unlike engineering, which allegedly dominates the higher education system.

I am personally baffled by these arguments; so often they appear to conflate academic disciplines with business practices in ways that most practitioners I engage with would not endorse. (Try asking an engineer how much they learned in school, versus on the job, about what it’s like to work in a corporate setting.) But beyond the strange extrapolation from academic disciplinary disputes (which are so often about the internal bureaucracies of universities it is, I’d argue after learning the hard way, unwise to take them seriously from either an intellectual or political perspective), there is also a profound absence of some fields from the debate, as framed in these reports.

I’m referring to the quantitative social sciences, such as economics and quantitative sociology, or what might be more be more generally converging on computational social science. These are the disciplines that one would need to use to understand the large-scale, systemic impact of technology on people, including the ways costs and benefits are distributed. These disciplines deal with social systems and include technology–there is a long tradition within economics studying the relationship between people, goods, and capital that never once requires the term “sociotechnical”–in a systematic way that can be used to predict the impact of policy. They can also connect, through applications of business and finance, the ways that capital flows and investment drive technology design decisions and corporate competition.

But these fields are awkwardly placed in technology ethics and politics. They don’t fit into the engineering vs. humanities dichotomy that entrances so many graduate students in this field. They often invoke mathematics, which makes them another form of suspicious, alien, insufficiently diverse expertise. And yet, it may be that these fields are the only ones that can correctly diagnose the problems caused by technology in society. In a sense, the progressive framing of the problems of technology makes technogy’s ills a problem of social context because it is unequipped to address them as a problem of economic context, and it wouldn’t want know that it is an economic problem anyway, for two somewhat opposed reasons: (a) acknowledging the underlying economic problems is taboo under hegemonic neoliberalism, and (b) it upsets the progressive view that more popularly accessible (and, if you think about it quantitatively, therefore as a result of how it is generated and constructed more diverse) humanistic fields need to be recognized as much as fields of narrow expertise. There is no credence given to the idea that narrow and mathematized expertise might actually be especially well-suited to understand what the hell is going on, and that this is precisely why members of these fields are so highly sought after by investors to work at their companies. (Consider, for example, who would be best positioned to analyze the “full stack supply chain” of artificial intelligence systems, as is called for by the AI Now report: sociologists, electrical engineers trained in the power use and design of computer chips, or management science/operations research types whose job is to optimize production given the many inputs and contingencies of chip manufacture?)

At the end of the day, the problem with the “technology ethics” debate is a dialectic cycle whereby (a) basic research is done by engineers, (b) that basic research is developed in a corporate setting as a product funded by capitalists, (c) that product raises political hackles and makes the corporations a lot of money, (d) humanities scholars escalate the political hackles, (e) basic researchers try to invent some new basic research because the politics have created more funding opportunities, (f) corporations do some PR work trying to CYA and engage in self-regulation to avoid litigation, (g) humanities scholars, loathe to cede the moral high ground, insist the scientific research is inadequate and that the corporate PR is bull. But this cycle is not necessarily productive. Rather, it sustains itself as part of a larger capitalist system that is bigger than any of these debates, structures its terms, and controls all sides of the dialog. Meanwhile the experts on how that larger system works are silent or ignored.

References

Fraser, Nancy. “Progressive neoliberalism versus reactionary populism: A choice that feminists should refuse.” NORA-Nordic Journal of Feminist and Gender Research 24.4 (2016): 281-284.

Greene, Daniel, Anna Laura Hoffman, and Luke Stark. “Better, Nicer, Clearer, Fairer: A Critical Assessment of the Movement for Ethical Artificial Intelligence and Machine Learning.” Hawaii International Conference on System Sciences, Maui, forthcoming. Vol. 2019. 2018.

Raicu, Irina. “False Dilemmas”. 2018.

Selbst, Andrew D., et al. “Fairness and Abstraction in Sociotechnical Systems.” ACM Conference on Fairness, Accountability, and Transparency (FAT*). 2018.

