Several of you wrote to rightly ask what came next. What I owed you, having brought you that far, was an honest accounting of the levers. I owe you FOUR articles. This the second . . . after the diagnosis.
I. The Question Left Open
The earlier essay (The Man in the Reading Room – Part I – Will He Be Right Again?) closed with what is, I think, the most important sentence in either piece of writing.
The diagnosis describes a default tendency, not an iron law.
That sentence is the hinge.
If the patterns the essay describes are inevitable, no lever matters. If they are the natural product of an unconstrained system, then constraint is possible, and what is at issue is which constraints, applied where.
The American record between 1933 and 1973 is the demonstration. The patterns receded. They receded because the political community of those decades chose to constrain them. The patterns returned after 1973 because the political community of the subsequent fifty years chose, in cumulative steps, not to.
The choices were not abstract. They were named statutes, named votes, named decisions. Glass-Steagall in 1933 separated commercial banking from investment banking. The Wagner Act in 1935 protected the right to organize. The Fair Labor Standards Act in 1938 set a minimum wage. The Sherman Act of 1890 had been the original. Each of these was a lever. Each was pulled in one direction at a particular historical moment. Each was, over time, modified or unwound.
The question for the present is whether equivalent levers, in modified form, are available now.
I think they are.
• •
II. Three Principal Levers
The First Lever: A Modern Glass-Steagall
The Banking Act of 1933 — Glass-Steagall, after its sponsors — separated by law the activities of commercial banking and investment banking. A bank that took deposits could not also speculate in securities. The law passed in the wake of the 1929 Crash, in the conviction that the speculative excesses of the late 1920s had been amplified by the conjunction of those two activities under a single roof.
The law stood for sixty-six years.
It was eroded through the 1980s and 1990s by a sequence of regulatory interpretations, and then formally repealed in 1999 by the Gramm-Leach-Bliley Act. The repeal was bipartisan. The mood of the moment was that the original separation was a relic, that the financial industry had matured beyond the need for it, that the world had changed.
Nine years later, in 2008, the financial system seized. The federal response authorized roughly $700 billion through the Troubled Asset Relief Program. The Federal Reserve’s balance sheet rose from under $900 billion to over $4 trillion. The institutions deemed too big to fail emerged from the crisis larger and more concentrated than they had entered it.
The case for restoring something like Glass-Steagall: the conjunction of deposit-taking and speculation under a single corporate roof creates the precise systemic risk the original law was written to prevent. The 2008 record is not proof that the 1999 repeal caused the crisis — the causes were many — but it is evidence that the modified structure permitted what the prior structure would not have.
The case against: the financial system of 2026 is not the financial system of 1933. Capital markets, derivatives, and global flows are now the substance of modern finance, not appendages to it. Reimposing a separation designed for an earlier industrial economy would be costly and might displace activity offshore without reducing systemic risk. Defenders of the present architecture argue that better capital requirements and improved resolution authorities — the Dodd-Frank approach — are more appropriate to the system as it actually operates.
Both cases have merit. The reader will have to decide which weighs more heavily.
What is not in serious dispute is that the financial sector’s share of all corporate profits rose from approximately 14 percent in 1980 to roughly 50 percent at its 2010 peak, and that the system has not since unwound that share. Whether that figure represents the natural maturation of a service economy, or the consequence of a missing structural separation, is the question on which the lever turns.
The Second Lever: A Structural Antitrust Standard
The Sherman Antitrust Act was passed in 1890. It was passed because Congress recognized a danger — that the trusts, the great combinations of the late nineteenth century, had grown to a scale that threatened not only commerce but republican government. Standard Oil controlled approximately 90 percent of American petroleum refining. The railroads were organized into a handful of empires. Meatpacking, sugar, tobacco, and banking each had their kings.
The Sherman Act, and the Clayton Act of 1914 that supplemented it, gave the federal government the authority to break such concentrations. Theodore Roosevelt invoked the authority. So did Taft. The breakup of Standard Oil was decided by the Supreme Court in 1911. The Alcoa case followed in the 1940s. The IBM prosecution. The AT&T divestiture. These were not edge events. They were policy.
