If the levers exist, why have they not been pulled?
This is the third essay about The Man in the Reading Room. The first essay described the diagnosis. the second offers ideas for the prescription. This article focuses on the realities of The Conditions of Action.
The first essay is here: (The Man in the Reading Room – Part I – Will He Be Right Again?).
The second essay is here: (The Man in the Reading Room – Part II – After the Diagnosis)
* * * * *
The Conditions of Action
Why It Was Possible Then
The New Deal was not the natural product of American politics.
Six conditions are worth naming.
A forcing condition that no political coalition could ignore
The crash of 1929 and the three years that followed produced a crisis whose magnitude no living American had seen. Industrial production fell by nearly half. GDP per capita fell from $858 in 1929 to $455 in 1933. Unemployment reached 24.9 percent. Roughly nine thousand banks failed. The political space for ordinary inaction had collapsed because ordinary inaction had fallen alongside it. Bread lines formed in cities that had been the wonders of the industrial world a decade earlier. Denial was not a viable political posture.
An electoral mandate of historic scale
The 1932 election produced a Democratic majority of historic dimensions: a 313-to-117 margin in the House, a fifty-nine-to-thirty-seven majority in the Senate, the presidency, and the support of both the urban-industrial North and the agrarian South. The 1934 midterm expanded those majorities further, an outcome unprecedented for the party in power. The coalition’s internal contradictions were real, but the shared experience of the crisis held them together. A working majority for action existed.
Institutional capacity capable of governing
The federal government of 1933, though small by later standards, was capable of conceiving, drafting, and administering large new programs. The professional civil service, the regulatory commissions inherited from the Progressive Era, and the universities that supplied policy talent were intact. Roosevelt could appoint a Frances Perkins, a Marriner Eccles, a Robert Jackson, and they would arrive ready to govern. The agencies that had to be built new—the Securities and Exchange Commission, the National Labor Relations Board, the Social Security Board—were built fast and staffed fresh, not yet enmeshed with the industries they regulated.
A judiciary that, by 1937, accommodated the architecture
The early New Deal faced real constitutional resistance. The Supreme Court struck down the National Industrial Recovery Act in 1935 and the Agricultural Adjustment Act in 1936. But the resistance ended. The 1937 “switch in time,” the Court’s decision in West Coast Hotel v. Parrish, upholding minimum-wage legislation, followed by sustaining decisions on the Wagner Act and Social Security Act, settled the constitutional question. The architecture was permitted to operate.
A labor movement organized as a political force
Union density stood at about 11 percent of the workforce in 1933, but it was rising sharply, and it would peak at approximately 35 percent in 1954. The political organization of working-class voters in the industrial cities was the spine of the New Deal coalition. Labor supplied votes, money, and street-level mobilization. The Wagner Act and the Fair Labor Standards Act did not pass against the political weight of organized labor. They passed because organized labor had political weight.
An information environment with shared factual ground
In 1933, the country had three or four wire services, a few national radio networks, and a press corps that, while partisan, operated within a shared set of factual conventions. Roosevelt’s first fireside chat, on March 12, 1933, reached an estimated sixty million listeners, roughly half the adult population at that time.
A single voice could address the country and be heard. This is not nostalgia. It is a description of an information environment in which a citizen, regardless of political disposition, encountered the same basic facts as the next citizen.
Each of these six conditions, taken alone, is unremarkable. What was remarkable was the simultaneity. The forcing condition produced the political coalition. The political coalition could draw on existing institutional capacity. The judiciary, after 1937, accommodated this. The labor movement supplied the political force. The information environment allowed the case to be made.
The changes from 1933 to 1938 happened because, for those five years, all of this was true at once.
It is also worth saying what the five years were preceded by. The New Deal did not begin in 1933. It was preceded by more than a decade of intellectual preparation: the Pujo Committee’s 1912 documentation of financial concentration, the Brandeis briefs, the La Follette investigations, the Progressive-era statutes that established the regulatory model, and the academic and institutional work of the 1920s on banking reform and unemployment insurance. The five-year sprint of legislation was the visible portion of a far longer process.
Why It Is Harder Now
The conditions of 2026 differ from those of 1933 in ways that matter. Five differences are worth naming, with the third, the legislative environment proper, taking the most space because it is where the other four converge.
The absent forcing condition
No present crisis has produced the political space that 1929 to 1933 did. The 2008 financial crisis was real and severe, but the federal response—TARP, the Federal Reserve’s expansion of its balance sheet, and Dodd-Frank—was sufficient to prevent a second Depression and, therefore, the political conditions it would have produced.
