---

The Imprint and the Exit

Generational succession as the experiment societies keep running — and the first one registered in advance. A Registration Paper — the first in this archive (Armchair Scholar, the independent paper series in which this is No. 011), to the archive's knowledge. Rules in force: no prophecy, no obituary, no monocause, no cycles. Draft v5 (archive panel); the predictions become binding only when the frozen table and scoring script (Artifact A) are posted to a public archive under a timestamp no one can edit — the deposit; until then this registration is a draft.

· · ·

Society persists despite the mortality of its individual members, through processes of demographic metabolism and particularly the annual infusion of birth cohorts.

— Norman B. Ryder, "The Cohort as a Concept in the Study of Social Change," American Sociological Review 30(6), 1965, 843–861 (verbatim — the abstract's opening sentence, confirmed at source)

Abstract

Societies with continuous institutions keep running the same experiment without writing down its hypotheses. A cohort is formed by shared events in its formative window; it rises into institutional power; it holds that power while its formation shapes what the institutions do; then it leaves — on schedules only mortality finally enforces, and that officeholders' own choices substantially set along the way; the registration prices the choosing. What does the exit change? Four families of answer recur — the behavior is carried by the imprinted persons and fades as they go; by the cohort's numerical weight and fades with demography; by coalitions that recruit successors and never fades; by asset positions that transfer rather than fade — and the historical record cannot decide among them, because in every completed case the answers were composed after the exit by writers who knew the ending. Part I of this paper builds a grammar for reading generational successions — four carriers, four clocks, with stated scope conditions — and reads six completed cases through it, labeled throughout as retrodictions that establish the grammar's reach and adjudicate nothing. Part II revokes the after-the-fact privilege for one live case. The American Baby Boom cohort's exit is underway — Generation X already outnumbers Boomers in the House while the Senate remains three-fifths Boomer, and the cohort holds 51.6 percent of household wealth — and the paper registers the four accounts' predictions before the outcome arrives: a public succession ledger; twelve scored rows, three per account — window-symmetric at one dated window each, the live-bet count disclosed at 2/1/2/1 — each carrying a registered risk grade, an exhaustive-and-exclusive confounder list, and a decision rule frozen in a scoring script that is itself an imprint gate; a fifth registered column in which nothing moves on any cohort clock and the generational frame fails for these observables, the account of The Living Room and the Long Grip included as applied — a column that can also return not-evaluable, the registered outcome when the instrument or the exit's own progress cannot support adjudication; and a scoring protocol — biennial public memos, a public tribunal in the form of a frozen script anyone can run and whose outside runs supersede the archive's own, an objective shock log, an amendment rule that permits only striking, an independent timestamp proving the predictions predate the outcomes, and a continuity clause. The paper adjudicates nothing and forecasts nothing. Societies will run the experiment again. This is what it looks like to run one on the record.

PART I — THE GRAMMAR

§ IThe Experiment Societies Keep Running

Institutions outlive the people who staff them; that is what makes them institutions. But they are staffed, at every moment, by people formed somewhere else in time — shaped in a window a generation or two before the present, by wars, depressions, broadcasts, and revolutions the current moment did not choose. Ryder called the mechanism demographic metabolism: society persists by continuously replacing its members, and social change arrives substantially through that replacement rather than through persuasion. The observation is old. What has never followed from it is a method. When a formed cohort takes institutional power — and some cohorts, formed hard and born large, take a great deal of it — the society acquires a question: how much of what its institutions do is the cohort, and what happens when it leaves?

The question has been answered many times, always the same way: afterward. This paper's claim is that the answering-afterward is not a blemish but the reason the question has never been settled — and that the present decade offers, for the first time with public data at quarterly resolution, the chance to answer one in advance.

The load-bearing assumption, stated so it can be doubted. Everything below assumes that institutional behavior has cohort structure — that the birth cohort of the people who staff an institution is a meaningful unit of analysis, distinguishable in principle from party, from period, from age itself, and from the institution's reproduction of its own personnel. A long institutionalist tradition doubts exactly this, and a life-cycle tradition doubts it differently: what looks like formation may be the common aging of officeholders. The paper does not assume either doubt away; it registers them. The fifth column of Part II's table — nothing moves on any cohort clock — is the assumption's failure given standing beside the hypotheses built on it, with a consequence scoped honestly in advance: if the fifth column scores, the generational frame failed for these observables, in this case — and the account this archive argued in The Living Room and the Long Grip (its earlier paper on the Vietnam-broadcast imprint) fails with it, as applied.

§ IITwo Caricatures

Every argument about generational power is conducted between two caricatures, in every era that has the argument.

The Obituary holds that the cohort is the politics: wait out the old men and the institutions exhale. Its tell is the actuarial sneer — the mortality table mistaken for a political program. It fails morally, converting a living generation into an obstruction with a date; this paper's second rule exists because the genre rewards exactly that. And it fails empirically, because it predicts everything and prices nothing: which behaviors fade, on which schedule, in which venues — the Obituary never says.

The Iron Law holds that nothing changes: institutions reproduce themselves, coalitions recruit successors shaped to fit, and cohort talk is costume jewelry on the same machine. It survives every observation — change is the machine adapting, persistence is the machine persisting — and so it, too, prices nothing.

Between the caricatures stands the classical prior for this paper's whole question: Pareto's circulation of elites (1901/1968; Bottomore 1993 for the modern survey) — elites always circulate, and what replacement changes is the analytical question, not the assumed answer. The four accounts of §3 are rival specifications of what circulates.

Both caricatures are unpriced bets. The four accounts of §3 are the caricatures grown up: real research programs, real evidence, and — held to this paper's standard — real prices.

§ IIIThe Four Carriers

Ask what carries the behavior and the literatures sort into four families, each with a carrier, a decay driver, and a clock. Stating them this way is Part I's contribution — with its lineage owned rather than implied: the composition-versus-conversion decomposition underneath it is Firebaugh's (1989), the cohort-replacement partition is half a century old, and the carrier question has been run on Congress itself once before (Born, 1979, tracing an institutional change to entering cohorts rather than sitting members). What those literatures do not state is the four-way carrier partition an exit interrogates, with scope conditions under which it contracts — and that is the thing stated here, because it is the carrier that an exit interrogates.

A1 — Imprint. Carrier: the formed persons. Mannheim's generation-units, given quantitative form in the generations-of-voting literature: political events weighted by age at exposure, weights peaking in late adolescence (Ghitza, Gelman and Auerbach, AJPS 67(3), 2023 — generational evidence strongest for white voters, a limit that travels with every use of this authority) — with direct longitudinal persistence evidence in the 1965–1997 Youth–Parent Socialization Panel (Jennings & Niemi 1974, 1981; Jennings, Stoker & Bowers, JOP 2009), and one caveat the paper's own authority imposes: the model's formation boundaries do not respect the Census cohort — its peak formation years (1952, pro-Democratic; 1968, pro-Republican) fall inside and just past the 1946–64 span, so imprint-generations and Census bins are distinct objects, a tension §9 registers rather than hides. Decay driver: the persons' exit from position. Clock: office — retirement and mortality, chamber by chamber, court by court. Signature: behavior tracks who holds office, not how many cohort members are alive, vote, or own.

