LL2-19 digest — Ch19 Hungry for innovation: from GM crops to agroecology#

Late lessons from early warnings II (EEA 2013), Part C “Emerging issues”, pp. 458–485. Authors: Quist, Heinemann, Myhr (biosafety scientists at GenØk and Canterbury), Aslaksen (ecological economist) and Funtowicz (post-normal science). Quist was also on the report’s editorial team, and Funtowicz on its advisory board (p. 5). No panels or replies. The report says case-study chapters were peer reviewed, but reviewers are not identified (p. 5).

Core argument#

This is less a classic late-lessons case than a contrast between two innovation “pathways” for agriculture: - “Top-down” technology transfer, exemplified by GM crops. - “Bottom-up” participatory innovation, exemplified by agroecology.

The central claim is that innovation policy framed around competitiveness and intellectual property rewards innovations that can be “packaged and sold” to the largest markets and “largely bypass the poor” (p. 460). The result is an “unavoidable” bias toward proprietary seed products (p. 465). Knowledge-intensive, site-specific management innovations are starved of investment, however valuable (p. 466). Without governing the direction of innovation, “current indications are that we will design agriculture to fail” (p. 458).

Key evidence#

• Narrow delivery. After about 30 years of R&D and about 20 of commercialisation, only two traits matter commercially: herbicide tolerance and insect resistance. Five countries account for 91% of GM area. The 159 million ha in 2011 is framed as “just 3 %” of world agricultural land (p. 463; Figs 19.1–19.2, pp. 464–465).
• Patchy adoption (pp. 464–465). The chapter gives three reasons: rejection by high-income consumers, who get no direct benefit; developed-country subsidies undermining developing-country markets (OECD support “almost USD 1 billion per day”); and high seed and input costs. Poor countries that follow the export model risk a “loss leading treadmill”.
• Box 19.1, herbicide-tolerant crops (pp. 462–463). Glyphosate-resistant weeds create a “treadmill”, reversion to tillage and other herbicides, and “deskilling”. Stacked tolerance traits will deepen neglect of integrated weed management (Mortensen). The package does not fit smallholders. Yields owe more to conventional breeding than to GE traits. Transfer to poor farmers “appears to be” a top-down approach that “has not, and will not” benefit them, a verdict resting on one of Heinemann’s own papers.
• Knowledge pipeline (pp. 466–467, 470). R&D is shifting from public to private (the CGIAR budget is 12% of the six largest firms’ combined R&D). Public-sector patenting is read as a funding feedback loop. Other points: a “modest” anti-commons effect (Murray & Stern), seed concentration past a CR4 threshold, and UK POST’s warning of reduced incentives for conventional breeding.
• Box 19.2, “a late lesson case in the making?” (pp. 468–470). Risk assessment treats “no evidence of harm” as safety. It relies on developer-funded, confidential, underpowered studies, surrogate proteins, 90-day trials, variance-inflating comparators and weak monitoring. Material access is restricted (26 scientists’ statement), risk research is under-funded (1% of USDA biotech research, 1992–2002), and dissenting scientists face career risk. Health “indications” rely heavily on Séralini 2012. The box concedes “it is plausible that there simply are no effects to be found” (p. 468).
• Agroecology evidence (pp. 473–474). Badgley 2007 (organic matches yields once conversions mature); UNEP-UNCTAD (+116% across 114 African cases); Pretty (+79% across 286 projects; 2.13-fold across 40 African projects); farmer field schools; Rodale and Wisconsin trials.
• Water-stress case (pp. 474–475). GE drought tolerance will lag breeding and marker-assisted selection because the trait is multigenic. Soil and water management, which the chapter’s list opens with “locally-adapted drought tolerant varieties”, matter more but cannot be patented.