The 2020 COVID intervention, on the same principle, transferred trillions to households and asset markets and prevented the deeper labor-market collapse that would have generated political pressure for structural reform.
- The case that the difficulty is decisive: without a forcing condition of comparable magnitude, the political space for major statutory reform does not open.
- The case that the difficulty is overstated: the patterns Part I documents have themselves become a slow-motion forcing condition, visible to a citizenry that, on present polling, no longer requires bread lines to recognize structural failure.
Both have merit.
What is not in dispute is that slow decline does not produce political unanimity the way sudden collapse does. The adaptive state has, since 1933, demonstrated a capacity to absorb crises that would once have produced political reform. That capacity is itself a feature of the present difficulty.
The fragmented political coalition
In 2026, neither party commands a majority of the scale that the 1933 Democratic Party commanded. No president since Lyndon Johnson in 1965 has held a durable, ideologically usable congressional majority comparable to Roosevelt’s or Johnson’s for purposes of structural economic reform. Also, the working coalitions that would be required to enact any of Part II’s levers cross the present partisan boundaries in ways that neither party’s leadership has shown sustained interest in pursuing.
- The case that the difficulty is decisive: sixty-vote Senate majorities for structural economic reform are not constructible from the present alignment of voters, and the partisan polarization that produces this state of affairs is not a passing weather pattern but a structural feature of the present moment.
- The case that the difficulty is overstated: trans-partisan coalitions on antitrust have already formed—the Department of Justice’s monopoly cases against Google and Apple, the Federal Trade Commission’s challenges to Meta and Amazon, all of which have drawn at least some bipartisan support—and the political incoherence of the present moment is not stable.
Both have merit.
The legislative environment, properly understood
This is the densest of the five differences and requires the most care. It is where the four others converge into procedural reality—the absence of a forcing condition, the fragmentation of the political coalition, the changed disposition of the judiciary, the asymmetry of organized power.
Their convergence has a name: the legislative environment. The question is not abstract. It is whether a Congress operating under present rules and doctrine could, in any plausible scenario, enact a statute carrying the structural weight of Glass-Steagall, the Sherman Act, or the Wagner Act.
Three procedural features make this harder to achieve than before.
- First, the Senate cloture rule. Since 1975, ending debate on ordinary legislation has required sixty votes. The 1933 sequence faced no comparable threshold: the cloture rule was reformed in 1975, the filibuster was used sparingly throughout most of the twentieth century, and the New Deal statutes were enacted before the filibuster’s modern weaponization.
In 2026, any reform substantial enough to address the patterns Part I documents must take one of two paths. It must secure sixty votes—a threshold neither party has approached on a contested economic question for many years. Or it must fit the narrow channel of budget reconciliation, which is restricted to provisions affecting spending, revenues, or the debt limit.
Reconciliation can produce tax legislation and entitlement changes. It cannot produce a Glass-Steagall, a Sherman Act, or a Wagner Act. Those are regulatory architectures, not budget items.
- Second, the changed disposition of the federal judiciary. The Supreme Court’s 2024 decision in Loper Bright Enterprises v. Raimondo overruled Chevron deference, the doctrine under which courts, for forty years, had deferred to reasonable agency interpretations of ambiguous statutes.
The Court’s articulation of the major-questions doctrine in West Virginia v. EPA in 2022 elevated judicial review of agency actions of “vast economic and political significance.” SEC v. Jarkesy, in 2024, narrowed the ability of regulatory agencies to impose civil penalties through in-house adjudication. Each decision alone is a contestable matter of constitutional doctrine. Taken together, they alter what regulatory architecture can accomplish without explicit statutory authorization.
The 1933 Congress could write the Wagner Act in eight pages and trust the National Labor Relations Board to interpret and enforce it. A 2026 Congress writing the same statute would face a judicial environment in which broad delegations to a new agency are read narrowly, the agency’s interpretations are reviewed without deference, and the agency’s enforcement procedures are themselves subject to challenge. The statute that worked in 1935 would, if drafted identically in 2026, function differently.
- Third, the role of money in elections. The Supreme Court’s 1976 decision in Buckley v. Valeo struck down expenditure limits while preserving contribution limits. Citizens United v. Federal Election Commission, in 2010, held that independent expenditures by corporations and unions cannot be banned on anti-corruption grounds. McCutcheon v. FEC, in 2014, struck down aggregate contribution limits. Arizona Free Enterprise Club v. Bennett, in 2011, invalidated triggered matching-fund mechanisms in public-financing systems.
