Why Scientists Can't Rebuild a Polaroid Camera [César Hidalgo]
TL;DR
César Hidalgo argues that knowledge follows three fundamental laws governing its growth, diffusion, and valuation, emphasizing that knowledge is collective, non-fungible, and embodied in networks rather than individuals or books—with critical implications for why development efforts fail and how organizations actually learn.
🧠 The Nature of Knowledge 3 insights
Non-rival but non-fungible properties
Knowledge can be shared without depletion (non-rival) unlike physical tools, but it cannot be aggregated like commodities because it consists of unique, categorical components similar to words in language—requiring representations more akin to machine learning than traditional economics.
Collective embodiment over individual possession
Knowledge is not contained in books or manuals but is situated in teams, organizations, and networks; you cannot throw engineering manuals into a gorge and build a bridge because knowledge requires dynamic, embodied social interaction to function.
Three distinct knowledge types
Factual knowledge (easily transmitted facts), conceptual knowledge (integrative stories that connect facts), and procedural knowledge (how to perform tasks like DNA sequencing) operate differently in economic growth and innovation.
📈 Laws of Growth and Learning 3 insights
Power law learning at individual/team scale
Learning curves for individuals and firms follow power laws (square root functions) where rapid initial improvement plateaus over time, documented in studies from typewriting classes (Thurstone, 1916) to Liberty ship production during WWII.
Exponential growth at industry scale
When transitioning from firm-level to industry-level knowledge accumulation, growth shifts qualitatively from power law to exponential (Moore's curve), representing different mathematical mechanisms of collective learning.
Experience dominates documentation
Shipyard studies show efficiency gains came purely from accumulated production experience rather than technology upgrades, capital investment, or written procedures—highlighting why tacit knowledge cannot be captured in wikis or manuals.
🌐 Diffusion and Organizational Architecture 3 insights
Distance and relatedness constraints
Knowledge diffuses more effectively over shorter distances through social networks and moves more easily between related activities (the principle of relatedness), explaining why geographic clustering and industrial ecosystems emerge predictably.
Architectural innovation disrupts incumbents
Changing how components interact—like Barnes & Noble's retail model versus Amazon's fulfillment centers, or propeller versus jet aircraft—requires complete organizational redesign rather than component replacement, explaining why established firms often fail during technological transitions.
Network reconfiguration as learning mechanism
Organizations learn by reconfiguring networks of people, tools, and ideas (adjusting 'weights' in the organizational graph) rather than just individual education, as when employees shift roles or teams based on compatibility and capability.
⚠️ Economic Development Failures 2 insights
Boneheaded development policies
Many economic development projects fail by treating knowledge as a fungible quantity that can be stockpiled in 'science parks' or 'knowledge cities,' ignoring that knowledge cannot be treated like capital in a barrel and requires specific network conditions to grow.
Democratic knowledge beyond academia
Economic knowledge includes tacit, experiential wisdom from car mechanics and bakers—not just scientifically validated truths—meaning economies function through distributed practical expertise that cannot be centralized or rapidly trained.
Bottom Line
Economic development and organizational competitiveness depend on treating knowledge as a collective, experiential, and non-fungible resource that grows through network reconfiguration and accumulated practice, rather than as a transferable commodity that can be documented, centralized, or rapidly scaled through capital investment alone.
More from Machine Learning Street Talk
View all
He won a Nobel here for AlphaFold. Then he left. - John Jumper
Nobel laureate John Jumper explains how AlphaFold solved the 50-year protein structure prediction problem by collapsing years of experimental work into minutes, while emphasizing its narrow scope as a starting point for biological research rather than a universal model of life.
The Ex-Congressman Who Says AI Isn't Unstoppable — Brad Carson
Former Congressman and Pentagon official Brad Carson argues that AI development is not inevitable and can be controlled through strategic regulation, particularly by treating AI as products subject to liability laws rather than granting them human rights, while leveraging chip controls and mandatory testing to shape the technology's future.
Inference, not prediction — Prof. Michael I. Jordan on what modern AI is still missing
Professor Michael I. Jordan critiques the hype around AGI and prediction-based LLMs, arguing that modern AI lacks economic and social thinking; he advocates for 'inference' systems grounded in game theory and market dynamics that respect human agency and create collective value.
Inference, not prediction — Prof. Michael I. Jordan on what modern AI is still missing
Professor Michael I. Jordan critiques AGI as harmful PR that demoralizes young engineers, arguing that true intelligence requires economic and social systems thinking—treating billions of humans as agents in value-creating ecosystems—rather than isolated prediction engines built without intellectual foundations.