Whittaker, Meredith et al. “AI Now Report 2018”. 2018.

The secret to social forms has been in institutional economics all along?

A long-standing mystery for me has been about the ontology of social forms (1) (2): under what conditions is it right to call a particular assemblage of people a thing, and why? Most people don’t worry about this; in literatures I’m familiar with it’s easy to take a sociotechnical complex or assemblage, or a company, or whatever, as a basic unit of analysis.

A lot of the trickiness comes from thinking about this as a problem of identifying social structure (Sawyer, 200; Cederman, 2005). This implies that people are in some sense together and obeying shared norms, and raises questions about whether those norms exist in their own heads or not, and so on. So far I haven’t seen a lot that really nails it.

But what if the answer has been lurking in institutional economics all along? The “theory of the firm” is essentially a question of why a particular social form–the firm–exists as opposed to a bunch of disorganized transactions. The answers that have come up are quite good.

Take for example Holmstrom (1982), who argues that in a situation where collective outcomes depend on individual efforts, individuals will be tempted to free-ride. That makes it beneficial to have somebody monitor the activities of the other people and have their utility be tied to the net success of the organization. That person becomes the owner of the company, in a capitalist firm.

What’s nice about this example is that it explains social structure based on an efficiency argument; we would expect organizations shaped like this to be bigger and command more resources than others that are less well organized. And indeed, we have many enormous hierarchical organizations in the wild to observe!

Another theory of the firm is Williamson’s transaction cost economics (TCE) theory, which is largely about the make-or-buy decision. If the transaction between a business and its supplier has “asset specificity”, meaning that the asset being traded is specific to the two parties and their transaction, then any investment from either party will induce a kind of ‘lock-in’ or ‘switching cost’ or, in Williamson’s language, a ‘bilateral dependence’. The more of that dependence, the more a free market relationship between the two parties will expose them to opportunistic hazards. Hence, complex contracts, or in the extreme case outright ownership and internalization, tie the firms together.

I’d argue: bilateral dependence and the complex ‘contracts’ the connect entities are very much the stuff of “social forms”. Cooperation between people is valuable; the relation between people who cooperate is valuable as a consequence; and so both parties are ‘structurated’ (to mangle a Giddens term) individually into maintaining the reality of the relation!

References

Cederman, L.E., 2005. Computational models of social forms: Advancing generative process theory 1. American Journal of Sociology, 110(4), pp.864-893.

Holmstrom, Bengt. “Moral hazard in teams.” The Bell Journal of Economics (1982): 324-340.

Sawyer, R. Keith. “Simulating emergence and downward causation in small groups.” Multi-agent-based simulation. Springer Berlin Heidelberg, 2000. 49-67.

Williamson, Oliver E. “Transaction cost economics.” Handbook of new institutional economics. Springer, Berlin, Heidelberg, 2008. 41-65.

Transaction cost economics and privacy: looking at Hoofnagle and Whittington’s “Free”

As I’ve been reading about transaction cost economics (TCE) and independently scrutinizing the business model of search engines, it stands to reason that I should look to the key paper holding down the connection between TCE and privacy, Hoofnagle and Whittinton’s “Free: Accounting for the Costs of the Internet’s Most Popular Price” (2014).

I want to preface the topic by saying I stand by what I wrote earlier: that at the heart of what’s going on with search engines, you have a trade of attention; it requires imagining the user has have attention-time as a scarce resource. The user has a query and has the option to find material relevant to the query in a variety of ways (like going to a library). Often (!) they will do so in a way that costs them as little attention as possible: they use a search engine, which gives an almost instant and often high-quality response; they are also shown advertisements which consume some small amount of their attention, but less than they would expend searching through other means. Advertisers pay the search engine for this exposure to the user’s attention, which funds the service that is “free”, in dollars (but not in attention) to the users.