The 1933 Congress operated under campaign-finance norms in which corporate political expenditure was subject to far broader regulation. A 2026 Congress contemplating reform of any sector—financial services, technology platforms, agricultural processing—must do so knowing that the regulated industries can deploy political expenditure at scales the trusts of 1900 could not have matched.
- The case that the difficulty is decisive: a Congress operating under sixty-vote cloture, the post-Loper Bright judicial environment, and the post-Citizens United campaign-finance environment is a Congress structurally less able to produce New Deal-scale legislation than the Congress of 1933.
- The case that the difficulty is overstated: each of these procedural features is itself the product of statute, doctrine, or interpretation that has been changed in the past and could be changed again, none of which is constitutionally entrenched. The cloture threshold is a Senate rule. Chevron was overruled, which is to say doctrine can move.
Citizens United is a constitutional decision, but the campaign-finance landscape contains state-level experiments—New York City’s small-donor matching system, in particular—that have changed donor composition meaningfully without running afoul of the doctrine.
Both have merit.
What is not in dispute is that the procedural environment of 2026 imposes design constraints on any reform that the 1933 environment did not. A modern Glass-Steagall, written today, would have to be drafted with the cloture threshold, the major-questions doctrine, and the campaign-finance environment in mind from the first draft. That is a different drafting problem than the one Carter Glass and Henry Steagall faced.
It is not necessarily an unsolvable one.
But it is different.
The asymmetry between organized capital and organized labor
Capital in 2026 is more concentrated, more politically organized, and more legitimate in the public mind than it was in 1933.
Senate Caucus Room 301. 1933. Ferdinand Pecora, a Sicilian immigrant who had left school at sixteen, questioned Charles Mitchell of the National City Bank. Mitchell was the most powerful banker in America. The questioning was patient. It was methodical. It was public. Within days, Mitchell had resigned. Within eighteen months, the executives whose names had been household names across the 1920s had become the men the country was legislating against.
The Pecora hearings of 1932 to 1934 publicly broke the prestige of the financial sector. No comparable proceeding has broken the prestige of the platform monopolies, the private equity firms, or the largest financial institutions of 2026.
Union density stood at 10 percent of the workforce in 2025, near its lowest level in a century, against the 11 percent rising trajectory of 1933. The political asymmetry between the two organized constituencies is, on the data, the inverse of what it was when the Wagner Act passed.
- The case that the difficulty is decisive: organized labor was not merely supportive of the New Deal; it was constitutive of it, and its absence as a political force of comparable scale is the central 2026 difference from 1933.
- The case that the difficulty is overstated: labor approval among the public stood at 68 percent in 2025, the highest in nearly six decades, and union organizing in service-sector industries—Amazon warehouses, Starbucks stores, the United Auto Workers’ 2023 strike—has produced visible gains that the post-1981 trajectory did not predict.
Both have merit.
The fractured information environment
In 2026, the country does not share the set of factual conventions that allowed Roosevelt’s fireside chats to reach half the adult population. The wire services and three networks have been replaced by an algorithmically mediated environment in which a citizen’s information diet is curated by platforms whose interests are not aligned with the construction of shared political understanding.
By the mid-2020s, most American adults reported getting news at least sometimes from social media; a substantial share reported getting it primarily that way.
- The case that the difficulty is decisive: a republic that cannot share facts cannot deliberate about legislation like the Wagner Act because it requires a shared understanding of what the labor market actually is.
- The case that the difficulty is overstated: every generation has overestimated the unprecedented quality of its own information environment, and the present fragmentation contains both old vices and new affordances—Substack, podcasting, longer-form independent journalism—that the broadcast era did not.
Both have merit.
In 1933, simultaneity worked in its favor. In 2026, simultaneity works against it. The five differences reinforce one another. The absence of a forcing condition reduces the urgency of the coalition. Coalition fragmentation makes the cloture threshold binding. The procedural environment makes the judicial environment more consequential. The asymmetry of organized power shapes the information environment. None of these can be addressed alone.
This is the political problem.
It is not a complaint.
It is a description.
What Would Have to Be True
The five differences named in the previous section are not constitutional features of the American republic. They are the present state of conditions that, in the historical record, have moved before.
The patterns the man in the Reading Room described before they had names receded when those conditions moved. They returned when the conditions hardened. The question is what would have to be true for them to move again.
Four conditions are worth naming, each treated with the same evenhanded discipline.