A2 — Size. Carrier: numerical weight. The Easterlin lineage (Birth and Fortune, 1980: a cohort's fortunes vary inversely with its size) and its political descendants (Munger, Generation Gap, 2022: size plus cohort consciousness as ballast). Decay driver: demography of the living cohort. Clock: mortality and electorate composition. Signature: behavior tracks population and electorate share; office turnover without demographic decline changes nothing.

A3 — Sorting. Carrier: coalitions that outlive their members. Parties recruit successors shaped to fit; the behavior belongs to the coalition, not the cohort — and the recruitment has named evidence in the same panels that support A1: family transmission of partisanship (Jennings, Stoker & Bowers 2009) is one way coalitions inherit members. Decay driver: none intrinsic. Clock: none. Signature: no fade on any cohort schedule — an account that survives nulls, which is why any honest test must bind it so inertia can never silently score in its favor. Part II does.

A4 — Wealth structure. Carrier: asset positions. Wealth does not fade; it transfers — to heirs and fiduciaries, on bequest schedules. Decay driver: none; migration instead. Clock: the transfer curve. Signature: wealth-linked behavior concentrates in successor hands while decoupled from every office and electorate series.

Scope conditions, stated rather than discovered. The grammar is not culture-free. A3 presupposes durable party-like coalitions; A4 presupposes heritable private capital. Where a setting lacks one or both — single-party states, revolutionary property regimes — the grammar contracts to the carriers the setting supports, and the honest reading says so rather than stretching. This bounds every general claim in the paper: the grammar is a reading instrument wherever its carriers exist, not a law of history anywhere. Nor do the four carriers exhaust the space: cohort effects can themselves be endogenous to predecessors' policy choices (Levy & Razin, 2025) — a fifth candidate carrier the grammar registers as outside its partition rather than silently absorbing.

§ IVThe Record: Six Completed Successions, Read as Reach

Everything here is retrodiction, labeled. The cases establish three things only: formative-cohort grips recur across societies and centuries; the observable trajectories of a single succession demonstrably diverge — some series fall on the cohort's schedule while others persist beyond it; and that divergence is the historical motivation for registering predictions domain by domain. The cases attribute no trajectory to any carrier — §4's own closing paragraph explains why they cannot — and prove no account.

The Founders' exit (United States, to 1825). The revolutionary cohort held the presidency for the office's first thirty-six years — every holder through Monroe was of the founding generation — and the first post-founder succession arrived as the contested, four-way, House-decided transition of 1824–25, followed within a decade by a rebuilt party system under new alignments. Trajectories diverged: a politics of founding-era deference ended in the same decade the founding persons did, while party machinery was rebuilt almost immediately — two trajectories, stated as trajectories; which carrier drove either is exactly what this section's closing paragraph says the record cannot supply. (The standard survey of the 1824–25 transition: Howe, What Hath God Wrought, Oxford, 2007.)

The Civil War generation (United States, 1865–1910s). Veterans of the Union dominated the presidency from Grant through McKinley — with the two exceptions named, as the parallel construction demands: Cleveland, who hired a substitute under the Conscription Act, and Arthur, whose service was militia staff — and around the cohort grew the era's great domestic program: the Civil War pension system, which at its early-1890s peak consumed on the order of forty percent of federal receipts, and which by 1910 covered roughly eighteen percent of all Americans over sixty-five (Skocpol, Protecting Soldiers and Mothers, 1992; page-cites at imprint, with the coverage denominator — all persons versus men over sixty-five — settled there). The pension apparatus contracted on the cohort's actuarial schedule — with a long transfer tail, stated as a trajectory: benefits ran to widows for another century, and last to a veteran's disabled daughter, until 2020 — while the same era's party system, the Civil War alignment itself, persisted decades beyond the veterans. One succession; trajectories that diverged by domain. Which carrier drove which trajectory, the record cannot say — that is the closing paragraph's point, and the design requirement it produces.

The front generation (Europe, interwar). The cohort formed in the trenches carried its formation into every interwar politics at once — Wohl's The Generation of 1914 (Harvard, 1979) is the standard account — and into fascist, communist, and democratic variants simultaneously: the standing caution that imprint fixes salience, not direction. Carried as a scope lesson for A1, at conceptual grade.

The GI generation (United States, 1961–1993). Every president from Kennedy through Bush was formed in the Second World War era — the phrasing is deliberate; the edge case is Carter, who entered Annapolis in 1943 under the accelerated wartime program and graduated with the Class of 1947 in June 1946, formed by the wartime academy but not the war. With the cohort in office came a distinctive reasoning style: the Munich analogy, whose grip on the 1965 escalation decisions Khong documented (Analogies at War, Princeton, 1992). Whether the analogy's authority faded with the cohort is — the paper says this against its own earlier draft — a labeled conjecture, not a finding: the analogy revived fluently in the 2002–03 Iraq debate, at Syria in 2013, and again over Ukraine, and no corpus test of its trajectory is registered here. Part II struck rhetoric observables for exactly this undecidability; Part I claims no exemption.

The two socialist gerontocracies (USSR 1966–85; China 1978–97). The record's most instructive paired comparison — nothing in this record is controlled: the same kind of cohort — a revolutionary founding generation — exited two opposite ways. The Politburo deferred succession to actuarial limits (average age fifty-five in 1966, sixty-eight by 1982 — the era anatomized in Beissinger, "In Search of Generations in Soviet Politics," World Politics 38(2), 1986), and the corrections of 1985 arrived all at once. China engineered the alternative: from 1978, an age-based exit norm was constructed from above and millions of revolutionary-generation cadres were retired — in the phrase attached to Manion's study — Paretian on its face — the largest bloodless circulation of elites in history (Retirement of Revolutionaries in China, Princeton, 1993; the phrase is the Princeton edition's characterization). A third completed variant now exists: Cuba's scheduled handoff, 2018–2021 — the founding cohort's exit by announced term limit, completed when Raúl Castro left the party leadership at the Eighth Congress (NBC News, 2021) — a constitutionalized exit sitting between Moscow's freeze and Beijing's program, read through the charismatic-succession frame Hoffmann built for the transition's first phase (International Political Science Review 30(3):229–248, 2009 — a source that predates the completion it frames, cited for the frame alone). The case also exercises §3's scope conditions as written: without competitive parties or heritable private capital, the grammar contracts to its first two carriers, and says so. The set establishes that exit management is a variable, not a constant — deferred rupture and engineered metabolism bracket a range, and the American case sits inside it as unmanaged drift. It also corrects a phrase this paper's own first draft used: retirement schedules are not beyond politics — Beijing wrote them by policy. Only mortality is.

Italy, 1992–94. Near-total elite replacement inside three years (Cotta & Verzichelli, 2007) — and institutional behavior that changed far less than the replacement promised. The standing warning, attached to every window in Part II: even total succession can deliver less than its celebrants expect.

What the six cases cannot do — the hinge. Each case above was stated as trajectories, not attributions, because attribution is exactly what retrodiction cannot supply. The pension state "obviously" faded with the persons — but the veteran cohort's electoral weight collapsed on the same schedule, and the pension flow was the cohort's asset: three carriers, one trajectory, no referee. The party system "obviously" persisted like a coalition — or like a wealth structure, or like the period's sectional geography. Every carrier moved together, and the reading was chosen after the ending was known. That is not a counsel of despair; it is the design requirement for Part II. The six cases establish that the grammar has reach — the carriers are real, the trajectories diverge, domains split. Only a registered succession can establish which carrier does the work, because only registration removes the privilege that has made every generational theory — this archive's own The Living Room and the Long Grip included — unfalsifiable in practice.

