brain/
sourcestock-market

Distilled understanding of technology adoption S-curves — an evaluation lens for stock-market signals

Thread snapshot: how S-curve position (remaining runway), curve steepness (adoption velocity), and the cost-curve tipping point can sharpen how stock-market evaluates technology-driven causal chains — plus the hard caveat that cost is forecastable but adoption timing is not.

Source

Distilled understanding of technology adoption S-curves — an evaluation lens for stock-market signals

Distilled from thread vault/threads/technology-adoption-s-curves on 2026-06-16. This source captures the thread's current understanding as of the distillation; re-distill if the thread has significantly advanced since. Focus: full thread scope, framed for use as a stock-market signal-evaluation lens.

Summary

Technology adoption follows an S-shaped (logistic) curve, driven underneath by a Wright's-Law cost curve. For a project that already surfaces tradeable causal chains, the S-curve adds an orthogonal evaluation layer the current conviction model omits: how much of the move is left (curve position vs. ceiling), how fast it arrives (curve steepness), and what dates the knee (the cost/performance tipping point). The single most important caveat for trading use: cost curves are forecastable; adoption timing and the eventual ceiling are not — so position and velocity should be expressed as ranges with explicit falsifiers, and the lens must be applied per-segment (solar ≠ EV-battery ≠ robotaxi), never per-theme.

Key entities

  • tony-seba — the popularizer of cost-curve disruption. His record is the cautionary spine of the whole lens: his cost-curve forecasts (solar cheapest power, ~$100/kWh batteries) were right and early; his adoption-timing forecasts (95% of US miles via robotaxi TaaS by 2030) missed badly. "Right spot, wrong timeframe."
  • rethinkx — Seba's think tank; the fully-worked framework (converging cost curves → S-curve from ~1-2% to ~80%, incumbents underestimate by extrapolating linearly). Also a caution on source incentive — an advocate, so its forecasts are directional arguments, not neutral base cases.
  • clayton-christensen — disruptive-innovation theory; useful for the direction of a disruption (low-end/new-market foothold climbing up-market), contested as a predictive tool.

Key concepts

The adoption S-curve and how to read position — technology-adoption-s-curve

Adoption is sigmoidal: slow under ~1-2% penetration, a non-linear acceleration past a tipping point, deceleration toward an ~80% ceiling. The logistic model y = L/(1 + e^(−k(x−x₀))) gives three parameters that map onto three trading questions — L (ceiling: how big), k (steepness: how violent the re-rate), x₀ (inflection: before or after the knee). The same S-curve arises from several mechanisms (information epidemics, firm heterogeneity, density dependence), and real curves are often asymmetric, so symmetric fits can misplace "how much is left." (Well-evidenced across the canon and the practitioner framing.)

Wright's Law is the engine — and the forecastable part — wrights-law

Unit cost falls a roughly constant % per doubling of cumulative production (median learning rate ~21% across 150 technologies). It is the cause beneath adoption: cost crosses a parity threshold → the new technology beats the incumbent → adoption accelerates. Live exemplar: AI inference ("LLMflation") fell ~1,000× in three years at fixed capability. Two caveats that matter for sizing: learning rates change for ~66% of technologies ("past learning ≠ future learning"), and falling per-capability cost can coexist with rising frontier spend (AI). (Strongly evidenced.)

Curve steepness is set by feedback — and is rising — curve-steepness-and-adoption-velocity

"Some S-curves are steeper than others." Steepness k is governed by learning curves, economies of scale, network effects, and lock-in (Bass imitation coefficient q). Time-to-50%-penetration has compressed across eras (telegraph 56y → smartphone 5y → AI ~3y), so modern curves re-rate faster and the window to take a position is shorter. The strength/durability of these feedback mechanisms is the leading indicator of a steep curve and a defensible eventual share. (Well-evidenced.)