Hoofnagle and Whittington make a very different argument about what’s going on with “free” web services, which includes free search engines. They argue that the claim that these web services are “free” is deceptive because the user may incur costs after the transaction on account of potential uses of their personal data. An example:

The freemium business model Anderson refers to is popular among industries online. Among them, online games provide examples of free services with hidden costs. By prefacing play with the disclosure of personal identification, the firms that own and operate games can contact and monitor each person in ways that are difficult for the consumer to realize or foresee. This is the case for many games, including Disney’s “Club Penguin,” an entertainment website for children. After providing personal information to the firm, consumers of Club Penguin receive limited exposure to basic game features and can see numerous opportunities to enrich their play with additional features. In order to enrich the free service, consumers must buy all sort of enhancements, such as an upgraded igloo or pets for one’s penguin. Disney, like others in the industry, places financial value on the number of consumers it identifies, the personal information they provide, and the extent to which Disney can track consumer activity in order to modify the game and thus increase the rate of conversion of consumers from free players to paying customers.

There are a number of claims here. Let’s enumerate them:

  1. This is an example of a ‘free’ service with hidden costs to users.
  2. The consumer doesn’t know what the game company will do with their personal information.
  3. In fact, the game will use the personal information to personalize pitches for in-game purchases that ‘enrich’ the free service.
  4. The goal of the company is to convert free players to paying customers.

Working backwards, claim (4) is totally true. The company wants to make money by getting their customers to pay, and they will use personal information to make paying attractive to the customers (3). But this does not mean that the customer is always unwitting. Maybe children don’t understand the business model when they begin playing Penguin Club, but especially today parents certainly do. App Stores, for example, now label apps when they have “in-app purchases”, which is a pretty strong signal. Perhaps this is a recent change due to some saber rattling by the FTC, which to be fair would be attributable as a triumph to the authors if this article had influence on getting that to happen. On the other hand, this is a very simple form of customer notice.

I am not totally confident that even if (2), (3), and (4) are true, that that entails (1), that there are “hidden costs” to free services. Elsewhere, Hoofnagle and Whittington raise more convincing examples of “costs” to release of PII, including being denied a job and resolving identity theft. But being convincingly sold an upgraded igloo for your digital penguin seems so trivial. Even if it’s personalized, how could it be a hidden cost? It’s a separate transaction, no? Do you or do you not buy the igloo?

Parsing this through requires, perhaps, a deeper look at TCE. According to TCE, agents are boundedly rational (they can’t know everything) and opportunistic (they will make an advantageous decision in the moment). Meanwhile, the world is complicated. These conditions imply that there’s a lot of uncertainty about future behavior, as agents will act strategically in ways that they can’t themselves predict. Nevertheless, agents engage in contracts with some kinds of obligations in them in the course of a transaction. TCE’s point is that these contracts are always incomplete, meaning that there are always uncertainties left unresolved in contracts that will need to be negotiated in certain contingent cases. All these costs of drafting, negotiating, and safeguarding the agreement are transaction costs.

Take an example of software contracting, which I happen to know about from personal experience. A software vendor gets a contract from a client to do some customization. The client and the vendor negotiated some sort of scope of work ex ante. But always(!), the client doesn’t actually know what they want, and if the vendor delivers on the specification literally the client doesn’t like it. Then begins the ex post negotiation as the client tries to get the vendor to tweak the system into something more usable.

Software contracting often resolves this by getting off the fixed cost contracting model and onto a cost-and-materials contact that allows billing by hours of developer time. Alternatively, the vendor can internalize the costs into the contract by inflating the cost “estimates” to cover for contingencies. In general, this all amounts to having more contract and a stronger relationship between the client and vendor, a “bilateral dependency” which TCE sees as a natural evolution of the incomplete contract under several common conditions, like “asset specificity”, which means that the asset is specialized to a particular transaction (or the two agents involved in it). Another term for this is lock-in, or the presence of high switching costs, though this way of thinking about it reintroduces the idea of a classical market for essentially comparable goods and services that TCE is designed to mitigate against. This explains how technical dependencies of an organization become baked in more or less constitutionally as part of the organization, leading to the robustness of installed base of a computing platform over time.