§ VWhy the Past Never Decides — the Fence, and What Part I Stakes

The identification degeneracy. Age, period, and cohort are linearly dependent, and the modern identification literature is precise about what that does and does not forbid: the linear components cannot be separated by any amount of data — algebraic "solutions" merely relocate the assumption — while nonlinear components are identified, and partial-identification bounds can recover much of the rest (Fosse & Winship, 2019a, 2019b; Bell & Jones, 2013; Luo, 2013). The degeneracy, so stated, binds this paper's own instrument, not only its rivals'. Within any single Congress, a member's birth cohort is their age; a cross-sectional "cohort effect" is an age effect wearing a method's clothes. Part II's scoring design exists because of this sentence: the registered quantity is an age-adjusted cohort increment, identified by pooling Congresses so that same-age comparisons across time track the 1946–64 boundary as it moves through the age distribution — with the residual identifying assumption named in §9 as an assumption. The move has a published nearest neighbor and a standing adversary: Ahlfeldt, Maennig and Mueller separate age from cohort across three hundred Swiss referenda — identifying age under arbitrary cohort effects, the mirror of this design — and find the observed generation gap driven substantially by aging (European Journal of Political Economy, 2022): the strongest evidence in the record that this paper's fifth column is live. The design answers it in the instrument rather than the prose: every scored decision must replicate across a registered family of age specifications, linear through spline; a placebo cohort boundary shifted six years must stay silent; and the age-adjusted increment ships beside the assumption-free nonlinear components (Fosse & Winship, 2019a) so anyone can run the bounds the archive does not (Artifact A, codebook §§3, 5). A registered design that merely named its assumption would be a disclosure; this one carries its own placebo. The evidential moment is when the carriers' clocks decouple; for the American case, that is now.

The after-the-fact privilege. Every generational theory in the record was composed by someone who knew how the succession ended. The consequence, unpriced by the field: theories built on completed successions can always be rebuilt to fit the next one, and so the argument is in its fifth decade without a casualty. The registration form — hypotheses filed before outcomes, frozen, scored in public (Nosek et al., PNAS 115(11), 2018) — exists to revoke the privilege; Part II transposes it, cost by cost. And the form has a political lineage the paper claims membership in, not priority over: Tetlock's forecasting program (Expert Political Judgment, 2005) and the tournament tradition it spawned (Tetlock & Gardner, 2015); the Social Science Prediction Platform, timestamping researchers' predictions before results (DellaVigna & Vivalt, from 2020); the Forecasting Collaborative's pre-registered societal forecasts (Nature Human Behaviour, 2023); the Long Now Foundation's public bets (2002–); and — from the cycle-adjacent program this paper fences against — Turchin's registered 2010 forecast of 2010s instability and his own 2020 retrospective scoring of it, named here precisely because the fence is against unpriced cycles, not registered ones. And the field has made forward, gradeable replacement claims before: Inglehart's postmaterialism program is a cohort-replacement forecast scored in public across four decades (Inglehart, 1977; Inglehart & Abramson, 1994); Rekker's twenty-one-country cohort analysis projects party-system change through Boomer replacement (Rekker, 2024); and The Emerging Democratic Majority is what an unregistered succession prediction looks like when it loses — its own authors scored it, honorably, two decades on (Judis & Teixeira, 2002). What none of these registered is a succession: rival carrier accounts bound in advance to one cohort's exit, window-symmetric with unequal stakes disclosed, the frame's own failure a scored outcome. The priority claim is that narrow, and made to the archive's knowledge.

The fence. A general grammar of succession sits one shelf from the cycle theories — turnings, archetypes, eighty-year wheels — and the adjacency is fatal if unfenced, because cycle theory is the after-the-fact privilege industrialized. The fence: this paper asserts no periodicity, no archetypes, no recurring generational types, no prophecy. The grammar reads any succession; it forecasts none; its only forward-looking claims are the twelve registered rows.

And what Part I itself stakes — because a fence is not an alibi. A framework whose every claim survives every outcome is the Iron Law in academic dress, and this paper declines the costume. Three of Part I's claims are exposed, by name, to Part II's results. If the fifth column scores, the hinge claim — that the accounts are separable at a watched exit — failed in the only case it was ever tested, and the grammar's advertised reach shrinks accordingly. If sorting sweeps the table while the ledger confirms the exit proceeding, the scope conditions were the story all along and the grammar demotes to a redescription of coalition politics — Part I's §3 would need rewriting, not reciting. And if no cross-series decoupling appears in the live case — if the clocks refuse to separate even now — then §4's "reach" was overread and the retrodictive divergence of trajectories was period all the way down. Part I cannot be proven by Part II; it can be damaged by it, and the paper says where.

PART II — THE REGISTRATION

§ VIThe Live Case

The American Baby Boom cohort — born 1946–1964; the oldest turn eighty this year — is the first formative-cohort grip whose exit can be watched in public data at quarterly resolution. This archive's The Living Room and the Long Grip argued one account of the grip: the Vietnam broadcast imprint. It called its own account "plausible and undervalued," which is the right grade for a retrodiction and the wrong place to stop, and its stated argument — quoted exactly, from the dek its revised edition still carries — is the claim this paper takes at its word: "The grip is best explained generationally rather than ideologically, and the paper expects it to loosen by succession rather than argument — an expectation it closes Section III by exposing to named tests with dates." The four accounts, instantiated for this case, disagree about the exit now underway; for the first time, the disagreement can be registered rather than narrated. The tradition studied this cohort's entry — Delli Carpini took the generation of the 1960s into the electorate (Stability and Change in American Politics, 1986), Rintala fixed the vocabulary (1963); the exit as a registered succession closes a symmetry the literature opened — the aging-in-power side has current shelves (Munger, 2022; McClean, 2024; Stockemer & Sundström, 2025; van der Brug & Kritzinger, 2026; Moyn, 2026), but none binds rival carriers to the exit in advance.

The paper's posture, stated once, because it is a conflict-of-interest disclosure and not a flourish: this archive argued the imprint account in The Living Room and the Long Grip. This paper holds the table — built so that account can lose on the same terms as its rivals. Neutrality is a design property with four checkable faces: the dated-window count (one per account); the risk grades (each row carries one, so a near-certainty can never impersonate a live bet); the live-bet census printed beside the count — 2/1/2/1 across A1/A2/A3/A4, an asymmetry disclosed rather than dressed, with the two near-certain rows (A1.3, A4.1) excluded from the fifth column's predicate so a gimme can neither save the frame nor fail it; and the confounder census (exhaustive and exclusive per row, so no account holds hidden escape hatches). Two referee rounds have now audited these faces; both audits are part of the record. The fifth column also has an advocate with a book: the case that the grip is age and wealth, not cohort, is argued at length in Moyn (2026) — the strongest published statement of the world in which the table's fifth column — the frame-failure row §9 defines — scores.

Registers, for the standalone reader: C marks conceptual claims carried on cited scholarship; E marks empirical claims re-pullable from named public sources; P marks the registered predictions — nothing else in the paper predicts anything.