The standard under which they were brought was a structural one. Concentration itself, beyond a certain scale, was understood to threaten the conditions of a free market and a free polity, regardless of whether consumer prices in any given quarter happened to be low.
Beginning in the 1980s, that standard was narrowed. The Chicago School argument — that antitrust law should concern itself only with consumer prices, not with concentration as such — became the operating consensus. Mergers and acquisitions that would have been blocked under the structural standard were permitted under the consumer-welfare standard, on the theory that as long as prices to consumers remained low, no harm was done.
The result, by 2026: four firms control roughly 70 percent of U.S. domestic air travel. Four firms control approximately 80 percent of U.S. beef processing. A small handful — Apple, Microsoft, Alphabet, Amazon, Meta, Nvidia — dominate the digital economy at scales unprecedented in American business history. Nvidia holds over 80 percent of the market for the chips that train artificial intelligence systems.
The case for restoring a structural standard: the framers of the Sherman Act understood concentration as a threat to a republic, not merely to consumer prices. The post-1980 consensus, by narrowing the inquiry, allowed concentrations to develop that the original act would have prevented. The recent enforcement actions — the Department of Justice’s monopoly case against Google’s search business, the Federal Trade Commission’s challenges to Meta’s acquisitions and Amazon’s marketplace conduct — represent a partial return to the older standard.
The case against: structural antitrust is a blunt instrument. Modern industries often exhibit network effects and economies of scale that produce concentration as a natural consequence of efficient operation. A standard that treats concentration as presumptively harmful will block mergers and dismantle firms whose scale produces real consumer benefits. The risk is that the cure damages the productivity gains that have made the modern economy possible.
Both cases have merit. The reader will decide.
What is not in dispute is that concentration ratios at the levels documented above are higher than at any point since the trust era of the early twentieth century, and that the post-1980 standard was a deliberate narrowing, not a discovery.
The Third Lever: Wagner-Act Protections Extended to Contingent Work
The National Labor Relations Act of 1935 — the Wagner Act — granted private-sector workers a federally protected right to organize and bargain collectively. The Fair Labor Standards Act of 1938 set a minimum wage and the forty-hour week. The Social Security Act of 1935 created the first federal pension system. Together, these statutes constituted the architecture of New Deal labor protection.
The architecture was designed for a particular form of employment. A worker had an employer. The employer had a workplace. The relationship was, in legal terms, full-time and continuing.
That model has, for a substantial share of the contemporary American workforce, eroded. By the mid-2020s, an estimated 36 percent of American workers had performed some form of independent or contingent work in the prior year — a figure drawn from McKinsey-class private surveys, broader than the Bureau of Labor Statistics’ narrower measure of contingent work as a worker’s main job. The platform-mediated categories — Uber and Lyft drivers, DoorDash couriers, Instacart shoppers, Amazon flex workers — are the most visible, but the pattern extends further: independent contractors, short-term project workers, and the various temporary and casual arrangements that have grown up alongside the conventional employment relationship.
The Wagner Act and its companions do not, as a matter of law, extend to most of these workers. The protections that defined the conditions of mid-twentieth-century American work — minimum wage guarantees, unemployment insurance, the right to organize, a path to employer-provided health coverage — are not, for these workers, present.
The case for extending the protections: the work is real work. The economic substance of the relationship between a platform and a driver is closer to the economic substance of an employment relationship than to the economic substance of an independent business. The classification under which contingent workers operate is a legal construct, not a description of what is actually happening. Extending the protections would honor the original intent of the New Deal architecture, which was to ensure that workers in the modern economy operated with a baseline of security.
The case against: the platform economy works because its costs are different from the costs of conventional employment. Imposing the full Wagner-Act framework on platform-mediated work would raise costs, reduce flexibility for workers who value it, and in some cases destroy the business models that make the work available at all. California’s Assembly Bill 5 of 2019, which attempted such an extension, was substantially reversed by Proposition 22 the following year — a reversal driven in part by the workers themselves, many of whom preferred the existing arrangement to the proposed alternative.
