Speed Asymmetry

In the [YOU] on AI Field Guide

The cycle that began with [YOU] on AI documents speed asymmetry in real time. The Berkeley researchers observed work intensification, task seepage, and boundary dissolution within eight months of AI tool adoption. The SaaS market correction erased a trillion dollars of value in weeks. The capability that the orange pill recognizes arrived as a phase transition—sudden, visceral, irreversible. The institutional response is arriving at the speed that institutional responses always arrive: slowly, fitfully, shaped by the distribution of power among the constituencies affected. The gap between those two speeds is the central human experience of the AI transition, and it is the gap that Basalla's framework makes analytically visible rather than merely felt.

The cycle's interest in Geoffrey Moore's whole product framework maps directly onto speed asymmetry. The reason the whole product does not yet exist for AI in healthcare, education, and the public sector is not a technical failure. It is a speed asymmetry: the capability arrived faster than the institutional infrastructure needed to make it deployable in those contexts. The regulatory clearances, the liability frameworks, the professional identity reconstructions, the workflow redesigns—these are institutional artifacts, and institutional artifacts evolve at institutional speed. The generic product is extraordinary. The whole product is years behind. The gap is speed asymmetry made concrete.

The most actionable implication of speed asymmetry for the cycle is that the appropriate human response is not to wait for the institutions to catch up passively but to actively build the selection environment that will receive the technology. Every educational norm about AI use, every organizational practice about protected thinking time, every regulatory framework being debated in legislative chambers is an institutional artifact being constructed in the gap between variation speed and selection speed. The people who build those artifacts—however unheroic, however anonymous—are the ones who will determine whether the gap produces a Luddite catastrophe or an electrification recovery.

Origin

Speed asymmetry is not a term Basalla used; it is an inference from his framework. Basalla documented the gap between technological variation and institutional selection across hundreds of cases in the history of technology, and in every case the gap was the space where the transition's human cost was concentrated. The power loom arrived in 1812; the Factory Acts arrived in the 1830s and 1840s. The gap was two decades. The industrial electrification of factories began in the 1890s; the eight-hour day became widespread in the 1920s. The gap was three decades. The personal computer arrived in the early 1980s; educational systems began systematically integrating it in the 1990s and 2000s. The gap was a decade.

The AI transition has compressed the timeline for variation by an order of magnitude without compressing the timeline for institutional selection. This compression is the novel feature of the present moment that Basalla's framework, built on prior transitions, did not anticipate. The pattern is the same; the speed is unprecedented. Whether the institutional selection speed can be accelerated to narrow the gap, or whether the gap will widen until its human cost forces a reactive rather than proactive institutional response, is the open strategic question of the transition.

Key Ideas

Variation speed vs. selection speed. Technological variations are produced at the speed of engineering: months from prototype to deployment, weeks from capability threshold to mass adoption. Institutional selection environments are built at the speed of democratic deliberation and political struggle: years from problem recognition to legislative response, decades from legislative response to cultural normalization. The gap between these speeds is not an accident or a failure. It is a structural feature of the evolutionary process. But the magnitude of the gap is variable, and the AI transition is producing the largest magnitude gap in the history of technology.

The accumulation of human cost. The gap between variation speed and selection speed is not a neutral space. It is the space in which human cost accumulates—in the form of work intensification, occupational displacement, skill devaluation, and the dissolution of the institutional structures that previously protected workers, students, and communities from the worst consequences of technological change. The cost is not distributed evenly: it falls on the people with the least institutional power to accelerate the selection response. The Luddites bore the cost of the 1812 gap. The question is who will bear the cost of the 2025 gap.

Speed asymmetry and the whole product gap. The whole product gap that Geoffrey Moore identifies—the distance between AI's generic capability and the institutional infrastructure needed to make it deployable in specific contexts—is speed asymmetry rendered in product terms. The capability arrived fast; the infrastructure is arriving slow. The healthcare, education, and public sector domains where AI could generate the most social value are precisely the domains where the whole product requires the most institutional construction, and institutional construction is the thing that cannot be accelerated simply by spending more engineering hours.

Narrowing the gap through proactive institution-building. Basalla's framework implies that the gap is not fixed. Prior transitions narrowed it through sustained institutional effort: labor organizing, regulatory advocacy, educational reform, cultural norm-setting. The people who built the Factory Acts, the eight-hour day, and the union movement were performing institutional construction in the gap between variation speed and selection speed. The AI transition requires the same kind of construction, at unprecedented speed, in the domains where the gap is currently widest: labor law, AI liability frameworks, educational curricula, professional certification standards for AI-assisted practice, and organizational norms about the conditions under which human judgment must remain primary.