This ebb and flow of contract negotiation with software vendors was a bit unsettling to me when I first encountered it on the job, but I think it’s safe to say that most people working in the industry accept this as How Things Work. Perhaps it’s the continued influence of orthodox economics that makes this all seem inefficient somehow, and TCE is the right way to conceptualize things that makes better sense of reality.

But back to the Penguins…

Hoofnagle and Whittington make the case that sharing PII with a service that then personalizes its offerings to you creates a kind of bilateral dependence between service and user. They also argue that loss of privacy, due to the many possible uses of this personal information (some nefarious), is a hidden cost that can be thought of as an ex post transaction cost that is a hazard because it has not been factored into the price ex ante. The fact that this data is valuable to the platform/service for paying their production costs, which is not part of the “free” transaction, is an indication that this data is a lot more valuable than consumers think it is.

I am still on the fence about this.

I can’t get over the feeling that successfully selling a user a personalized, upgraded digital igloo is such an absurd example of a “hidden cost” that it belies the whole argument that these services have hidden costs.

Splitting hairs perhaps, it seems reasonable to say that Penguin Club has a free version, which is negotiated as one transaction. Then, conditional on the first transaction, it offers personalized igloos for real dollars. This purchase, if engaged in, would be another, different transaction, not an ex post renegotiation of the original contract with the Disney. This small difference changes the cost of the igloo from a hidden transaction cost into a normal, transparent cost. So it’s no big deal!

Does the use of PII create a bilateral dependence between Disney and the users of Penguin Club? Yes, in a sense. Any application of attention to an information service, learning how to use it and getting it to be part of your life, is in a sense a bilateral dependence with a switching cost. But there are so many other free games to play on the internet that these costs seem hardly hidden. They could just be understood as part of the game. Meanwhile, we are basically unconcerned with Disney’s “dependence” on the consumer data, because Disney can get new users easily (unless the user is a “whale”, who actual pays the company). And “dependence” Disney has on particular users is a hidden cost for Disney, not for the user, and who cares about Disney.

The cases of identity theft or job loss are strange cases that seem to have more to do with freaky data reuse than what’s going on with a particular transaction. Purpose binding notices and restrictions, which are being normed on through generalized GDPR compliance, seem adequate to deal with these cases.

So, I have two conclusions:

(1) Maybe TCE is the right lens for making an economic argument for why purpose binding restrictions are a good idea. They make transactions with platforms less incomplete, avoiding the moral hazard of ex post use of data in ways that incurs asymmetrically unknown effects on users.

(2) This TCE analysis of platforms doesn’t address the explanatorily powerful point that attention is part of the trade. In addition to being concretely what the user is “giving up” to the platform and directly explaining monetization in some circumstances, the fact that attention is “sticky” and creates some amount of asset-specific learning is a feature of the information economy more generally. Maybe it needs a closer look.

References

Hoofnagle, Chris Jay, and Jan Whittington. “Free: accounting for the costs of the internet’s most popular price.” UCLA L. Rev. 61 (2013): 606.

Data isn’t labor because using search engines is really easy

A theme I’ve heard raised in a couple places recently, including Ibarra et al. “Should We Treat Data As Labor?” and the AI Now 2018 Report, is that there is something wrong with how “data”, particularly data “produced” by people on the web, is conceptualized as part of the economy. Creating data, the argument goes, requires labor. And as the product of labor, it should be protected according to the values and practices of labor movements in the past. In particular, the current uses of data in, say, targeted advertising, social media, and search, are exploitative; the idea that consumers ‘pay’ for these services with their data is misleading and ultimately unfair to the consumer. Somehow the value created by the data should be reapportioned back to the user.

This is a sexy and popular argument among a certain subset of intellectuals who care about these things. I believe the core emotional appeal of this proposal is this: It is well known that a few well-known search engine and social media companies, namely Google and Facebook, are rich. If the value added by user data were in part returned to the users, the users, who are compared to Google and Facebook not rich, would get something they otherwise would not get. I.e., the benefits for recognizing the labor involved in creating data is redistribution of surplus to The Rest of Us.