§ VIIThe Succession Ledger

Five series, each with a named source, an export a reader can operate, a cadence, an opening value with count and date, and pre-agreed thresholds: X₁, exit begun — Boomer share below plurality; X₂ — below a quarter; X₃ — below a tenth. Thresholds are properties of the series, not the world; they let windows condition on ledger states rather than calendar guesses. Two of the five (L2, L4) carry no registered rows and are disclosed as context series; the registration's evidence base is L1, L3, L5 and the vote families.

L1 — Congress, by chamber. Share of voting members born 1946–1964, per Congress. Registered primary: recomputation from the public legislators dataset keyed by the Congressional Biographical Directory's member identifiers (birth dates; pin carried to imprint after an environment failure this round, disclosed); Pew's profiles, the CRS tenure series (R41545), and the frozen FiveThirtyEight generation-composition file (66th–118th Congresses) as checks, with fallback rules stated in the codebook. Opening, 119th Congress, January 2025 seating: House 39 percent of 435 — X₁ crossed, Generation X the plurality at 41 percent; Senate 60 of 99 seated (one vacancy at the profile date) — 60.6 percent, X₁ not crossed. Medians 57.5 and 64.7, both falling. One caution rides the series (Pocasangre & Wahl, 2025): entering members are themselves older than their predecessors were at entry — median new-member age 50 in 2025 against 39 in 1979 — so the Boomer share can fall while the chamber's age profile barely moves; the cohort clock and the age clock are distinct clocks, which is the point of §9's age adjustment.

L2 — The Article III judiciary (context series; no rows). Cohort share of sitting judges by tier, computable from the Federal Judicial Center's Biographical Directory export (demographics with birth years, CSV, 1789–present). Opening value computed at imprint. Life tenure makes it the slowest clock — though not an apolitical one: judicial exits are strategically timed toward same-party presidencies (the retirement literature from Spriggs & Wahlbeck to Deschler & Sen) — one more reason the series contextualizes A1's long tail and scores nothing.

L3 — Wealth, by generation. Federal Reserve Distributional Financial Accounts, quarterly since 1989, scored on first-release vintages with a registered fallback if the Fed's generational buckets are redefined (recompute on birth years 1946–64 or mark unscoreable). Opening, 2026 Q1: Boomer share 51.6 percent (Gen X 26.1, Silent and earlier 11.3, Millennial and later 11.0). Beside the thresholds, the transfer clock at its registered spread: $124 trillion through 2048, $105 trillion of it to heirs, 81 percent of the total held by Boomer-and-older households (Cerulli); against a Visa analysis near $36 trillion over twenty years (as reported July 2026). The spread is the reliability statement — with scholarly warrant: the academic inheritance-flow literature (Piketty, QJE 2011; Alvaredo, Garbinti & Piketty, Economica 2017; Feiveson & Sabelhaus 2018; the current survey and distributional readings, Horioka 2025 and Zhu & Zinn 2024) finds recent U.S. trends genuinely uncertain, and the genre's founding projection (Havens & Schervish 1999) ran ahead of realized transfers for two decades (Wolff & Gittleman 2014).

L4 — Corporate leadership (context series; no rows). CEO cohorts from the annual Crist|Kolder Volatility Report; boards from Spencer Stuart's U.S. Board Index (fortieth edition, 2025; new-appointee average age 59.1). Proprietary PDFs with no archival guarantee — which is why nothing scores here; openings extracted at imprint.

L5 — The electorate. Boomer share of reported voters, from the Census Bureau's biennial Voting and Registration tables (its P20 report series; exact table, and the registered cohort assignment — single years of age from the public microdata, with the published-bracket mapping as fallback — in the codebook). The discriminator between office clocks and electorate clocks.

L6 — dropped, by name. Executive-branch and media-masthead composition carry no re-pullable series; two search rounds confirmed. A reader who builds the sixth is invited to file it.

§ VIIIThe Signatures

Each account as a function of the ledger, with its risk beside its prediction. A1 — Imprint: behavior tracks L1–L2 with at most one Congress's lag; the chamber asymmetry should produce divergence — a claim now registered as a quantity, not a gesture (§9, the interaction term), and registered two-sided: §4's own scope lesson is that imprint fixes salience, not direction, so the account's signature is cohort structure moving on the office clock, with the direction reported and interpreted, never bet on; the judiciary moves last; and the account accepts the tripwire-grade exposure that the House has already crossed. Whether cohort-scored behavior is moving there now is a question the author has not answered: no increment regressions have been run on any Congress, and none will be until the scoring script is frozen — the baseline is the first memo's job, not this draft's. A2 — Size: behavior tracks L5 irrespective of office; no chamber divergence net of registered constituency covariates; and — new this draft — a positive claim on the domain where numerical weight should bite hardest if it bites anywhere: old-age policy. A3 — Sorting: no fade on any cohort series, bound by the conjunction rule so nulls score only while the ledger confirms the exit proceeding; at risk in every succession pair where a same-party successor breaks the predecessor's baseline. A4 — Wealth structure: asset-linked behavior tracks L3 and the transfer clock, decoupled from office and electorate; the transfer itself now measured where it can fail — in flow data, not shares. The fifth column: none of it materializes; the frame fails for these observables, in this case, the account of The Living Room and the Long Grip included as applied.

§ IXThe Registration Table

Twelve rows, three per account. Every row carries: its decision rule; its risk grade — live bet, moderate, or near-certain, with the reference-class note — so the symmetry of counts can never disguise an asymmetry of stakes; and its confounder list, exhaustive and exclusive, drawn from the ten shock classes of §10 (in brief: S1 an attack on U.S. soil, S2 a new war authorization, S3 a chamber majority flip, S4 a federal franchise statute, S5 a recession onset, S6 a Senate rules change, S7 a financial crisis, S8 an estate-tax change, S9 a change in the party of the presidency, S10 the post-2030 reapportionment; exact definitions in §10) — only listed classes may flag the row; "none admissible" is a registered property. One calendar-dated window per account (T1 count 1/1/1/1); all else fires on ledger triggers with stated lags and tolerances.