Pattern data treated as a forcing condition before the next acute crisis
The political community might choose to treat the patterns Part I documents—the wealth-concentration data, the productivity-pay gap, the financial-sector profit share, the four-firm concentration ratios—as forcing conditions in their own right before the next acute crisis arrives.
- The case for the possibility: the antitrust enforcement actions of 2024 against Google, Meta, Amazon, and Apple suggest that pattern recognition without acute crisis is at least partially in operation, and that the political coalition for such recognition crosses the present partisan boundaries.
- The case against: the historical record is overwhelmingly that political communities act on acute crises and not on accumulating pattern data, and the same adaptive capacity that has prevented acute crises since 2008 also prevents the political clarification an acute crisis would produce.
You decide.
Coalitions that cross the present partisan alignments
The political community might construct, around one or two of Part II’s levers, working coalitions that do not map onto the present partisan boundaries—coalitions assembled from populist factions of both parties, organized labor and small-business associations, and civil-society groups whose concerns do not sort cleanly along the dominant axis.
- The case for the possibility: structural antitrust already commands at least partial support across the partisan boundary, and post-2020 service-sector labor organizing has crossed traditional political lines. There’s also historical precedent—the 1933 coalition itself crossed regional and ideological boundaries that observers in 1928 would have called impossible.
- The case against: leadership of both parties has consistently preferred partisan polarization to trans-partisan coalition-building, and the institutional incentives that produce that preference are not weakening.
You decide.
A rebuilt institutional capacity to govern
The political community might invest in the administrative state to a degree sufficient to make new regulatory architecture administrable. This includes the civil service itself, the data and analytical capacity of the regulatory agencies, the legal and economic capacity of the antitrust divisions, and the technological infrastructure on which administrative legitimacy depends in a digital economy.
- The case for the possibility: the historical precedent of the New Deal, which built capacity as it did statute, and the more recent example of the post-2008 expansion of the Federal Reserve’s analytical and supervisory capacity demonstrate that the administrative state is buildable when the political will exists.
- The case against: the political constituency for civil-service expansion in 2026 is small, and the cumulative effect of decades of skepticism toward administrative governance has produced a workforce, regulatory architecture, and legal environment that would resist reinvestment.
You decide.
Repair of the conditions of shared deliberation
The political community might address, through statute or cultural practice, the fragmentation of the information environment to a degree sufficient to allow shared deliberation. This is the civic lever named in Part II, here returned to in its political-economic rather than educational dimension.
- The case for the possibility: European jurisdictions have begun to regulate platform algorithms. American state-level experiments in disclosure, civic education, and small-donor matching have meaningfully changed the local political environment, and the present fragmentation, while real, has produced its own counter-currents in long-form independent journalism and citizen-led civic associations.
- The case against: First Amendment constraints, technological complexity, and bipartisan ambivalence about regulating platform-mediated speech make federal action unlikely on a meaningful timescale, and the cultural practices that would substitute for statutory action operate, as Part II noted, on a generation rather than a fiscal year.
You decide.
None of these four conditions is impossible. None is in active development at scale. They are conditions, not predictions. Whether they obtain is, as the entire trilogy has insisted, a political question.
The Conditions of Action
The 1933-to-1938 sequence happened because the conditions of those years made it possible. The conditions of 2026 do not, today, make it possible. That is the political question.
It is tempting to read this conclusion as despair. It should not be read that way. The historical record contains political moments in which conditions changed faster than any observer expected. The 1933 coalition did not exist in 1928. The civil rights legislation of 1964 and 1965 was not on the legislative horizon in 1954. The dismantling of the New Deal architecture, when it came after 1973, was not predicted in 1965. Conditions change, and they do so in directions that the observers of any particular moment do not predict.
There is one further point worth making, and it returns to the observation at the end of Section I. The 1933 sequence was preceded by more than a decade of intellectual preparation: the Pujo Committee’s 1912 documentation of financial concentration, the Brandeis briefs, the La Follette investigations, and the academic and institutional work of the 1920s. The five-year sprint of legislation was the visible portion of a far longer process.
The intellectual preparation for the present moment, if it is occurring, is occurring now. The data has been measured. The patterns have been described. The levers have been named. The conditions have been examined. Whether this is the early stage of a longer preparation or the late stage of an already completed one cannot be answered at the moment.
Conditions are not destiny. They are the present state of arrangements. Arrangements have been made before, in directions no observer of the moment predicted. They can be made again.
- Part I diagnosed the problem.
- Part II named the levers.
- Part III identified the conditions under which the levers can be pulled.
The conditions, like the levers, are not given.
They are made.
The action is up to you.