The Stickiness Paradox
Productive knowledge is sticky. It does not flow freely from place to place the way capital does. It adheres to specific social, institutional, and cultural arrangements and resists transfer. This stickiness is not a deficiency; it is a structural feature. Knowledge is sticky because the most valuable knowledge is tacit — embedded in context, inseparable from the specific conditions in which it operates. AI addresses the transfer problem for codifiable knowledge with unprecedented effectiveness. But here the paradox emerges: the smoothness that makes knowledge easy to access is precisely what makes it difficult to embed. The struggle of making knowledge work in a new environment — the repeated failures, corrections, adaptations — is not an obstacle to transfer. It is the mechanism of transfer. Remove the friction and the knowledge passes through without settling.
In the AI Story
The paradox converges with the critique Byung-Chul Han developed from an entirely different direction. Han argues that contemporary culture's drive to eliminate friction produces surfaces easy to traverse but impossible to grip — the smooth experience offers no resistance and therefore no engagement with the terrain. Hidalgo arrives at the structurally identical conclusion through information theory: knowledge flowing without resistance does not settle, because settling requires the friction through which the knowledge is tested, adapted, and integrated into the receiver's understanding.
Three dimensions of stickiness are particularly relevant to the AI moment. Interpersonal stickiness — the difficulty of transferring knowledge between individuals — has been partially addressed by AI's universal interface. But the tacit knowledge one individual holds — judgment, contextual awareness, embodied intuition — remains interpersonally sticky and transfers only through sustained, context-sensitive interaction no interface provides. Organizational stickiness — the difficulty of transferring knowledge between organizations — has been partially addressed for the codifiable layer, but a firm's institutional knowledge, its coordination mechanisms, its unwritten rules remain organizationally sticky. Geographic stickiness — the difficulty of transferring knowledge between locations — has been most dramatically reduced by AI; but the knowledge of what works in Lagos, what customers need, what institutions enable, remains geographically sticky.
The paradox produces a specific prescription: not the elimination of friction, which is happening already and cannot be reversed, but the relocation of friction to the level where it is productive. Not acquisition friction, which AI has rightly eliminated — the years of study, the institutional barriers that kept productive knowledge locked in specific populations. But embedding friction — the deliberate, structured processes through which accessed knowledge is tested against local conditions, adapted based on local feedback, challenged by local expertise, and integrated into local understanding.
Education that requires students to explain AI-generated output, identify its assumptions, and evaluate its applicability to specific contexts reintroduces the friction of engagement. Organizational practices that require teams to review and challenge AI-generated solutions before implementing them reintroduce the friction of collective evaluation. Development programs that require local adaptation of AI-accessed knowledge reintroduce the friction of embedding. The acquisition layer is smooth; the embedding layer must have grip.
Origin
The stickiness concept entered development economics through Eric von Hippel's 1994 paper on information stickiness in innovation, and was extended by Gabriel Szulanski's work on internal best-practice transfer. Hidalgo synthesized these strands with his own information-theoretic framework to produce a general theory of why productive knowledge resists transfer — and, crucially, why the resistance is functionally necessary rather than merely an obstacle to overcome.
Key Ideas
Friction is the mechanism of embedding. The struggle to make knowledge work in new conditions is not an obstacle to transfer — it is how transfer happens.
Smoothness prevents settlement. Knowledge that flows without resistance does not accumulate locally; it passes through the way water passes through a pipe.
Three stickiness dimensions persist unevenly. AI reduces geographic and interpersonal stickiness for codifiable knowledge; tacit stickiness remains in all dimensions.
Relocate friction rather than eliminating it. Remove acquisition friction; preserve embedding friction through educational, organizational, and institutional design.
Access is not accumulation. AI democratizes access to productive knowledge; crystallization into durable local capability requires the friction AI eliminates.
Debates & Critiques
Triumphalist readings of AI argue that the stickiness paradox overstates the importance of embedding — that access alone is transformative and societies will develop new modes of capability that do not require the old friction-based mechanisms. Hidalgo's response is empirical: the history of technology transfer programs is littered with cases where access without embedding produced short-term output and long-term dependency, and there is no reason to believe AI will be exempt from this pattern unless deliberate institutional investment in embedding accompanies the access.
Appears in the Orange Pill Cycle
Further reading
- von Hippel, Eric. "Sticky Information and the Locus of Problem Solving" (Management Science, 1994)
- Szulanski, Gabriel. Sticky Knowledge (2003)
- Hidalgo, César. Why Information Grows (2015)
- Han, Byung-Chul. Saving Beauty (2017)