What this covers
Henry Farrell examines how the architecture of the global economy—payment systems, supply chains, semiconductor networks, and digital platforms—has become a terrain of political coercion. The conversation ranges across the mechanisms through which states and firms exploit points of dependence for leverage, from the US Treasury's unintended disruption of European auto manufacturing through sanctions on aluminum suppliers, to China's ability to discipline Hollywood over political speech, to the US government's efforts to isolate Huawei by controlling semiconductor access. Farrell's central concern is the transition from a world in which the United States largely monopolized the ability to weaponize these interdependencies to one in which China, the EU, Russia, and other actors possess both motive and means to do the same, often without the institutional knowledge to avoid collateral damage.
The discussion moves through specific domains where this shift reshapes risk and stability. On platforms, Farrell argues that the equilibrium allowing self-regulation has broken down and that market structure—where companies like Amazon and Google are both player and marketplace—makes credible internal restraint impossible. On sovereignty and institutions, he explores how the EU might reclaim influence over payment systems, why Ireland's financial regulatory strategy is unsustainable, and how credible commitment mechanisms fail when power is concentrated. On information and democracy, he distinguishes between two threats: surveillance (which he treats as inevitable and partly overstated) and active security compromise (which he treats as serious), and argues democracy itself requires robust institutional design to function under conditions of disagreement and informational attack. Throughout, Farrell interrogates received wisdom—on encryption, Bitcoin, labor, and Karl Polanyi—while grounding abstract arguments in concrete examples of how networks become weapons.
Farrell argues that the global economy's dense networks—payment systems, supply chains, platforms—have become tools of coercion ('weaponized interdependence'), and that as power shifts from US unipolarity to multipolarity, more actors will exploit these chokepoints, raising the risk of mistakes, escalation, and instability.
- States increasingly use control of supply chains (chips), payment networks (SWIFT), and platforms to coerce others across jurisdictions
- The transition from US unipolar dominance to multipolar competition means less sophisticated actors will weaponize networks they don't fully understand
- Both old equilibria (platform self-regulation, US near-monopoly on coercion) have broken down
Governments increasingly weaponize network interdependence and platform control to extend power across borders.
- The assumption that the US always cares more than the EU about sanctions enforcement is wrong: while EU-Iran trade is minor, Europe's deep energy and economic ties to Russia mean that if a post-2020 US administration pursues aggressive Russia sanctions implicating European entities, Europe would push back hard—potentially via licensing requirements that withdraw US firms' permission to operate in Europe.
“Europe depends much more heavily on Russia for energy and has a wide variety of other economic relations with Russia. I think that’s the situation in which Europe would care much more substantially than the United States”
- When the US Treasury's OFAC sanctioned Russian oligarch Oleg Deripaska's companies, it did not realize it would disrupt the entire West European car manufacturing industry, which relies on specific aluminum products only Deripaska's companies make—illustrating how actors weaponize networks they do not fully understand, forcing Treasury into a complicated unwinding.
“The US Treasury Department OFAC sanctioned Deripaska but didn’t realize that when it was sanctioning Deripaska’s companies, it was going to mess up the entire West European car manufacturing industry, which relies upon specific aluminum products that only Deripaska’s companies can create.”
- The world is transitioning from US unipolar near-monopoly on weaponized interdependence to a multipolar world (US, China, EU, India, Russia as spoiler); because newer actors are less sophisticated about avoiding collateral damage and act competitively rather than imposing order, mistakes, escalation, and minor trade wars blowing up into economic warfare will become far more likely.
“we are moving from a world of unipolarity… to a world of multipolarity. And one of the things that we can expect from that is that multipolarity tends to lead to a lot of danger of things screwing up and things going wrong.”
- Rather than building an alternative payment system, the EU's better path to reducing US exposure would be to bring SWIFT—nominally European but heavily US-influenced—under a stronger European mandate, forcing it to comply with European rather than US demands; but this requires heavier EU-level institutional power that many member states resist.
“it would probably be less about trying to create an alternative system to SWIFT than trying to bring SWIFT more under its own command.”
- The US government is attempting to block Huawei by leveraging its control over supply networks—designating Huawei via the Entity List to threaten the chip supply relationships (e.g., Qualcomm) that Huawei depends on.
“The United States has used the so-called Entity List, which is a means of effectively designating Huawei as a problematic company, to put the supply-chain relationships, which Huawei needs, under threat.”
- US telecom companies historically took bids from Huawei in order to scare Ericsson, Nokia, and other suppliers into offering better terms, while presuming they would never actually use Huawei because doing so would create too much security trouble.
“they used to effectively take bids from Huawei in order to scare Ericsson and Nokia and other companies into offering them better terms, but with the presumption that they were never going to use Huawei”
Large platforms wield structural power by controlling both marketplace and market information simultaneously.
