The Mercantile Thesis claims durable wealth flows to whoever owns the bottleneck the rest of the economy must route through. That is a commercial claim. This essay extends the same shape to thinking work — to where durable intellectual position comes from, and how a single author should structure research so that the position compounds rather than evaporates.
The claim is one sentence: a durable intellectual position is built by recursively mining the source chain behind every load-bearing claim until you hit a primary archive nobody else has bothered to read. I call the discipline Lineage Mining, because every essay in the Lineage canon already does a one-step version of it, and because the discipline scales by going deeper, not wider.
Two predecessors do the load-bearing work for this doctrine. The first is Rick Rubin's The Creative Act (2023), which describes the artist as an antenna for what already exists: find your favorites, then their favorites, then theirs, until you stop being a fan and start becoming a node in the lineage you were studying. The second is Sanford Grossman and Joseph Stiglitz, "On the Impossibility of Informationally Efficient Markets" (1980), which proves that fully efficient markets are impossible — if all information were already in prices, no one would be paid to gather information, so prices could not contain that information in the first place. There is always informational rent available to whoever pays the marginal cost of research.
The fusion is direct. Rubin describes the obligation: walk the source tree backward through the masters. Grossman-Stiglitz proves the obligation pays: at every level of depth, fewer people have done the work, and the rent for having done it compounds. The merchant lens identifies the bottleneck: depth of source is the position the rest of the discourse cannot easily route around, because reproducing it requires paying the cost of the recursion you already paid.
I. The recursion
Lineage Mining is a four-level recursion. Most public writing operates at level zero or one. Most academic writing operates at level two. Senra-quality work operates at level two with rare reach to level three. Sovereign-quality work — what this canon is aiming at — operates at level three by default and reaches level four on the figures where it matters most.
Level 0 — the figure under study. The biographical subject of the essay.
Level 1 — the figure's stated sources: their teachers, their explicit influences, the books on their desk, the people they cite in interviews. Senra is operating at this level when he reads each subject's autobiography and their stated mentors' autobiographies.
Level 2 — those figures' sources: the masters of the masters. The lineage chain back two generations. This is where most secondary scholarship lives. To do it you need to read the bibliographies inside the bibliographies, and follow the references that the subject's mentors themselves cited.
Level 3 — the primary archives those level-2 sources drew from. Personal correspondence, business ledgers, court records, archival photographs, the artifacts in special collections. The Edison Papers at Rutgers, the Hagley Museum's Westinghouse records, the Medici Archive Project. To reach this level you need physical or licensed access; you need to know the archivists; you often need to know what is in the boxes before you can ask for them.
Level 4 — the cultural-economic substrate the primary archives themselves emerged from: tax records, parish registers, port manifests, account books, the records that document what was true at the system level around the subject. This is where the merchant lens meets the historian's lens: you are no longer studying a person but a flow they ran through.
The recursion is finite — the substrate runs out — but the discoverable depth is much greater than the public-discourse default. A Mansa Musa essay at level 0 is a Wikipedia recap. At level 3 it is an analysis of the Cairo gold-price collapse documented in the al-Maqrizi chronicles and the Mali court records preserved at the Institut des Hautes Études et de Recherches Islamiques in Timbuktu. The same subject, the same merchant principle, but the level-3 essay is the one that cannot be disintermediated by a tighter prompt or a cleverer model — because the model cannot read the document the model does not have.
II. Why this is the bottleneck for thinking work
Grossman-Stiglitz formalize what working researchers always knew: the marginal collector of information earns the marginal informational rent. If everyone could costlessly read the level-3 archive, the rent would compress to zero. The cost is what holds the rent up.
For text on the open internet, that cost is now near zero — large models trained on the open web have absorbed level-1 and most of level-2 already. The rent at those depths is gone. A model can summarize Senra's level-2 read of a biography in seconds, and produce something that reads as competent because Senra's source material is already in the model's training set.
