BOOK ONTOLOGY / CLUSTER
Books that theorize the same object, compared as a set: what each makes “technological revolution” mean, and the citation lineages that divide them.

What Is a Technological Revolution?

Situating Perez (2002) among the theorists of technological change. Carded members: Perez 2002 (the finance cycle) · Smil 2005 (the great discontinuity) · Ding 2024 (diffusion capacity). The field around them: Mokyr, Gordon, the general-purpose-technology economists, Freeman and Louçã, with Schumpeter and Kondratiev as the shared ancestors.
I · Five answers to one question

Everyone in this cluster is explaining the same thing: the bursts of technology that reorganize the economy since about 1770. They divide on what the unit is, what drives it, and whether it repeats.

Thinker (work)What a “technological revolution” isPrime driverTime-shapeFinance
Perez 2002
theorem
A great surge: a revolution plus its techno-economic paradigm plus a financial cycle, moving through five fixed phasesFinanceRecurring, about every 50 years; five surges, four phases eachthe engine
Mokyr 1990 / 2002
Lever of Riches; Gifts of Athena
The growth of useful knowledge (propositional and prescriptive); the Industrial EnlightenmentKnowledge and cultureCumulative and contingent; no cycle, no recurrenceabsent
Gordon 2016
Rise and Fall of American Growth
A one-off cluster of Great Inventions that raised the American standard of livingA special set of general-purpose inventionsOne wave, 1870–1970, not repeatable (the future disappoints)peripheral
Smil 2005
Creating the Twentieth Century
The greatest technical discontinuity in history, concentrated in 1867–1914Engineering and energyOne discontinuity; the century after only improved and diffused itabsent
Ding 2024
Technology and the Rise of Great Powers
A general-purpose technology whose economy-wide diffusion decides great-power racesInstitutions (diffusion capacity)GPT waves as the backdrop to power transitionsperipheral
GPT economists
Bresnahan-Trajtenberg 1995; Lipsey-Carlaw-Bekar 2005
A general-purpose technology: one pervasive input with innovational complementaritiesA pervasive enabling technologyTwo dozen GPTs across history; no phase lawabsent
Freeman & Louçã 2001
As Time Goes By
The same techno-economic paradigm, embedded in five subsystems of societyTechnology and institutions, co-evolvingRecurring long waves; timed half a cycle off Perezsecondary

The shared ancestors: Schumpeter (creative destruction, the entrepreneur and the financier as the two wheels) and Kondratiev (the half-century long wave). Every member reckons with both. Perez builds on Schumpeter and argues against Kondratiev; Smil argues against both; Ding nods to Schumpeter and ignores Kondratiev. Same two ancestors, opposite uses (section IV).

II · The finance outlier

Perez is the only one who makes finance the engine and the cycle a recurring law.

Mokyr, Gordon and Smil treat technology as secular and non-repeating: cumulative useful knowledge, one special century, one great discontinuity. Ding and the GPT economists treat it as diffusion or taxonomy. Only Perez runs a clock, and only Perez makes the clock financial. The decoupling and recoupling of financial and production capital is what drives her five-phase sequence, and the sequence is what repeats.

The citation base confirms it. Her heavy sources are financial historians: Kindleberger (21), Sobel (16), Susan Strange (12), Galbraith (10), Soros (9), Minsky (7), roughly seventy-five mentions of the mania-and-crash literature. The technology-history guild that anchors the others is absent or glancing: Mokyr zero, Maddison zero, Paul David zero, Rosenberg one, von Tunzelmann one, Nelson and Winter one. The one classic of industrial history she leans on is Landes (7). The book is a finance-history book wearing Schumpeterian clothes, which the title says plainly, and which is exactly what separates her from the technologists.

Consecration made the split visible. When Barry Eichengreen named three “books for the century” in economics (Foreign Affairs, 2022), he put Perez beside Chandler’s The Visible Hand and Gordon’s The Rise and Fall of American Growth. Gordon is her opposite number in this cluster: one special century against her five equal surges. The field files them together and reads them against each other.