Authors’ lessons and recommendations#

• Top-down approaches “will most likely fail” on food security and poverty because they feed a “feedback cycle that concentrates resources, knowledge, and influence” (p. 477).
• Providers “will always serve last” smallholders on plots under 2 ha (p. 476).
• The emerging GM late lesson is primarily institutional, not harm-based: “the persistence of the same institutional patterns” as asbestos, benzene and BSE (p. 470). The same sentence asserts that “early indications of harm are just emerging”.
• Precaution belongs inside scientific risk assessment, through choices about error types and explicit statements of what is unknown (p. 470).
• Recommendations (pp. 475–477):
• Rebalance innovation toward the public good.
• Operationalise IAASTD and SCAR 2012, with agroecology given top funding priority.
• Build a public sector “free from political incentives for top-down innovation”.
• Prefer IP instruments that permit farmer innovation.
• Adopt the STEPS “directions, distribution, diversity” agenda.
• Scientists should take responsibility for how science is governed.
• The chapter does hedge: “the right mix of innovation is essential”, and seed-based and environmental approaches “both have value in all agroecosystems at all times” (p. 470). Footnote 1 concedes that “hypothetically” GM need not be high-input (p. 460). “Either pathway” could cut food costs and offer a way out of poverty (p. 461).

Main mechanisms#

Appropriability steering direction (pp. 460, 465–466); homogenisation of environments to fit products (p. 466); framing power (p. 477); resistance treadmills (p. 462); lock-in through deskilling and lost seed networks (pp. 462, 472); public research drifting to private logics (pp. 467, 471); concentration narrowing the pipeline (p. 467); developer-controlled evidence biased toward null results (pp. 468–469). Alternatives whose proceeds “flow to adopters rather than the providers” are politically weak (p. 476).

Transferable insights (technology-neutral)#

1. Ownership-tied incentives under-supply non-excludable, context-specific solutions (pp. 460, 465–466). Moderate.
2. Products needing uniform conditions serve large, homogeneous users first (pp. 461, 464–466, 476). Moderate.
3. Defining “innovation” is a distributive choice; powerful actors shape the problem to fit their solution (pp. 461, 477). Suggestive to moderate.
4. Broad promises narrow to what pays (pp. 463, 466). Moderate to strong.
5. Single-tactic control of adaptive systems produces treadmills (p. 462). Strong.
6. Simplifying technologies erode the skills and infrastructure that alternatives need (pp. 462, 472). Moderate.
7. Public research adopting private incentives crowds out public-good work (pp. 466–467, 470). Moderate / suggestive.
8. Developer-controlled, underpowered, confidential evidence biases toward “no evidence of harm” (pp. 468–470). Moderate.
9. Absence of evidence is not evidence of absence (p. 468). Strong as a principle.
10. Recurring institutional patterns as an early warning (p. 470). Suggestive; risks unfalsifiability.
11. Distributed-benefit alternatives lack political champions (p. 476). Suggestive.
12. Concentrated production can mean global exposure through supply chains (p. 468). Strong as a structural point, though asserted without exposure data.

Main caveats#

• Asymmetric scrutiny. GM claims are held to strict evidential standards. Agroecology claims, mostly uncontrolled before-and-after project syntheses and an advocacy-institute trial plus a scenario model, are accepted as “speak[ing] for themselves” (p. 474).
• Selective citation. NRC 2010, WHO 2005, Marvier 2007 and World Bank 2007 are cited for narrow points, while their broader, more GM-favourable findings go unmentioned. Pre-2013 evidence of Bt-cotton gains and organic yield gaps (Seufert 2012) is absent. [background]
• Contested health evidence. It rests on the Séralini group and NGO sources. [background] Séralini 2012 was criticised by EFSA in late 2012, before this report appeared, and retracted in 2013. The report’s editors mention it only as a spur to regulatory data transparency (p. 660).
• Framing and omissions. [background] The “3 %” figure uses a denominator that includes pasture; measured against cropland, the share would be roughly 10%. The text implies only 17 countries grow GM, while Fig. 19.1 calls them “the 17 largest producing countries”. The Badgley characterisation needs checking. The rival explanation for low adoption (regulatory and trade barriers) is ignored. The dichotomy is conceded as “artificial” (p. 465) yet used throughout, and the conclusions outrun the hedges.
• No adversarial voice. Editorial and advisory-board ties go undisclosed in the chapter. Self-citation is heavy (Heinemann; the GenØk network), and the lead author’s own contested-science history is not mentioned [background].
• Much held up. The treadmill, concentration, narrow-trait and absence-of-evidence points were well founded and largely borne out [background]. The strongest parts are the political economy of innovation and the epistemics of risk assessment.