I don’t have a problem personally with that redistributive impulse. However, I don’t think the “data is labor” argument actually makes much sense.

Why not? Well, let’s take the example of a search engine. Here is the transaction between a user and a search engine:

  • Alice types a query, “avocado toast recipes”, into the search engine. This submits data to the company computers.
  • The company computers use that data to generate a list of results that it deems relevant to that query.
  • Alice sees the results, and maybe clicks on one or two of them, if they are good, in the process of navigating to the thing she was looking for in the first place.
  • The search engine records that click as well, in order to better calibrate how to respond to others making that query.

We might forget that the search engine is providing Alice a service and isn’t just a ubiquitous part of the infrastructure we should take for granted. The search engine has provided Alice with relevant search results. What this does is (dramatically) reduce Alice’s search costs; had she tried to find the relevant URL by asking her friends, organically surfing the web, or using the library, who knows what she would have found or how long it would take her. But we would assume that Alice is using the search engine because it gets her more relevant results, faster.

It is not clear how Alice could get this thing she wants without going through the motions of typing and clicking and submitting data. These actions all seem like a bare minimum of what is necessary to conduct this kind of transaction. Similarly, when I got to a grocery store and buy vegetables, I have to get out my credit card and swipe it at the machine. This creates data–the data about my credit card transaction. But I would never advocate for recognizing my hidden labor at the credit card machine is necessary to avoid the exploitation of the credit card companies, who then use that information to go about their business. That would be insane.

Indeed, it is a principle of user interface design that the most compelling user interfaces are those that require the least effort from their users. Using search engines is really, really easy because they are designed that way. The fact that oodles of data are collected from a person without that person exerting much effort may be problematic in a lot of ways. But it’s not problematic because it’s laborious for the user; it is designed and compelling precisely because it is labor-saving. The smart home device industry has taken this even further, building voice-activated products for people who would rather not use their hands to input data. That is, if anything, less labor for the user, but more data and more processing on the automated part of the transaction. That the data is work for the company, and less work for the user, indicates that data is not the same thing as user labor.

There is a version of this argument that brings up feminism. Women’s labor, feminists point out, has long been insufficiently recognized and not properly remunerated. For example, domestic labor traditionally performed by women has been taken for granted, and emotional labor (the work of controlling ones emotions on the job), which has often been feminized, has not been taken seriously enough. This is a problem, and the social cause of recognizing women’s labor and rewarding it is, ceteris paribus, a great thing. But, and I know I’m on dicey ground here, so bear with me, this does not mean that everything that women do that they are not paid to do is unrecognized labor in the sense that is relevant for feminist critiques. Case in point, both men and women use credit cards to buy things, and make telephone calls, and drive vehicles through toll booths, and use search engines, and do any number of things that generate “data”, and in most of these cases it is not remunerated directly; but this lack of remuneration isn’t gendered. I would say, perhaps controversially, that the feminist critique does not actually apply to the general case of user generated data much at all! (Though is may apply in specific cases that I haven’t thought of.)

So in conclusion, data isn’t labor, and labor isn’t data. They are different things. We may want a better, more just, political outcome with respect to the distribution of surplus from the technology economy. But trying to get there through an analogy between data and labor is a kind of incoherent way to go about it. We should come up with a better, different way.

So what’s a better alternative? If the revenue streams of search engines are any indication, then it would seem that users “pay” for search engines through being exposed to advertising. So the “resource” that users are giving up in order to use the search engine is attention, or mental time; hence the term, attention economy.

Framing the user cost of search engines in terms of attention does not easily lend itself to an argument for economic reform. Why? Because search engines are already saving people a lot of that attention by making it so easy to look stuff up. Really the transaction looks like:

  • Alice pays some attention to Gob (the search engine).
  • Gob gives Alice some good search results back in return, and then…
  • Gob passes on some of Alice’s attention through to Bob, the advertiser, in return for money.

