H Heuristics Navigating a Changing World

H Heuristics Research Report · HH-2026-31

The Architecture of Creation and Development

Diffusion, convergence and the structural principles through which complex economies emerge

Flagship report. The mathematics of diffusion — first written down for salt dissolving in water — as a disciplined way to see technology spread, productivity convergence and the accumulation of capability whole.

Hunter Hughes · · 51 pages

Read the PDF opens in a new tab Download 6.0 MB

Abstract

Development is the process by which societies create structure: by accumulating capabilities, linking them through networks, and channelling flows of knowledge, energy and capital toward places where they are scarce. This report argues that the mathematics of diffusion, first written down for salt dissolving in water, offers a disciplined way to see that process whole. Across physical, biological, technological and economic systems, recurring structures emerge from a small set of generative rules: matter and energy flow down gradients; local interactions assemble into networks and hierarchies; and those structures in turn reshape the flows that created them. That recurring logic is called the architecture of creation.

Development economics can be read as the study of how these generative processes operate within societies: how innovations diffuse, how productivity converges, how capabilities accumulate, and how initially uneven systems reorganise toward new states of development. The bridge is a single well-tested relationship. In 1855 Adolf Fick proposed that the flux of a substance is proportional to its concentration gradient; the same form — flux = conductance × gradient — governs heat, electric current and groundwater. The report shows that it also organises a large body of evidence on technology adoption and economic catch-up, provided it is extended in specific, testable ways.

The report has three aims: to set out a unifying framework of constraints, flows, emergence and development that is precise enough to generate hypotheses rather than metaphors; to establish when concepts from physical diffusion legitimately transfer to economic diffusion and when they do not; and to translate the framework into a diagnostic for policy — a way of identifying which constraints most slow the movement of productive knowledge into an economy, and where intervention would yield the greatest return.

The central implication for decision-makers is to diagnose the binding resistance before adding force. Development strategies often push harder on the gradient with more technology, more capital and more pilots. Because resistances add in series, lowering a small one achieves little, while lowering the largest one can multiply the effective rate of diffusion.

Key findings

  1. One equation, many substrates. Fick’s law, Fourier’s law of heat conduction, Ohm’s law and Darcy’s law share one structure: a flow driven by a gradient and limited by a resistance. In development the gradient is the gap between frontier and local practice, and the conductance is the set of capabilities — skills, power, finance, connectivity, institutions — that lets knowledge move.
  2. Pure diffusion is too slow; development behaves like reaction–diffusion. Pure diffusion spreads with the square root of time, but technologies spread far faster because adoption also reproduces locally through imitation and learning, producing travelling waves and S-curves.
  3. Network architecture governs speed. Random networks spread simple information fastest, while ideas requiring social reinforcement spread reliably only through clustered networks with a few long-range bridges — which implies that who adopts first matters as much as how many.
  4. Arrival has converged; intensity has not. Mobile subscriptions in Sub-Saharan Africa rose from about 2 to 90 per 100 people between 2000 and 2025, while fixed broadband, which needs complementary infrastructure, remains near 1 per 100 against roughly 39 in high-income economies.
  5. Convergence is real but slow, and it has weakened. Across 191 economies poorer countries grew faster between 2000 and 2023 (β = −0.43), implying a convergence speed of about 0.45 per cent a year and a half-life of roughly 150 years; rolling ten-year estimates peak for windows beginning in 2006–07 and fade to statistical insignificance from 2012–14.
  6. Bottlenecks behave like resistances in series and weak links in a chain. When capabilities are complements, the weakest one binds: in 2024 about 657 million people still lacked electricity, roughly 88 per cent of them in Sub-Saharan Africa, and access in low-income countries was 49 per cent. No amount of digital ambition substitutes for a missing power supply.
  7. Resilience is part of the architecture. Shocks erase accumulated stock and recoveries are frequently incomplete, so a development strategy that raises conductance without protecting the accumulated stock will leak much of what it builds.
  8. The practical rule follows from the series structure: diagnose the binding resistance before adding force, because lowering a small resistance achieves little while lowering the largest one can multiply the effective rate of diffusion.

Contents

  1. —Executive Summary
  2. 1The Architecture of Creation
  3. 2From Physical Diffusion to Economic Diffusion
  4. 3The Mathematics of Innovation Spread
  5. 4The Architecture of Economic Convergence
  6. 5The Bottlenecks of Development
  7. 6A Unified Architecture of Development
  8. 7Testable Hypotheses and Research Agenda
  9. 8Conclusion
  10. —References
  11. —Appendix A · Technical derivations
  12. —Appendix B · Data sources and methods

Data and method

The report presents a research framework and an analysis of publicly available evidence. Empirical figures draw on public data from the World Bank World Development Indicators, the Maddison Project Database, the International Telecommunication Union and Our World in Data, retrieved in October 2026; simulations and illustrative figures are identified in their source notes, and Appendix B documents all sources and methods. Technical derivations appear in Appendix A. Illustrative models and simulations are intended to clarify mechanisms and are not forecasts for any specific economy. Section 2.5 states the conditions under which concepts from physical diffusion legitimately transfer to economic diffusion, and Section 1.6 what does not transfer across domains; Section 7 converts the framework into testable hypotheses.