III · The negative space — the technology guild she does not cite

Who a book counts is its frame made measurable. The scholars who define the modern study of technological change are constitutive-to-heavy for Ding and Smil, and ghost-or-thin for Perez. Counts are body-text mentions in each carded book (Perez recount 2026-07-22, body only).

Scholar (line)PerezSmilDingWhat they are to the field
Mokyr useful knowledge0443The leading historian of technological change. Gifts of Athena is the same year as Perez, 2002. Ding’s second-heaviest source; her clean ghost.
Paul David path dependence, QWERTY01075Ding’s single heaviest source; evidentiary for Smil. Perez takes lock-in from Brian Arthur (4) and never cites David.
Rosenberg inside the black box1·58Constitutive for Ding (the 1963 machine-tool convergence). One glancing nod in Perez (Rosenberg-Frischtak 1984).
Maddison growth accounting05·The quantitative long-run growth historian; absent from Perez, consistent with her thin econometrics and her rejection of aggregate series.
von Tunzelmann steam, industrialization1·A neo-Schumpeterian tech historian, one of her own school; in her bibliography, cited once in the body.
Her finance base for contrast75Kindleberger 21, Sobel 16, Strange 12, Galbraith 10, Soros 9, Minsky 7. The mania-and-crash tradition that the technologists never touch.

In the bibliography but never cited in the body: Wallerstein (the world-systems ghost, alongside the uncited Arrighi and Braudel). Thin nods of one mention each: Nelson and Winter (1977, ‘natural trajectory’), Abramovitz (1986, catch-up). The pattern is consistent: the tradition that studies how technology itself accumulates is not her evidence base. Finance history is.

IV · Shared nodes, three stances

The three carded books share node ids, so the merged graph shows, without anyone hand-coding it, that the same ancestors are put to opposite uses. Format: relationship / engagement (mentions).

Shared nodePerez (finance)Smil (technology)Ding (institutions)
Schumpeter 1939builds-on / constitutive (33)
the declared foundation
argues-against / adversarial (1)
the cyclical Schumpeter rejected
acknowledges / courtesy (11)
a nod
Kondratiev 1935argues-against / adversarial (8)
rejects the dating, keeps the rhythm
argues-against / adversarial (2)parallel / none (0)
ignored
Mokyr 1990rival / ghost (0)draws-evidence / evidentiary (4)draws-evidence / evidentiary (43)
Paul David 1990ghost (0)draws-evidence / evidentiary (10)builds-on / constitutive (75)
Freeman & Louçã 2001sibling / evidentiary (6)
her own school
·subsumes / extractive (15)
mined, framework dropped
Perez 2002 (how they treat her)argues-against / adversarial (1)
his 1867–1914 would cut her surges 3 and 4
acknowledges / courtesy (4)

The keystone reads down the Mokyr and David rows: the cumulative-knowledge and path-dependence economists are Ding’s two heaviest sources and real sources for Smil, and neither appears in Perez. Reading across the Schumpeter row: one ancestor, three stances (foundation, foil, courtesy). Smil is the only member who engages Perez directly, and he does it to dispute her boundary between the third and fourth surges.

V · Method and members

Counts are body-text mentions from full pdftotext extractions of each book; the Perez column is a body-only recount (bibliography and index excluded) done 2026-07-22 while confirming her citation anatomy, which corrected several first-pass weights and two work identities. Smil and Ding counts are from their card networks. A dot means the node is not surfaced in that card; a check means present in the bibliography with a single body cite. Carded members: perez-2002 · smil-2005 · ding-2024. In the field, not yet carded: Mokyr (The Gifts of Athena, the cumulative-knowledge antithesis) and Gordon (The Rise and Fall of American Growth, the one-century challenger) are the two most useful next cards, since both are already shared nodes across the set. Shared node ids let these single-book graphs union into one field map as the cluster grows.