So Alice gives up attention but gets back search results and the advertisement. Gob gets money. Bob gets attention. The “data” that matters is not the data transmitted from Alice’s computer up to Gob. Rather, the valuable data is the data that Alice receives through her eyes: of this data, the search results are positively valued, the advertisement is negatively valued, but the value of the bundled good is net positive.

If there is something unjust about this economic situation, it has to be due to the way consumer’s attention is being managed by Gob. Interestingly, those who have studied the value of ‘free’ services in attentional terms have chalked up a substantial consumer surplus due to saved attention (Brynjolfsson and Oh, 2012) This appears to be the perspective of management scientists, who tend to be pro-business, and is not a point repeated often by legal scholars, who tend to be more litigious in outlook. For example, legal scholarship has detailed the view of how attention could be abused through digital market manipulation (Calo, 2013).

Ironically for data-as-labor theorists, the search-engine-as-liberator-of-attention argument could be read as the view that what people get from using search engines is more time, or more ability to do other things with their time. In other words, we would use a search engine instead of some other, more laborious discovery mechanism precisely because it would cost us net negative labor. That absolutely throws a wrench in any argument that the users of search engines should be rewarded on dignity of labor grounds. Instead, what’s happened is that search engines are ubiquitous because consumers have undergone a phase transition in their willingness to work to discover things, and now very happily use search engines which, on the whole, seem like a pretty good deal! (The cost of being-advertised-to is small compared to the benefits of the search results.)

If we start seeing search engines as a compelling labor-saving device rather than a exploiter of laborious clickwork, then some of the disregard consumers have for privacy on search engines becomes more understandable. People are willing to give up their data, even if they would rather not, because search engines are saving them so much time. The privacy harms that come as consequence, then, can be seen as externalities to what is essentially a healthy transaction, rather than a perverse matter of a business model that is evil to the bone.

This is, I wager, on the whole a common sense view, one that I’d momentarily forgotten because of my intellectual milieu but now am ashamed to have overlooked. It is, on the whole, far more optimistic than other attempt to characterize the zeitgeist of new technology economy.

Somehow, this rubric for understanding the digital economy appears to have fallen out of fashion. Davenport and Beck (2001) wrote a business book declaring attention to be “the new currency of business”, which if the prior analysis is correct makes more sense than data being the new currency (or oil) of business. The term appears to have originated in an article by Goldhaber (1997). Ironically, the term appears to have had no uptake in the economics literature, despite it being the key to everything! The concept was understood, however, by Herbert Simon, in 1971 (see also Terranova, 2012):

In 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.

(A digitized version of this essay, which amazingly appears to be set by a typewriter and then hand-edited (by Simon himself?) can be found here.)

This is where I bottom out–the discover that the line of thought I’ve been on all day starts with Herbert Simon, that the sciences of the artificial are not new, they are just forgotten (because of the glut of other information), and exhaustingly hyped. The attention economy discovered by Simon explains why each year we are surrounded with new theories about how to organize ourselves with technology, when perhaps the wisest perspectives on these topics are ones that will not hype themselves because their authors cannot tweet from the grave.

References

Arrieta-Ibarra, Imanol, et al. “Should We Treat Data as Labor? Moving beyond” Free”.” AEA Papers and Proceedings. Vol. 108. 2018.

Brynjolfsson, Erik, and JooHee Oh. “The attention economy: measuring the value of free digital services on the Internet.” (2012).

Calo, Ryan. “Digital market manipulation.” Geo. Wash. L. Rev. 82 (2013): 995.

Davenport, Thomas H., and John C. Beck. The attention economy: Understanding the new currency of business. Harvard Business Press, 2001.

Goldhaber, Michael H. “The attention economy and the net.” First Monday 2.4 (1997).

Simon, Herbert A. “Designing organizations for an information-rich world.” (1971): 37-72.