The scoring instrument (Artifact A — an imprint gate). All vote rows are scored by a frozen script and codebook, deposited with an independent timestamp before the first baseline is computed. Core, registered here: member-level logistic regression of vote position on party, Congress, birth-cohort bins (pre-1946; 1946–64; 1965–80; 1981+), and the chamber-by-cohort interaction, estimated on both chambers pooled across every Congress from the 117th forward with files at memo date — the pool is a rule, never a frozen list — so that same-age cross-Congress contrasts identify cohort net of the age control: the age-adjusted cohort increment, reported as a within-party average marginal effect at the registered common age. Inference is a member-clustered Wald test at α = .05 under a per-memo Holm correction — the standard adjustment for testing several rows at once, applied with its running-maximum rule — across all primary rows scored in that memo; the equivalence bound is five within-party percentage points on a cluster-bootstrap standard error, with equivalence rows scoreable only when that standard error is at or under three points — an underpowered null is a registered outcome, not a quiet score; and every scored decision must replicate across the registered family of age specifications, linear through spline, with a six-year-shifted placebo cohort boundary that must stay silent — a placebo that fires flags the instrument, not the world. The vote families the live wire scores, glossed here so a reader of the paper alone knows what they are: O1 is surveillance, O2 war powers; full mechanical definitions in the codebook. The residual identifying assumption — no Congress-specific age-by-vote-family interactions beyond the registered controls — is an assumption, named as one and now watched: the placebo boundary and the specification family are its registered auditors, and every memo publishes the Fosse–Winship nonlinear components for anyone who wants to run the bounds the archive does not. The codebook fixes: the direction coding of every roll call in each family (oversight-direction rules per bill section); mechanical family membership (CRS policy-area and subject terms; referral committees; final-passage and amendment votes; a bridging rule restricting cross-Congress comparisons to comparable vote types); the O5 start-date audit (O5 is the family of congressional votes on age and term limits for federal office) — answered now rather than deferred: the class is non-empty as of September 28, 2023, when the House Judiciary Committee marked up the congressional term-limits amendment (H.J.Res. 11) and it failed there 17–19, member positions on the record — four Republicans joining the noes, a tally the archive's panel verified against the record after a first search missed the markup; the family's environment includes North Dakota's Measure 1 (June 2024, 60.8 percent, presumptively unenforceable under Thornton, untested by any candidacy to date) and a standing population of introduced amendments — so every O5 row is window-scoped by construction; the CPS bracket-to-cohort mapping; and the DFA vintage rule. Four registered provisions enter before the freeze, because the amendment rule bars them after. First, member military service (combat and non-combat where codable) is a registered covariate for every O1–O2 row — the Boomer bins carry Vietnam-era veterans as the post-Boomer bins do not, and service moves exactly these votes (Gelpi & Feaver, APSR 2002; Feaver & Gelpi 2004; Bianco 2005; Fiorelli, LSQ 2025) — with both models pre-specified and the interpretation rule stated now: an increment robust to the covariate is imprint beyond service; an increment absorbed by it is service-mediated imprint, and scores as such. Second, a sub-bin robustness split (1946–55; 1956–64) is a labeled output of every memo — the paper's own imprint authority places formation boundaries inside the Census span — with the primary rows scored on the registered bins while the split watches for two imprints canceling; GGA's scope note travels with it (the model's generational evidence is strongest for white voters, a named limit on what the authority licenses). Third, terms served — computed mechanically from the member file — is published as a labeled model pair, never a scoring model: tenure is causally downstream of cohort, so conditioning on it would absorb true cohort structure; a scored row whose increment reverses sign under the tenure model is tenure-flagged, the label traveling with the score. Fourth, the exit-selection ledger: every congressional departure is logged by mode — death, defeat, retirement, higher office — because the mortality clock is not the dominant exit mode in Congress and the design prices that rather than assuming it away; a row scored over a window in which voluntary exits exceed seventy percent of Boomer departures carries a selection label wherever its score is reported, and each memo publishes the contrast restricted to seats vacated by death or defeat as a labeled output. Prior data contact, disclosed: none — the author has run no increment regressions on any Congress, and the disclosure extends to family construction: the seed family map for the 117th–119th ships frozen in the deposit, its directions coded by the mechanical rules before any regression runs, and later Congresses' maps are rule-built and SHA-published in a dated note before each memo's regressions. The first run is the frozen script's, published as the t₀ baseline in the first memo with a signed statement — together with the power gate every memo must print: the party contrast the instrument must recover, the bootstrap standard error of the gap, cell sizes, and roll-call counts, so a table of nulls can never be read without the check that the instrument could see anything at all.

Imprint rows. A1.1 — the live wire (dated; risk: live bet). By the close of the 121st Congress (January 2031; tolerance one Congress), the chamber × post-Boomer interaction in the pooled O1–O2 model is Holm-significant, with the House's within-party post-Boomer-minus-Boomer gap at least 5 points in magnitude, stable across the registered specification family (an unstable decision is spec-fragile, a registered outcome that does not score) — two-sided, per §8: the bet is that cohort structure moves on the office clock; the direction is reported and interpreted as a labeled output, never wagered. This is the divergence claim, registered as a quantity. Confounders: S1, S2, S3, S9, S10. A1.2 — the chamber echo (risk: live bet). Within one Congress of the Senate crossing X₁ (+1 tolerance), the Senate gap matches the House's sign at ≥ 3 points in magnitude (the smaller chamber, the wider bound). Confounders: S1, S2, S3, S6, S9, S10. A1.3 — the reform items of The Living Room and the Long Grip (risk: near-certain — re-priced by the archive panel against the demonstrated base rate: the class produced a recorded committee tally in the 118th alone; the row is disclosed as cheap and sits outside the fifth column's predicate). Within two Congresses (+1) of both chambers passing X₁, at least one O5-family measure receives a recorded floor or committee vote within that window. Confounders: none admissible.

Size rows. A2.1 — no divergence (risk: live bet). Over A1.1's window, the same pooled model with registered constituency covariates (ACS five-year cohort shares and median age, vintage rule in the codebook) returns a chamber × cohort interaction inside the five-point equivalence bound on the bootstrap standard error, across the specification family. One primary model; A1.1 and A2.1 claim disjoint regions of its estimand, and the covariate-adjusted estimand is published beside it as its own quantity each memo. Confounders: S1, S2, S3, S10. A2.2 — the electorate crossing (dated; risk: near-certain on the crossing, live on the attribution — graded as its parts). L5 plurality loss in a federal election in the 2028–2036 band (tolerance one cycle, symmetric; the L5 opening value is computed and registered at deposit in the baseline run, and if that baseline shows the crossing already behind us, the crossing sub-claim scores never-triggered-retrospective — a registered outcome — leaving only the attribution sub-claim live). Attribution rule: movement in the vote families is dated by the first memo showing a significant gap change and attributed to the nearest preceding crossing; if office and electorate crossings land within one cycle of each other, the row scores unattributable — a registered outcome, not a failure. Confounders: S4. A2.3 — the old-age-policy clock (risk: moderate; the account's positive row). In the O6 family — Social Security and Medicare benefit-expansion measures, mechanically defined in the codebook — aggregate support holds or rises while the Boomer electorate share retains plurality, then declines within two Congresses (+1) of L5 plurality loss. Numerical weight, biting where it should bite hardest if it bites anywhere — and the row names what it bets against: Campbell's policy-feedback account (How Policies Make Citizens, 2003), on which Social Security regenerates senior mobilization with every incoming cohort and predicts no post-plurality decline, and Curry & Haydon's member-age findings (2018), on which age rather than cohort drives senior-issue attention. If A2.3 fails, those explanations are already on the record. Confounders: S5, S9.

Sorting rows. A3.1 — party swallows cohort (dated; risk: live bet). Through the 122nd Congress, the age-adjusted increment stays inside the equivalence bound in every scored Congress — under the conjunction rule: the null scores for sorting only in Congresses where the House Boomer share has declined at least one point from the 119th baseline. Confounders: S1, S2, S10. A3.2 — seat-level inheritance (risk: live bet). For two Congresses after each same-party Boomer-to-post-Boomer seat succession (+1), the successor's O1–O2 support sits within five points of the predecessor's own last-two-Congress record; party-flip successions are excluded and logged; the row is scored only once the panel holds twenty pairs, and scores for sorting when at least seventy percent of accumulated pairs sit inside tolerance. Confounders: S10 — the reapportionment redraws the seats the pairs live in. A3.3 — the boldest refusal (risk: moderate). Within the window from both-chambers X₂ to one Congress past it, no recorded O5 vote — window-scoped, not cumulative, because the class's history is already non-empty (the audit). Under the conjunction rule. Confounders: none admissible.