- Platform companies like Amazon, Google, and Facebook are both market players and the marketplace itself, which makes it impossible for them to credibly guarantee to other market actors that they will not exploit their unique access to information and ability to shape market conditions—and even well-intentioned leaders cannot stop ambitious executives from pushing the envelope, potentially undermining the platform's own long-run position.
“they’re not only businesses, but they’re also effectively microeconomies in their own right. They’re both market players and they are the marketplace itself for key aspects of the information economy.”
- Businesses operate to make profits, so when there are clear political risks associated with certain actions (such as China's coercion over the Daryl Morey Hong Kong tweet or Hollywood removing the Taiwanese flag from Top Gun), large firms in particular will not be courageous—meaning outsourcing cultural and normative decisions to business yields predictable capitulation.
“when there are clear political risks associated to business doing certain kinds of things, by and large, businesses are not going to be especially courageous. And this is particularly likely to be true of big firms”
- The long-standing equilibrium in which the US persuaded everyone that platform self-regulation was acceptable has broken down in the US, Europe, and authoritarian societies alike, so we will increasingly see governments use platform companies as instruments to extend their reach into other jurisdictions and impose universal restrictions.
“That equilibrium has broken down. It has broken down in the US. It has broken down in Europe, and it has broken down in authoritarian societies in particular.”
- There is a tradeoff in tech regulation: if you prioritize security, you want to keep big companies around because only large firms can create collective security goods and impose broad standards; if you prioritize privacy and rights, you want an equilibrium with far fewer big firms—so the fight is fundamentally between people who value security and people who value privacy.
“If you value security... then the answer is, you probably want to keep big companies around... If, alternatively, you value things like privacy and other kinds of rights, then you probably want to move towards an equilibrium in which there are far, far fewer big firms.”
- Applying North and Weingast's 'tying the king's hands' theory—where the Glorious Revolution let the British monarch credibly commit not to renege on debts, becoming a source of power—Facebook fails the analogy because Zuckerberg controls the board and majority of voting shares, making it impossible for him to credibly commit to constraining his own power, especially once such constraints threaten the business model.
“this is very, very difficult for Zuckerberg or anyone who is in Facebook to really credibly commit to tie their own hands in this kind of way.”
Content moderation and technological architecture shape political discourse by filtering interlocutors and distributing speech.
- The perception that Twitter leans left may result not from the service's inherent valences but from the truncation of much of the Right from the small-l-liberal conversation under Trump—such that people committed to Popperian/Gellnerian debate now find their interlocutors disproportionately on the left, with avid Trump followers offering little unexpected to engage with.
“to what extent is that perception the results of the truncation of a big chunk of the Right out of the conversation?”
- Sites like 8chan and 4chan are 'meme factories' producing content through Darwinian competition for attention, but harmful content becomes epidemic only because larger platforms like Facebook provide the dissemination reach; in a world where everything was at 8chan's scale, the spread of toxic content across internet users would be far less, even if other problems would arise.
“you need two things for that to work. First of all, you need a process of generation, and secondly, you need some kind of process of dissemination. You need other platforms which have far greater reach”
- Philip K. Dick's strength is not politics or character but his sense of massive complex systems that don't work well and the profound unreality in his novels; he—more than Orwell, Huxley, or Zamyatin—captures our actual dystopia, which is not a state telling everyone what to think but a world where nobody knows what to believe is true and there are multiple versions of the truth, anticipating phenomena like QAnon and Trump reinforcing each other in a feedback loop.
“is really the person who captures the kind of dystopian characteristics of the society that we live in at the moment, which isn’t a society where we have a state telling everybody what to think, but it’s a world in which nobody knows quite what to believe is true anymore.”
- Beyond Hayek's account of how markets aggregate diffuse tacit knowledge through the price mechanism, democracy has its own techniques for capturing information: the clash of different ideas, perspectives, and partisanship—though messy and rancorous—carries information value that markets cannot capture, so the ideal of democracy as machinery is to extract as much information as possible from people's disagreements while keeping the peace among them.
“my ideal for democracy is, as a machinery, to try and get as much information out of those differences as possible, to take things like partisanship... but to recognize that the clash of different ideas, the clash of perspectives also has a lot of information value and does things that markets can’t capture.”
- When an idea like weaponized interdependence takes off, it effectively escapes its creators' control as people apply it to a much wider, more diffuse range of phenomena than originally intended—a plausible fate for anyone who has an idea that takes off.
“you see the idea effectively escaping your control. So people are talking about weaponized interdependence with respect to a much wider and broader and diffuse range of phenomena than what we talked about.”
No technical or political solution exists to police speech across platforms serving billions with limited resources.