Terranova, Tiziana. “Attention, economy and the brain.” Culture Machine 13 (2012).

the make or buy decision (TCE) in the software and cybersecurity

The paradigmatic case of transaction cost economics (TCE) is the make-or-buy decision. A firm, F, needs something, C. Do they make it in-house or do they buy it from somewhere else?

If the firm makes it in-house, they will incur some bureaucratic overhead costs in addition to the costs of production. But they will also be able to specialize C for their purposes. They can institute their own internal quality controls. And so on.

If the firm buys it on the open market from some other firm, say, G, they don’t pay the overhead costs. They do lose the benefits of specialization, and the quality controls are only those based on economic competitive pressure on suppliers.

There is an intermediate option, which is a contract between F and G which establishes an ongoing relationship between the two firms. This contract creates a field in which C can be specialized for F, and there can be assurances of quality, while the overhead is distributed efficiently between F and G.

This situation is both extremely common in business practice and not well handled by neoclassical, orthodox economics. It’s the case that TCE is tremendously preoccupied with.


My background and research is in the software industry, which is rife with cases like these.

Developers are constantly faced with a decision to make-or-buy software components. In principle, they can developer any component themselves. In practice, this is rarely cost-effective.

In software, open source software components are a prevalent solution to this problem. This can be thought of as a very strange market where all the prices are zero. The most popular open source libraries are very generic , having little “asset specificity” in TCE terms.

The lack of contract between developers and open source components/communities is sometimes seen as a source of hazard in using open source components. The recent event-stream hack, where an upstream component was injected with malicious code by a developer who had taken over maintaining the package, illustrates the problems of outsourcing technical dependencies without a contract. In this case, the quality problem is manifest as a supply chain cybersecurity problem.

In Williamson’s analysis, these kinds of hazards are what drive firms away from purchasing on spot markets and towards contracting or in-house development. In practice, the role of open source support companies fills the role of being a responsible entity G that firm F can build a relationship with.

Williamson on four injunctions for good economics

Williamson (2008) (pdf) concludes with a description of four injunctions for doing good economics, which I will quote verbatim.

Robert Solow’s prescription for doing good economics is set out in three injunctions: keep it simple; get it right; make it plausible (2001, p. 111). Keeping it simple entails stripping away the inessentials and going for the main case (the jugular). Getting it right “includes translating economic concepts into accurate
mathematics (or diagrams, or words) and making sure that further logical operations are correctly performed and verified” (Solow, 2001, p. 112). Making it plausible entails describing human actors in (reasonably) veridical ways and maintaining meaningful contact with the phenomena of interest (contractual or otherwise).

To this, moreover, I would add a fourth injunction: derive refutable implications to which the relevant (often microanalytic) data are brought to bear. Nicholas Georgescu-Roegen has a felicitous way of putting it: “The purpose of science in general is not prediction, but knowledge for its own sake,” yet prediction is “the touchstone of scientific knowledge” (1971, p. 37).

Why the fourth injunction? This is necessitated by the need to choose among alternative theories that purport to deal with the same phenomenon—say vertical integration—and (more or less) satisfy the first three injunctions. Thus assume that all of the models are tractable, that the logic of each hangs together, and that agreement cannot be reached as to what constitutes veridicality and meaningful contact with the phenomena. Does each candidate theory then have equal claimsfor our attention? Or should we be more demanding? This is where refutable implications and empirical testing come in: ask each would-be theory to stand up and be counted.

Why more economists are not insistent upon deriving refutable implications and submitting these to empirical tests is a puzzle. One possibility is that the world of theory is set apart and has a life of its own. A second possibility is that some economists do not agree that refutable implications and testing are
important. Another is that some theories are truly fanciful and their protagonists would be discomfited by disclosure. A fourth is that the refutable implications of favored theories are contradicted by the data. And perhaps there are still other reasons. Be that as it may, a multiplicity of theories, some of which are
vacuous and others of which are fanciful, is an embarrassment to the pragmatically oriented members of the tribe. Among this subset, insistence upon the fourth injunction—derive refutable implications and submit these to the data—is growing.