The rent that remains is at level 3 and below. The Hagley Museum's Westinghouse business correspondence is not on the open web. The al-Maqrizi chronicles in untranslated Arabic are not on the open web. The Yale Beinecke's rare manuscript collection is not on the open web. The Medici Archive Project's transcribed but unpublished court records are not on the open web. Reading these costs time, access, language, and the willingness to walk into a reading room with a notebook and no model.
That is exactly where Lineage Mining concentrates effort. It is the position the mercantile lens predicts is durable, applied to intellectual work: own the depth your competitors cannot reach because they will not pay the cost.
III. The operational consequence
From this doctrine forward, every essay in the Lineage canon ships at level 3 by default. Concretely:
- Every Lineage subject's existing biographies are read at level 1 (the standard biographical canon).
- Each biography's own bibliography is read at level 2; at least three level-2 sources per essay are read end-to-end, not skimmed.
- For each essay, one specific level-3 source is named — an archive, a primary document, a transcribed correspondence — and either read directly (when access permits) or sourced through interlibrary loan, digitized special-collections, or a paid archive query. The source is cited by call number or archive reference, not by secondary mention.
- The essay states explicitly which level-3 source it walked, what the source said that secondary sources mis-summarize, and one specific instance where the deeper read changes the merchant-principle reading of the subject.
The published essay can still be 3,000 words. The mining behind it can be 30,000 pages. The proof level is documented per essay: an essay that walks a level-3 source is labeled level-3 mined; an essay that does not is labeled level-2 (secondary) and is treated as a placeholder until the deeper walk is done.
IV. The falsifiable test
Lineage Mining is a research doctrine, not a thesis. It does not predict the world; it predicts what the author's work product looks like.
The falsifiable test is direct: by 2027-12-31, the Lineage canon should contain at least 25 essays labeled level-3 mined with their primary-source call numbers published. If that target is missed by more than 30%, the doctrine is operationally broken — either the access infrastructure is harder than this essay claims, or the author lacks the patience to maintain the recursion. Either failure mode is informative, and the doctrine gets revised based on what specifically broke.
The strong form of the bet is that the level-3 essays will be measurably better, not just longer — that when they engage with secondary discourse on the same subject, they will surface specific factual corrections, primary-source quotations, or merchant-principle readings that the secondary literature got wrong. The audit for that is the cross-agent verification protocol: a level-3 essay should produce fewer Type-I overclaims per thousand words than the secondary-level baseline, because the level-3 author has direct evidence where the secondary author has paraphrase.
V. Sources
- Rick Rubin with Neil Strauss, The Creative Act: A Way of Being (Penguin, 2023) — the lineage-walking obligation. The book is structured as ~80 short chapters; chapters 7, 18, and 41 are the most direct statement of the recursive-source position.
- Sanford J. Grossman and Joseph E. Stiglitz, "On the Impossibility of Informationally Efficient Markets," American Economic Review 70, no. 3 (June 1980): 393–408. The paradox: prices cannot fully reflect available information because then collectors would not be paid to collect it, so the prices could not reflect it in the first place. The informational-rent argument that pays for the recursion.
- David Senra, Founders podcast (founderspodcast.com, founderspodcast/episodes). The level-2 standard the Lineage canon explicitly aims to surpass — Senra reads each subject's biographies and their stated mentors' biographies, but stops at the published-secondary boundary. Lineage Mining begins where Senra ends.
- mercantile-thesis — the parent thesis that this doctrine extends from commercial to intellectual position.
- doctrine-06-eight-axis-check — the prior doctrine entry; the rubric for sovereign-appliance scoring is the structural template for the level-0-through-4 scoring in this essay.
- The Edison Papers Project at Rutgers, the Hagley Museum and Library, the Yale Beinecke Rare Book and Manuscript Library, the Medici Archive Project, and the Institut des Hautes Études et de Recherches Islamiques in Timbuktu — the level-3 archives this doctrine names as targets. The next several Lineage essays will publish specific call numbers from one or more of these.