- There is no satisfactory solution—politically or technically—to policing free speech on platforms: there are no universal standards, and even if there were, machine learning will never keep up with the multifarious ways humans use language to communicate nasty things, especially for companies with billions of users and limited staff.
“machine learning simply is not going to ever be able to keep up with the multifarious ways in which human beings can use language in order to communicate nasty and unpleasant things to each other.”
- Research suggests that increasing Twitter's character limit (from 140 to 280) has had consequences for perceptions of real argument happening, with a tendency toward polarization when people see the other side's actual statements, though Twitter also exposes users to interesting conversations they would not otherwise encounter.
“There has been some research which suggests that the increase in the character limit has actually had consequences for perceptions of real argument happening as opposed to people just setting out stances.”
- Because the world is fundamentally insecure and any technology will have exploitable back doors, the best approach is to accept that some surveillance is inevitable while making critical systems secure and robust through multiple redundancies—prioritizing robustness over economic efficiency.
“a world which is much less going to be about trying to pare things down to the minimum in search of economic efficiency and much more about robustness, redundancy, and trying to think carefully through the fact that we live in a fundamentally insecure world”
- Democracy is likely to be overwhelmed under circumstances where people disagree too much on basic facts and basic information, raising the question of how to make democracy less vulnerable and more robust against informational attacks—not only Russian-type attacks but also those domestic political actors use to pursue short-term aims.
“democracy is likely to be overwhelmed under circumstances where people disagree too much on basic facts, on basic information.”
Social trust depends on institutional design that manages inequality and constrains power, not diffuse optimism.
- Social trust is best understood through institutions rather than diffuse social optimism: the social production function is roughly that you need rough levels of economic inequality which allow for institutions that are reasonably fair, which in turn create the conditions for social trust.
“you need rough levels of economic inequality which allow for institutions which are reasonably fair, which in turn, create the conditions for social trust.”
- It is essentially impossible to imagine a world with protection for large segments of the population that does not roughly resemble labor unions; though they may take very different, more disaggregated forms in a platform economy, something union-like is necessary—and even economists like Acemoglu and Robinson and libertarian-leaning thinkers like Lindsey and Teles now recognize unions are needed to maintain the underlying conditions for democracy.
“I think labor unions are going to replace labor unions... it’s more or less impossible to imagine a world in which you have protection for large segments of the population that doesn’t roughly resemble labor unions.”
- Although Farrell opposes the specific EU decision forcing Facebook to remove content globally, he argues that given the choice between US non-regulation and authoritarian regulation, a robust European Union—which has both the regulatory clout and a willingness compatible with liberal norms—is the most preferable actor to bring platform companies to heel.
“I would much prefer to have a robust European Union than any of the other actors that I can think of that have the clout and the ability to try and bring the platform companies to heel.”
Economic structures accumulate drag that periodically requires cleansing, whether through regulatory intervention or preference cascades.
- There is an implicit break in the Trump administration between those (like Trump) who think in terms of trade victories and those concerned with long-term security, leading to a likely compromise where Trump offers concessions on Huawei that the defense establishment, which sees Huawei as a long-term threat, will dislike.
“there’s an implicit break between, on the one hand, people like Donald Trump who think primarily in terms of trade and some very strange ideas about what victory in trade is, and on the other hand, people who are more concerned with the long-term security relationship”
- Elizabeth Warren's economic thinking is best understood through Mancur Olson's The Rise and Decline of Nations: an Olsonian view that drag, dross, and corruption build up in markets over time, so that to allow markets to reach their full potential you periodically need to cleanse them—a law-and-economics framework turned in a different ideological direction.
“what Elizabeth Warren is pursuing is very much an Olsonian view of how markets work: that drag and dross and corruption builds up and that in order to allow markets to achieve their full potential, you basically need to cleanse them at a certain point.”
- Ireland's transformation from poor to wealthy was a creative fit into the neoliberal global power structure via a combination of a regulatory play (lax regulation of financial and platform activities) and a taxation play (exotic tax arrangements); the regulatory play kickstarted genuine native industry that will persist, but the tax game is clearly ending, leaving the Irish state with a looming fiscal headache compounded by Brexit.
“a lot of what Ireland has done over the last number of years is to figure out a combination of a regulatory play and a taxation play.”
- Italy's economic trap—an over-elaborated bloated public sector alongside a dynamic private sector stuck in a previous (manufacturing) cycle—may foreshadow China's future: China's massive state sector coexisting with dynamic small private firms cannot sustain private dynamism indefinitely without the public economy providing appropriate public goods, and China's lack of democracy may make this worse two or three decades hence.
“I wonder whether China may suffer from a version of the Italian disease two or three decades from now, and perhaps even more starkly because of the lack of democracy.”