References

Williamson, Oliver E. “Transaction cost economics.” Handbook of new institutional economics. Springer, Berlin, Heidelberg, 2008. 41-65.

Discovering transaction cost economics (TCE)

I’m in the process of discovering transaction cost economics (TCE), the branch of economics devoted to the study of transaction costs, which include bargaining and search costs. Oliver Williamson, who is a professor at UC Berkeley, won the Nobel Prize for his work on TCE in 2009. I’m starting with the Williamson, 2008 article (in the References) which seems like a late-stage overview of what is a large body of work.

Personally, this is yet another time when I’ve discovered that the answers or proper theoretical language for understanding something I am struggling with has simply been Somewhere Else all alone. Delight and frustration are pretty much evening each other out at this point.

Why is TCE so critical (to me)?

  • I think the real story about how the Internet and AI have changed things, which is the topic constantly reiterated in so many policy and HCI studies about platforms, is that they reduced search costs. However, it’s hard to make the case for that without a respectable theorization of search costs and how they matter to the economy. This, I think, what transaction cost economics are about.
  • You may recall I wrote my doctoral dissertation about “data economics” on the presumption (which was, truly, presumptuous) that a proper treatment of the role of data in the economy had not yet been done. This was due mainly to the deficiencies of the discussion of information in neoclassical economic theory. But perhaps I was a fool, because it may be that this missing-link work on information economics has been in transaction cost economics all along! Interestingly, Pat Bajari, who is Chief Economist at Amazon, has done some TCE work, suggesting that like Hal Varian’s economics, this is stuff that actually works in a business context, which is more or less the epistemic standard you want economics to meet. (I would argue that economics should be seen, foremost, as a discipline of social engineering.)
  • A whole other line of research I’ve worked on over the years has been trying to understand the software supply chain, especially with respect to open source software (Benthall 2016; Benthall, 2017). That’s a tricky topic because the idea of “supply” and “chain” in that domain are both highly metaphorical and essentially inaccurate. Yet there are clearly profound questions about the relationships between sociotechnical organizations, their internal and external complexity, and so on to be found there, along with (and this is really what’s exciting about it) ample empirical basis to support arguments about it, just by the nature of it. Well, it turns out that the paradigmatic case for transaction cost economics is vertical integration, or the “make-or-buy” decision wherein a firm decides to (A) purchase it from an open market, (D) produce something in-house, or (C) (and this is the case that transaction cost economics really tries to develop) engage with the supplier in a contract which creates an ongoing and secure relationship between them. Labor contracts are all, for reasons that I may go into later, of this (C) kind.

So, here comes TCE, with its firm roots in organization theory, Hayekian theories of the market, Coase’s and other theories of the firm, and firm emphasis on the supply chain relation between sociotechnical organizations. And I HAVEN’T STUDIED IT. There is even solid work on its relation to privacy done by Whittington and Hoofnagle (2011; 2013). How did I not know about this? Again, if I were not so delighted, I would be livid.

Please expect a long series of posts as I read through the literature on TCE and try to apply it to various cases of interest.

References

Benthall, S. (2017) Assessing Software Supply Chain Risk Using Public Data. IEEE STC 2017 Software Technology Conference.

Benthall, S., Pinney, T., Herz, J., Plummer, K. (2016) An Ecological Approach to Software Supply Chain Risk Management. Proceedings of the 15th Python in Science Conference. p. 136-142. Ed. Sebastian Benthall and Scott Rostrup.

Hoofnagle, Chris Jay, and Jan Whittington. “Free: accounting for the costs of the internet’s most popular price.” UCLA L. Rev. 61 (2013): 606.

Whittington, Jan, and Chris Jay Hoofnagle. “Unpacking Privacy’s Price.” NCL Rev. 90 (2011): 1327.

Williamson, Oliver E. “Transaction cost economics.” Handbook of new institutional economics. Springer, Berlin, Heidelberg, 2008. 41-65.