Wealth rows. A4.1 — the share holds (dated; risk: near-certain — re-priced by the archive panel: failure requires a sustained decline near 1.2 points per year against a share that rose in the latest quarter, and the share statistic carries the age-wealth gradient the vote rows adjust for and this row does not; disclosed as such, it sits outside the fifth column's predicate). The DFA Boomer share stays above 40 percent through 2035 Q4, scored on first-release vintages. Confounders: S7, S8. A4.2 — asset politics on the asset clock (risk: live bet). In the estate-and-basis family — defined now: measures carrying CRS subject terms for estate, inheritance, and gift taxation or stepped-up basis, referred to Finance or Ways and Means, final-passage and amendment votes — the age-adjusted increment stays inside the equivalence bound through one Congress past both-chambers X₁, scored only in memos where A1.1's increment is simultaneously live — the divergence-across-families condition as a conjunction rule. Confounders: S5, S8, S9. A4.3 — transfer in the flows (risk: moderate). In the IRS Statistics of Income estate-tax aggregates — public, annual — real total gross estate value on returns above a fixed real threshold ($20 million in 2026 dollars) for 2030–2034 exceeds the 2020–2024 total by at least half. The transfer, measured where it can fail: in flows, not shares — because the last announced great transfer underdelivered against its projections (Havens–Schervish 1999; Wolff–Gittleman 2014), which is exactly the failure this row exists to detect. Confounders: S7, S8.

The fifth column (R-0). Evaluated at the terminal date of the last live-bet or moderate row — not at a fixed calendar date — R-0 scores FOR only when three things hold at once: no live-bet or moderate row has scored FOR its account under its stated rule (the two near-certain rows sit outside this predicate — a gimme can neither save the frame nor fail it); the instrument was seen working — the power gate passed in at least half the scored memos and no placebo flag stands unresolved; and the ledger confirmed the exit proceeding for at least two scored Congresses. If the first holds and either check fails, R-0 returns not-evaluable — a registered outcome distinct from failure and from survival: the table could not adjudicate the frame, and says so instead of pretending. Consequence when R-0 scores, registered: the generational frame failed for these observables, in this case; the load-bearing assumption of §1 takes the hit; the account of The Living Room and the Long Grip takes it as applied. What R-0 does not license: any claim about domains the table never scored — and what a not-evaluable R-0 does not license is any claim at all.

§ XThe Scoring Protocol — and What Registration Buys

Cadence. A public scoring memo each Congress, beginning with the 120th; pre-trigger memos publish ledger updates, the conjunction status, and nothing that banks — the asymmetry statement covers them explicitly, so early quiet can never be read as a null quietly scoring. Tribunal. The frozen script is the scorer. Anyone can run it; a published outside run that conflicts with the archive's own supersedes it if the discrepancy stands unresolved for one memo cycle. "Public tribunal" means exactly this and nothing grander. The shock log. Ten classes, objective, enumerated: S1 — an attack on U.S. soil with one hundred or more deaths; S2 — enactment of a new authorization for use of military force or a declaration of war; S3 — a chamber majority flip; S4 — a federal franchise or turnout statute enacted; S5 — an NBER-dated recession onset; S6 — a Senate floor-rules change restricting amendment votes; S7 — a financial crisis (NBER dating plus a twenty-five percent real equity drawdown); S8 — a federal estate-tax statutory change; S9 — a change in the party of the presidency (oversight behavior runs on out-party incentive; Kriner & Schickler, 2016); S10 — the first Congress seated under the post-2030-census apportionment (the 123rd, January 2033). The class-to-row mapping is the confounder lists of §9 — exhaustive and exclusive; a class not listed on a row cannot flag it; rows marked "none admissible" cannot be flagged at all. A flagged row is shock-flagged, unscored, and extends its window once, by one Congress, by rule; each class may flag a given row at most once in the row's lifetime; and a second flag from among the rare classes (S1, S2, S4, S5, S6, S7, S8) converts the row to shock-terminated, a registered outcome outside the fifth column's predicate. The routine-politics classes (S3, S9, S10) extend but can never terminate — an ordinary election cannot quietly kill a row. No memo discretion anywhere in the chain. Freeze and amendment. The table freezes at publication, deposited with the script under an independent timestamp (OSF Registrations or Zenodo, DOI-pinned); before the first memo, the archive additionally invites independent forecasts on the twelve rows, in the Social Science Prediction Platform's manner — an external benchmark no prior succession theory ever had. The calendar windows of §9 are written against a deposit by the close of the 119th Congress; if deposit slips past January 2027, every dated window re-bases by the slippage, once, in the deposit itself — a pre-freeze liberty, disclosed here and exercised nowhere else. After the first in-window data arrive, amendment is restricted to one act: striking a row, publicly, with reasons. Nothing may be loosened, redefined, or added — the vote-family definitions travel inside the frozen codebook precisely so that redefinition is impossible, and the seed family map for the 117th–119th deposits frozen beside them — later maps are rule-built and SHA-published before each memo's regressions run. Continuity. If the archive misses two consecutive memos, the registration lapses and is scored as of the last memo published. A paper about succession carries a succession plan.

And the concession that bounds Part II. Registration buys discipline, not identification. One run of history; no counterfactual America; no control chamber; the shock log will fill; the identifying assumption behind the age-adjustment is an assumption; and registrations are kept by execution, not intention — adherence rates in the discipline that invented them are sobering (Claesen, Gomes, Tuerlinckx & Vanpaemel, 2021), which is what the continuity clause and the outside-run tribunal are for. What the table buys is narrower and real: it makes each account's failure legible — dated, directional, unspinnable — and strips every account, this archive's own first among them, of the privilege Part I diagnosed. Where identification is impossible, pre-commitment is what honesty has left; the alternative is four accounts retrofitted to whatever the 2030s deliver, and the certainty of learning nothing.

§ XIThe Ledger of Claims and Non-Claims

Register C. Succession as a recurring experiment among societies with continuous institutions (Ryder, ASR 30(6):843–861); the four-carrier grammar with scope conditions; near-equivalence about any completed past, separability at a watched exit — the hinge, now itself exposed to the table's outcome; the APC degeneracy — stated with the identification literature's precision — binding this paper's instrument too, and the age-adjusted, specification-swept, placebo-guarded design it forces; the after-the-fact privilege as the field's structural vice; §4's reach findings as trajectory claims only — trajectories within single successions diverge by domain; exit management is a variable, bracketed by Moscow's freeze and Beijing's program; the no-cycles fence; registration transposed to history buys discipline, not identification; the tribunal-outward precedent; the structural register about a living generation; neutrality as a checkable design property; and Part I's three named exposures to Part II (§5). Register E, re-pullable, with counts and dates. The DFA shares (2026 Q1, first release); the House 39 percent of 435 and Senate 60 of 99 seated (119th Congress, January 2025 profiles) with medians 57.5/64.7; the FJC export mechanism; the CPS P20 family; the cohort boundary arithmetic; Skocpol's pension figures as re-dated (early-1890s receipts peak on the order of forty percent; eighteen percent of over-65s covered by 1910; pages at imprint); the pension tail to 2020; Manion's characterization (page at imprint); the Politburo ages (Beissinger, 1986, 288–314); the 1824–25 fact pattern (Howe, 2007); Spencer Stuart's 59.1 (fortieth edition, 2025). Placeholders permitted in a draft and barred at imprint: the freeze is gated on zero open E-flags. Register P. The twelve rows and the fifth column. Nothing else predicts anything — including A1.2's trigger phrasing, which carries an implicit timing belief (Senate actuarial turnover) that the paper licenses explicitly: trigger-based rows assume their triggers arrive within the registration's life; if a trigger never fires, its rows score never-triggered, a registered outcome distinct from failure.

