Lobbying operates within the existing knowledge framework. Epistemic strategy shapes the framework itself—what counts as knowledge, who can produce it, which questions are legible as policy questions. Standard capture is a chess game where industry buys better players. Epistemic strategy is industry redesigning the chessboard.
Canonical Moments from Research
Description
Robert Kehoe establishes that the burden of proof lies on opponents to prove harm, not on industry to prove safety. This explicitly inverts the epistemological ground rules of regulation itself.
Impact
Prevented regulatory action for 40+ years. Kehoe became essentially the only researcher producing authoritative knowledge about lead toxicity, concluding current levels were safe.
Sources
Markowitz & Rosner (2002), Needleman (1997)
Description
"A Frank Statement to Cigarette Smokers" and establishment of TIRC. Industry pledges to fund independent research but creates parallel knowledge institution to contest existing science.
Impact
Made uncertainty itself the product. Distributed doubt proved more resilient than Kehoe's monopoly—one Patterson could break the lead case, but tobacco's parallel institutions sustained doubt for decades.
Sources
Oreskes & Conway (2010), Brandt (2007), Proctor (2011)
Description
American Petroleum Institute strategy documents explicitly frame uncertainty as a political objective. Later shift from contesting science to making action contingent on economic cost.
Impact
Rendered science politically inert by shifting debate to economic terrain where industry holds advantage. Climate action only passed when reframed as economic opportunity (IRA). "When I think climate, I think jobs."
Sources
PBS Frontline (2022), Supran & Oreskes (2017), Brulle (2014)
Description
Legal-institutional infrastructure (trade secret law, IP protections, computational barriers) makes independent knowledge production structurally impossible. The object of study itself is inaccessible.
Impact
Qualitative break from previous cases. Patterson could measure lead in ice cores because lead isn't proprietary. No independent researcher can measure what proprietary AI models do because model weights are trade secrets. The system performs the Kehoe function automatically.
Sources
Crawford (2021), Zuboff (2019), Whittaker (2021)
Description
Collectively produced data (search queries, social media posts, behavioral patterns) is processed through proprietary systems and the output becomes corporate IP inaccessible for democratic governance.
Impact
Primitive accumulation of knowledge. The public creates the raw material but cannot access the resulting knowledge. This is doubly illegitimate—not only can the public not see what systems do, the systems were built from what the public involuntarily provided.
Sources
Federici (2004), Harvey (accumulation by dispossession)
Stage 1 (Lead): Monopolize scientific authority. Be the only researcher; prevent consensus from forming. Object of study: accessible.
Stage 2 (Tobacco): Manufacture doubt against existing science. Create parallel institutions; make uncertainty the product. Object of study: accessible.
Stage 3 (Fossil Fuels): Concede science, shift debate to economic terms. Render science politically inert; win on cost terrain. Object of study: accessible.
Stage 4 (Tech): Control existence of knowledge itself. Own the object of study; make independent knowledge structurally impossible. Object of study: inaccessible.
The first three stages operate on the relationship between industry and independently producible knowledge. Stage 4 is qualitatively different—the industry controls reality's accessibility. What began as improvised corporate tactics has become embedded in the institutional infrastructure of American capitalism.