Backwash and Spread Effects
Every center of economic growth produces two kinds of effects on the regions that surround it. Spread effects carry benefits outward: demand for suppliers' goods, transfer of knowledge, expansion of markets, the rising tide that optimists celebrate. Backwash effects pull resources inward: talent migrates toward opportunity, capital flows toward higher returns, institutions adapt to serve the center's priorities rather than the periphery's needs. Myrdal's empirical finding, documented across multiple continents and decades, was that backwash tends to dominate in the absence of deliberate intervention. The AI economy replicates this dynamic at digital speed. Spread effects are visible — tools available to anyone with connectivity — while backwash effects operate through talent extraction, tool optimization toward center-priorities, and the capture of surplus by capital concentrated at the center.
The Material Prerequisites of Extraction — Contrarian ^ Opus
There is a parallel reading that begins not with the flow of talent and value but with the physical infrastructure that enables any digital economy to function. The server farms consuming electricity equivalent to small nations, the submarine cables requiring naval protection, the rare earth mines in Congo and Bolivia — these material realities reveal backwash and spread as second-order effects of a more fundamental extraction. Before talent can be siphoned from Bangalore to San Francisco through remote work, lithium must be extracted from the Atacama, processed in China, assembled into devices in Shenzhen, powered by grids that burn coal in some places and harness wind in others. The geography of this material substrate tells a different story about center and periphery.
The populations most affected by AI's reshaping of work experience neither spread nor backwash but something more immediate: the disappearance of entire categories of livelihood without corresponding appearance of alternatives. The call center worker in Manila whose job evaporates doesn't experience "backwash" — there is no flow to trace, only absence. The content moderator in Nairobi training the very systems that will replace her participates in her own obsolescence. From this vantage point, Myrdal's framework assumes a stability that AI disrupts: there must be recognizable centers and peripheries for flows between them to be measured. But when the center becomes fully abstracted — a distributed network of servers, algorithms, and capital with no fixed geography — and the periphery becomes everyone whose labor can be simulated, the analytical frame itself requires reconstruction. The question isn't how value flows but how the infrastructure of value creation is being fundamentally rewired.
In the AI Story
The mechanism of backwash is not mysterious. Talent moves to where talent is rewarded. Capital moves to where returns are highest. Institutions evolve to serve the interests of the powerful, who are concentrated at the center. Each movement reinforces the others in a self-sustaining spiral. The periphery loses the very people and resources that would have enabled it to develop capacity on its own terms — a loss that spread effects, however real, are insufficient to offset because they are diffuse, indirect, and do not counteract the concentrated, direct, self-reinforcing dynamics of extraction.
The AI economy amplifies backwash through three specific mechanisms. First, AI tools make talented individuals in the Global South immediately visible to recruiters and platforms in the developed world, redirecting their talent toward center-priorities rather than local needs. Second, the remote-work revolution means talent extraction no longer requires physical migration — a brilliant engineer in Bangalore works for a San Francisco company while generating value that accrues to California shareholders. Third, a novel form of cognitive extraction operates through the tools themselves: every interaction with Claude Code in Lagos contributes data that improves the model, with the improvement owned by the company at the center rather than the contributor at the periphery.
Segal's account of the Trivandrum training illustrates simultaneous spread and backwash with inadvertent precision. The spread effect is real: engineers gained capability that transforms their individual productivity. The backwash effect is equally real: the twenty-fold productivity gain accrues first to the company headquartered in the United States, serving a global market, generating revenue that flows to shareholders in the developed world. Segal's choice to keep and grow the team rather than convert productivity into headcount reduction is admirable but contingent — dependent on specific values at a specific company, not on institutional structures that would reliably replicate the choice across the economy.
Myrdal's insistence on holding spread and backwash in the same analytical frame distinguishes his approach from both the triumphalist narrative (which sees only spread) and the catastrophist narrative (which sees only backwash). The framework refuses to resolve the tension by choosing a side, because both forces operate simultaneously, and the question of which dominates in any given context is determined not by the technology but by the institutional environment within which the technology is deployed.
Origin
Myrdal introduced the concepts in Economic Theory and Under-Developed Regions (1957) and elaborated them in Asian Drama (1968). The empirical basis came from his extensive fieldwork across South and Southeast Asia, where he documented how industrial development in regional centers systematically drained talent and capital from surrounding peripheries. The framework became foundational to development economics and heterodox institutional analysis, providing the analytical vocabulary through which persistent global inequality could be described as structural rather than transitional.
Key Ideas
Two forces, one frame. Spread and backwash operate simultaneously; analyzing one in isolation produces systematic distortion.
Backwash dominates absent intervention. The gravitational pull of the center is structurally stronger than the diffusive force of the spread.
Digital extraction without migration. AI enables talent capture that does not require physical relocation — the periphery develops the capacity, the center captures the return.
Cognitive raw material. User interactions improve the models, with improvements owned by the center — a novel extractive form with colonial structural parallels.
Tool optimization as hidden backwash. Models are trained on center-language data and optimized for center-workflows, making tool performance a subtle form of advantage concentration.
Appears in the Orange Pill Cycle
Scales of Analysis and Power — Arbitrator ^ Opus
The validity of each perspective depends entirely on which scale and timeline we examine. At the scale of individual careers over 2-5 years, Edo's account dominates (90/10) — the Trivandrum engineers genuinely gained capability that transforms their market position, and spread effects through accessible AI tools create real opportunities. At the scale of regional development over 10-20 years, the contrarian view gains ground (65/35) — the material prerequisites of AI infrastructure and the replacement of entire job categories reshape economic geography more fundamentally than talent flows. The question "where does value accumulate?" yields different answers than "what happens to specific workers?"
When we examine mechanisms of change, both views prove partially correct but incomplete. Myrdal's spread/backwash framework accurately captures flows between existing centers and peripheries (Edo 70/30), but misses the more radical possibility that AI dissolves the stability of center-periphery relations themselves (Contrarian 60/40). The physical infrastructure of computation creates new geographies of extraction that don't map onto traditional development patterns, while simultaneously the death of distance through remote work scrambles conventional patterns of brain drain. Neither view alone explains why some regions might leapfrog development stages while others face wholesale economic exclusion.
The synthesis requires holding three dynamics simultaneously: traditional spread/backwash effects (which remain real), the material base of digital infrastructure (which creates new forms of dependency), and the dissolution of stable center-periphery geography (which scrambles both preceding dynamics). Perhaps the proper frame isn't about flows between fixed positions but about who controls the infrastructure that determines whether positions exist at all. The question shifts from "how does the center extract from the periphery?" to "who decides what constitutes a viable economic position in an AI-mediated economy?"
Further reading
- Gunnar Myrdal, Economic Theory and Under-Developed Regions (Duckworth, 1957)
- Gunnar Myrdal, Asian Drama, vol. III (Pantheon, 1968)
- Raúl Prebisch, The Economic Development of Latin America and Its Principal Problems (ECLA, 1950)
- Immanuel Wallerstein, The Modern World-System (Academic Press, 1974)