Non-claims. No outcome forecasts. No adjudication — this archive argued A1 in The Living Room and the Long Grip, a relationship disclosed in the paper because it is a conflict of interest; this paper holds the table. No moral ledger, no obituary. No cycles, archetypes, or periodicity. No travel beyond the scope conditions. No APC decomposition presented as identification. No re-litigation of The Living Room and the Long Grip. No §4 case argued as account-evidence — trajectory language only, enforced as a tripwire. And no claim of priority beyond the archive's knowledge.

Tripwires. T1: dated windows 1/1/1/1, binding every amendment — and amendment itself now restricted to striking. T2: any row a stranger cannot score from public data with the frozen artifacts is struck; the codebook exists because a referee enumerated nine ways a stranger would fail, and each has a registered resolution. T3: the fifth column, horizon-coherent, three-condition-gated so an instrument failure can never impersonate a frame failure, with the account of The Living Room and the Long Grip inside the blast radius, as applied. T4: shocks flag by rule and never excuse. T5: retrodiction labels in §4, enforced against the claims register as well as the prose — attribution language is a section failure wherever it appears.

§ XIIThe Close

Societies keep running this experiment, and keep grading it after the students leave the room: the founders' republic re-founded itself and called it destiny; the pension state contracted with its veterans and the obituaries called it obvious; Moscow froze and Beijing engineered, and the theorists explained both, afterward, fluently. Part I says these successions can be read. Part II says the reading can, once, be done honestly — before the ending is known, in public, under rules that would let the archive's own account lose first.

The archive closes on the falsifier it most fears: row A1.1, the live wire — the archive's own account failing at the first fence, in the chamber that has already crossed, under a scoring script the archive froze before it ever ran. The paper would rather lose that row than shelter it; an instrument that cuts is worth more than an account that survives. And the invitation, with the overclaim removed: other formed cohorts are aging in power on clocks less public than America's — Iran's revolutionary generation above all — and the grammar, the ledger design, and the protocol are offered to whoever can assemble the series there. Societies will keep running the experiment. The only question this paper answers today is what it looks like, finally, to run one on the record.

Notes

Registers. C marks conceptual claims carried on cited scholarship; E marks empirical claims re-pullable from named public sources, with counts and dates; P marks the registered predictions — the twelve rows and the fifth column — and nothing else in the paper predicts anything.

Rules. No prophecy (windows condition on ledger states; the one dated window per account carries stated tolerance). No obituary (the subject is a living generation; the register stays structural). No monocause (four accounts, equal resolution, symmetric by dated-window count, with the live-bet census printed beside it and every risk grade disclosed). No cycles (no periodicity, archetypes, or turnings; the fence names Turchin's registered forecast precisely because the fence is against unpriced cycles, not registered ones).

Review record. One internal adversarial round (eleven majors — among them the within-Congress age–cohort collinearity, a tautologous wealth row, non-exclusive head-to-heads, and a broken fifth column — all repaired in v3) and the standing coverage round (thirteen integrations in v4, from the forecasting lineage to the veteran-legislator covariate to the discovery that the O5 vote class was already non-empty). Referee reports and response letters are filed with the archive. A full archive panel followed (four referees, blind and parallel, August 2026 — methods in protocol-audit mode over Artifact A itself): decision memo and raw reports in the repo (reviews/011/), disposition MAJOR REVISION, executed as v5 — the instrument rebuilt (v1.0: clustered Wald, bootstrap equivalence, one registered A1.1 quantity, the pool as a rule, the state machine), the identification hardened (specification family, placebo boundary, power gate, exit-selection ledger, tenure watch, S9/S10), the fifth column given its three-condition predicate, and the bets re-priced in the open (A1.3 and A4.1 to near-certain, outside R-0; the live-bet census printed).

Artifact A. The frozen scoring script and codebook ship with the paper at v1.0 — the August 2026 archive-panel rebuild; v0.9 stays filed as history: the registered model (both-chambers pooled logit, chamber-by-cohort interaction, within-party average-marginal-effect gap, member-clustered Wald, Holm with running maximum, cluster-bootstrap equivalence), the specification family and placebo boundary, the power gate, the state machine that emits a status for every row with no hand step, the veteran pair with its interpretation rule, the tenure and Fosse–Winship labeled outputs, the exit-selection ledger, the sub-bin split, and the nine stranger-in-2035 resolutions. Prior data contact: none — no increment regression has ever been run, and no family map has ever been tabulated against cohort patterns; the first run is the frozen script's. The single consolidated gate: run Artifact A v1.0 end-to-end on the 117th–119th Congresses, compute the remaining ledger openings (the L5 opening now registered at deposit) in the same run, and deposit table, script, codebook, seed family map, and baseline memo with an independent timestamp (OSF/Zenodo). The registration clock runs from deposit.

Verification record. The Ryder epigraph verified verbatim at source (the abstract's opening sentence); Wohl, Khong, Manion, Hoffmann, Beissinger, and Howe pinned; Skocpol's figures re-dated to the 1890s receipts peak; the Manion superlative attributed to the Princeton edition's characterization; Spencer Stuart's 2026 index and the Visa attribution fixed. The August 2026 archive panel's verification pass (thirty-eight checks) and the editor's re-verification are filed in the repo; its corrections — the dek quotation from The Living Room and the Long Grip, the pension tail's last recipient, the Hoffmann scope, the North Dakota and Spencer Stuart figures, the H.J.Res. 11 tally — are executed in this draft. Page-level cites complete in the deposit's reference file.