Both cases have merit. The reader will decide.
What is not in dispute is that the share of the workforce operating outside the New Deal architecture has grown to a level where the question can no longer be deferred as marginal.
• •
III. Two Notes in Closing
There are two further levers, less central than the three above, that any reader of the first essay may already be considering. I will address them briefly.
The Procedural Lever
It is one thing to identify a lever. It is another to ensure that future legislative and regulatory decisions are evaluated against the patterns the historical record has revealed.
The 1999 Glass-Steagall repeal was not assessed, at the time of its passage, against the financial-sector concentration data that had accumulated by that year. The Commodity Futures Modernization Act of 2000 was not assessed against derivatives growth. NAFTA in 1994 was not assessed against the manufacturing-employment loss it would produce in concentrated regions. In each case, the legislation was evaluated against narrower criteria — competitiveness, modernization, free trade — and the cumulative effects on the patterns the New Deal had been built to constrain were not, in any disciplined way, on the table.
A requirement that major financial deregulation, antitrust modification, or labor-protection rollback carry a published cumulative-impact assessment — measured against the patterns this essay and its predecessor have described — would not change the politics. But it would put the relevant data in front of the political community at the moment of decision.
Whether such a requirement is desirable depends on whether the reader believes the political community has been operating without adequate information, or merely operating on values the reader does not share. Reasonable people will differ.
The Civic Lever
The framers of the American Constitution built elaborate checks against the concentration of political power. They did not, on the whole, build checks against the concentration of private economic power. The twentieth-century regulatory apparatus was that check — embedded in statute rather than in constitutional text, and dismantlable, as it was dismantled, across decades.
A statutory check requires continuous political will to maintain. Continuous political will requires a citizenry that understands what the check is for. The civic education of an American voter in 2026 is, on the evidence, weaker on the structural questions of political economy than the civic education of an American voter in 1936 or 1966. The wealth concentration data, the productivity-pay gap, the financial-sector profit share, the four-firm concentration ratios — these are not, on the whole, part of standard civic literacy.
Whether they should be is the question on which the civic lever turns. The argument for: a republic that loses sight of its own structural conditions cannot defend them. The argument against: civic education is slow, indirect, and easily captured by whichever side of a debate is doing the educating. Both arguments carry weight.
The lever, if it is a lever at all, operates on a generation, not on a fiscal year.
• •
IV. After the Diagnosis
There is a sentence in the first essay I have come back to several times since writing it.
The default has been operating, in the United States, for roughly fifty years.
The sentence is descriptive. It is not, in itself, a call to action. It does not say that the default ought to be interrupted. It says only that the default has been operating, that the patterns Marx described in the 1860s have, since 1973, expressed themselves in the American economy with measurable accuracy, and that the political community of those fifty years has, in cumulative steps, chosen the path that allowed it.
The choice now is whether, and through which levers, to choose differently.
I have not made the case for any of the three principal levers above. I have laid each of them out, named the historical statute that originally embodied it, described what it would do, and described what the strongest argument against it is. I have done the same for the two supporting levers. The exercise is not advocacy. It is the kind of orderly description a man tries to produce when he believes the citizens he is writing for are capable of weighing the evidence themselves.
There is one personal point I will press. The question the original essay closed on was not rhetorical. Whether a free people will choose, again, to govern the economic forces in its midst. The choice exists. The levers are real. The historical record contains the precedents. The data has been measured.
The 1933-to-1938 sequence — Glass-Steagall, the Securities Acts, the Banking Act, the Wagner Act, Social Security, the Fair Labor Standards Act — was the work of five years. Not five decades. Five years. The political community of that moment chose, and the apparatus they built held for forty years before it began to be dismantled.
Whether the political community of the present moment, presented with comparable data and comparable patterns, will make a comparable choice is not for me to decide. It is not for any one observer to decide. It is for the citizens of the republic, deliberating in their assemblies and casting their votes, to decide.
The patterns are visible. The data is on the table. The levers are named.
The rest is yours.
* *
Leave a Reply