- Ireland secularized faster than almost any country in history (under 30 years) because it was a Timur Kuran-style preference cascade: mutually reinforcing beliefs held because everyone believed everyone else believed them collapsed rapidly when a shock (a bishop revealed to have secretly fathered and funded a child for 17 years) triggered mass preference revelation; the change was fast because it was 20-25 years overdue, switching people from one equilibrium to another.
“this is a Timur Kuran story, where you have the set of beliefs and expectations which are mutually reinforcing because everybody believes that everybody else believes them. And when this collapses, the collapse can happen very, very quickly indeed.”
Technology and culture operate through evolutionary dynamics where ideas escape creators' control and vary widely in uptake.
- Bitcoin purports to be a technological hack but is really an attempt at a sociological hack—creating value out of wasted computer cycles—that has not particularly worked; it succeeded only within a community large enough to survive, but will never take off beyond that community, while its underlying technologies will be adopted where useful for financial services.
“it purports to be a technological hack. It’s really an effort at a sociological hack, and I don’t think it’s particularly worked.”
- Karl Polanyi's The Great Transformation has two key flaws: his ideal alternative is a Christian socialism (derived from Tawney) involving nostalgia for a quasi-medieval organic society that cannot be reestablished under liberal conditions; and his account of how 'society' reacts against the market via the 'double movement' relies on a vague big abstraction—'society'—that Farrell, distrusting such abstractions as social causes, finds an inadequate substitute for the real micromechanics.
“the other thing that, of course, is very, very vague and wafty in Polanyi is how it is that “society” reacts against the market... Personally, I tend not to believe in big abstractions such as society as explanations or as social causes.”
- Gene Wolfe's The Book of the New Sun is 'Proust with rockets and ray guns'—Proustian in style and concern with time, but Proust in reverse: where Proust describes modernity overtaking aristocracy, Wolfe imagines a world where modernity has disappeared, supplemented by medieval ways of organizing the world, embodying science fiction's power to make the strange familiar and the familiar strange (the Novalis sense).
“I think of him almost as being Proust in reverse. Proust is describing a world in which the modern world is overtaking aristocracy... What Wolfe does... is to take that and to reverse this and to imagine a world in which modernity has disappeared.”
- While much of Marx is flat-out wrong and the analytic Marxist project drove itself into a hole, the underlying idea of combining rationalist perspectives with a direct concern for understanding power relations in the Marxist and Weberian way is coming back to the fore and is extremely valuable.
“the basic underlying idea, which is that if you bring together rationalist perspectives with a direct concern with power relations... in the Marxist and the Weberian way, that this is something which is coming to the fore again, which is extremely valuable.”
- Crooked Timber served a catalyst function because it emerged around the start of the Iraq War when there was no organized, coherent intellectual Left in the US—places like Dissent were running on accumulated 1960s-70s energy without orienting to present circumstances—but today there is a vibrant, thriving, factional Left engaging with policy again, so there is far less need for such a blog.
“we came into being at a moment when a lot of the energy of the Left had run out. You had places like Dissent and other places which were really running on the accumulated energy of the 1960s and 1970s”
- A productive intellectual method is to read widely and play with many ideas, most of which fail, throwing out variations (e.g., as micro threads on Twitter) at a rapid rate to see which take root—an evolutionary dynamic where most variations fail but the successful ones can then succeed far better.
“You can think about it as evolutionary dynamic. You throw out variations. Most of those variations fail. Some of those variations take root, and then those variations can succeed far, far better.”
Surveillance technology poses ongoing tradeoffs between inevitable monitoring and system robustness requiring careful institutional process.
- A number of US allies have opted for Huawei because they reason they are already being surveilled by the United States and will likely be surveilled by China regardless, so the marginal surveillance cost of using Huawei is small relative to cheaper telecom infrastructure.
“there is a good reason why a number of countries, including US allies, have decided to opt for Huawei because they figure they’re already being listened into by the United States. They’re probably going to be listened into by China anyway.”
- Legal access to heavily encrypted information (a 'back door') is acceptable provided there is clear legal process around it, achievable through careful institutional design like the reformed FISA court that now includes public advocates—though it risks capture by the security state and would require shared standards among advanced industrial democracies.
“I think that it is okay to have this as long as there is clear legal process around it.”
- The threat from Huawei should be separated into two tradeoffs: surveillance (listening), which is widely overstated and unavoidable, versus the more serious risk that such technology could be linked into systems to pose an active security threat—such as causing communication breakdowns during a military attack.
“The more complicated, more difficult question is whether or not Huawei, or other kinds of technology companies like it, could be linked into other forms of systems in ways which might pose an active security threat which goes further than just listening.”