References

Ahlfeldt, G. M., W. Maennig & S. Q. Mueller (2022). "The generation gap in direct democracy: Age vs. cohort effects." European Journal of Political Economy 72:102120. · Alvaredo, F., B. Garbinti & T. Piketty (2017). "On the Share of Inheritance in Aggregate Wealth: Europe and the USA, 1900–2010." Economica 84(334):239–260. · Bartels, L. M. & S. Jackman (2014). "A Generational Model of Political Learning." Electoral Studies 33:7–18. · Beissinger, M. R. (1986). "In Search of Generations in Soviet Politics." World Politics 38(2):288–314. · Bell, A. & K. Jones (2013). "The impossibility of separating age, period and cohort effects." Social Science & Medicine 93:163–165. · Bianco, W. T. (2005). "Last Post for 'The Greatest Generation.'" Legislative Studies Quarterly 30(1):85–102. · Born, R. (1979). "Generational Replacement and the Growth of Incumbent Reelection Margins in the U.S. House." APSR 73(3):811–817. · Bottomore, T. B. (1993). Elites and Society, 2nd ed. Routledge. · Campbell, A. L. (2003). How Policies Make Citizens. Princeton. · Cerulli Associates (2024). U.S. wealth-transfer projection: $124T through 2048. · Claesen, A., S. Gomes, F. Tuerlinckx & W. Vanpaemel (2021). "Comparing dream to reality: an assessment of adherence of the first generation of preregistered studies." Royal Society Open Science 8(10):211037. · Cotta, M. & L. Verzichelli (2007). Political Institutions in Italy. Oxford. · Crist|Kolder Associates. Volatility Report (annual). · CRS R41545. Congressional Careers: Service Tenure and Patterns of Member Service, 1789–2025. · Curry, J. M. & M. R. Haydon (2018). "Lawmaker Age, Issue Salience, and Senior Representation in Congress." American Politics Research 46(4). · Delli Carpini, M. X. (1986). Stability and Change in American Politics. NYU Press. · DellaVigna, S., D. Pope & E. Vivalt (2019). "Predict Science to Improve Science." Science 366:428–429. · Deschler, J. & M. Sen (2024). "The Role of Judge Ideology in Strategic Retirements in U.S. Federal Courts." Journal of Law & Empirical Analysis. · Easterlin, R. A. (1980). Birth and Fortune. Basic Books. · Feaver, P. D. & C. Gelpi (2004). Choosing Your Battles. Princeton. · Federal Judicial Center. Biographical Directory of Article III Federal Judges: Export. · Federal Reserve. Distributional Financial Accounts (quarterly, 1989–). · Feiveson, L. & J. Sabelhaus (2018). "How Does Intergenerational Wealth Transmission Affect Wealth Concentration?" FEDS Notes. · Fiorelli (2025). "Military Experience and the Use of Force." Legislative Studies Quarterly. · Firebaugh, G. (1989). "Methods for Estimating Cohort Replacement Effects." Sociological Methodology 19:243–262. · Forecasting Collaborative (2023). "Insights into the accuracy of social scientists' forecasts of societal change." Nature Human Behaviour 7:484–501. · Fosse, E. & C. Winship (2019a). "Analyzing Age-Period-Cohort Data: A Review and Critique." Annual Review of Sociology 45:467–492. · Fosse, E. & C. Winship (2019b). "Bounding Analyses of Age-Period-Cohort Effects." Demography 56(5):1975–2004. · Gelpi, C. & P. D. Feaver (2002). "Speak Softly and Carry a Big Stick?" APSR 96(4):779–793. · Ghitza, Y., A. Gelman & J. Auerbach (2023). "The Great Society, Reagan's Revolution, and Generations of Presidential Voting." AJPS 67(3):520–537. · Havens, J. J. & P. G. Schervish (1999). "Millionaires and the Millennium." Boston College SWRI. · Hoffmann, B. (2009). "Charismatic Authority and Leadership Change: Lessons from Cuba's Post-Fidel Succession." International Political Science Review 30(3):229–248. · Horioka, C. Y. (2025). "Bequests." NBER Working Paper 33927. · Howe, D. W. (2007). What Hath God Wrought. Oxford. · Inglehart, R. (1977). The Silent Revolution. Princeton. · Inglehart, R. & P. R. Abramson (1994). "Economic Security and Value Change." APSR 88(2):336–354. · Jennings, M. K. & R. G. Niemi (1974; 1981). The Political Character of Adolescence; Generations and Politics. Princeton. · Jennings, M. K., L. Stoker & J. Bowers (2009). "Politics across Generations." Journal of Politics 71(3):782–799. · Judis, J. B. & R. Teixeira (2002). The Emerging Democratic Majority. Scribner. · Khong, Y. F. (1992). Analogies at War. Princeton. · Kriner, D. L. & E. Schickler (2016). Investigating the President: Congressional Checks on Presidential Power. Princeton. · Levy, G. & R. Razin (2025). "The dynamics of cohort effect in politics." Journal of Public Economics 247. · Long Now Foundation (2002–). Long Bets. longbets.org. · Luo, L. (2013). "Assessing Validity and Application Scope of the Intrinsic Estimator Approach to the Age-Period-Cohort Problem." Demography 50(6):1945–1967. · Manion, M. (1993). Retirement of Revolutionaries in China. Princeton. · Mannheim, K. (1928/1952). "The Problem of Generations." In Essays on the Sociology of Knowledge, trans. P. Kecskemeti. Routledge and Kegan Paul, 276–320. · McClean, C. T. (2024). "Young versus old politicians and public spending priorities." Journal of Economic Behavior & Organization 225:88–106. · Moyn, S. (2026). Gerontocracy in America: How the Old Are Hoarding Power and Wealth — and What to Do About It. Farrar, Straus and Giroux. · Munger, K. (2022). Generation Gap. Columbia. · Nosek, B. A., C. R. Ebersole, A. C. DeHaven & D. T. Mellor (2018). "The preregistration revolution." PNAS 115(11):2600–2606. · NBC News (2021, April 16). "Cuba's Raul Castro confirms he's stepping down as head of Communist Party." · Pareto, V. (1901/1968). The Rise and Fall of the Elites. Bedminster. · Pew Research Center (2025). 119th Congress demographic profiles. · Piketty, T. (2011). "On the Long-Run Evolution of Inheritance: France 1820–2050." QJE 126(3):1071–1131. · Pocasangre, O. & D. Wahl (2025). The Age Divide. Protect Democracy / New America. · Rekker, R. (2024). "Electoral change through generational replacement: An age-period-cohort analysis of vote choice across 21 countries between 1948 and 2021." Frontiers in Political Science 6:1279888. · Rintala, M. (1963). "A Generation in Politics: A Definition." Review of Politics 25(4):509–522. · Ryder, N. B. (1965). "The Cohort as a Concept in the Study of Social Change." ASR 30(6):843–861. · Skelley, G. (2023). "Congress Today Is Older Than It's Ever Been." FiveThirtyEight, with the data_aging_congress file. · Skocpol, T. (1992). Protecting Soldiers and Mothers. Harvard. · Social Science Prediction Platform (2020–). socialscienceprediction.org. · Spencer Stuart (2025). U.S. Board Index, 40th ed. · Spriggs, J. F. II & P. J. Wahlbeck (1995). "Calling It Quits: Strategic Retirement on the Federal Courts of Appeals, 1893–1991." Political Research Quarterly 48(3). · Stockemer, D. & A. Sundström (2025). "Age Inequalities in Political Representation: A Review Article." Government and Opposition 60(1):271–288. · Tetlock, P. E. (2005). Expert Political Judgment. Princeton. · Tetlock, P. E. & D. Gardner (2015). Superforecasting. · Turchin, P. (2010). "Political instability may be a contributor in the coming decade." Nature 463:608; (2020) PLOS ONE 15(8):e0237458. · van der Brug, W. & S. Kritzinger (2026). "Generations and age in electoral research." European Political Science Review 18(3):389–409. · Wohl, R. (1979). The Generation of 1914. Harvard. · Wolff, E. & M. Gittleman (2014). "Inheritances and the Distribution of Wealth." Journal of Economic Inequality 12(4). · Voteview (Lewis et al.). Congressional roll-call database. · Zhu, L. & A. Zinn (2024). The Great Inequality Transfer. Urban Institute.

The companion deck — The Imprint and the Exit — sits beside this paper, with Artifact A (the frozen scoring script and codebook), the execution brief, corpus index, claim outline, ledger specification, registration table, and both referee rounds kept in the archive's working files. The clock runs from the deposit — the timestamped public filing §10 describes; until then, this is a draft that can still be struck, which is the last moment anything about it may be loosened. Score the table — the rules are public, and any run of the frozen script counts. Comments welcome at contactme​@​marshallcahill.com.