Cost forecastable, timing not — the core asymmetry — forecastability-of-technological-progress

Wright's Law beat five rival laws across 62 technologies, with forecast error growing predictably (~2.5%/yr of horizon); Farmer & Lafond turn this into a probability that one technology out-costs another by a future date. But the Bass ceiling M and coefficients p,q are badly identified, and learning rates shift — so adoption timing and the ceiling are the unreliable parts. Anchor conviction on the cost curve; treat the inflection date as a distribution. (Well-evidenced; the asymmetry is the load-bearing insight for trading.)

Quantifying adoption — bass-diffusion-model and diffusion-of-innovations

Bass decomposes adoption into innovation (p) and imitation (q) bounded by market potential M; q/p flags network-effect steepness. Rogers' adopter segmentation (innovators 2.5% → early adopters 13.5% → early majority 34% → …) lets a penetration % be mapped to a curve stage. The recurring warning: M (the ceiling, which sets remaining runway) is the hardest parameter to estimate and is highly sensitive to assumptions. (Well-evidenced.)

Convergence — technology-convergence

Seba's claim that disruptions fire when several independently-improving cost curves intersect (the smartphone example). For a portfolio this is the qualitative case for treating reinforcing theses as a correlated cluster, not independent bets — relevant to how exposure is budgeted. (Single-framework-sourced; directional.)

Open questions from the thread

  • s-curve-position-in-stock-evaluation — the direct bridge: how curve position, steepness, and the cost tipping point should be encoded into mechanism pages and the Signal contract. This is the Phase-2 design question this distill is meant to seed.
  • detecting-s-curve-inflection-in-real-time — can the knee be spotted prospectively? This is the hard part and bounds how much weight the lens can bear.
  • adoption-underestimated-or-overhyped — for a given name right now, are we in the structural-underestimation regime (the edge) or near a hype-cycle peak (the trap)? Segment-specific.

Relevance to stock-market

This project surfaces ideas by tracing causal chains to tradeable beneficiaries; the S-curve lens does not replace that — it is an evaluation layer that answers questions the current conviction model (which scores whether a chain is true, per-step) leaves open: how much of the move is left, how fast it arrives, and when the knee dates. A chain can be fully confirmed yet already played out near saturation (little remaining upside), or only partial yet sitting right at the inflection (large remaining upside if it confirms). Those are orthogonal to evidence-strength and are exactly where the lens earns its keep.

Three concrete encodings, matching the angles prioritized for Phase 2:

  1. Remaining-runway on upside. Tag a mechanism with the beneficiary technology's curve position (penetration % vs. an honest range for the ceiling L). This qualifies valuation.base_case_upside_pct and position sizing — buying near the inflection captures the steep part; buying near saturation does not. The asset-pricing literature backs the direction: adoption lags convert "growth options" into "assets in place," so the risk premium (and expected return) is structurally higher earlier on the curve.
  2. Velocity → conviction/horizon. Curve steepness k informs the Signal classification.horizon and the imminence/magnitude of the re-rate — a steep, network-effect-driven curve argues for a nearer horizon and a larger move; a shallow one warns you are early and bleeding time-decay.
  3. Cost-curve tipping point as catalyst. A Wright's-Law parity crossing is a datable forcing function — a catalysts[] entry (and possibly a new KNOWN_CLUSTERS value for "cost-curve disruption") so the trader can reason about the knee as a trigger rather than a vibe.

The discipline this lens imposes on the project, straight from the evidence: weight cost-curve evidence heavily (it's the forecastable part), but never treat an adoption date as confirmed — Seba's record and the IEA's repeated solar underestimation show consensus is biased against steep cost curves over the long run, while near-term timing is routinely over-hyped (Amara's Law). Encode adoption timing as a range with explicit falsifiers, and read the regime per-segment: in 2026 solar/storage are beating forecasts while EV battery demand and robotaxi adoption are undershooting.

The canonical chain this implies is filed in the thread as cost-curve-tipping-point-to-s-curve-adoption — when this source is ingested, that mechanism is the natural template for an S-curve-aware evaluation field on stock-market mechanism pages.

Provenance

Referenced by
Concepts
S-Curve Evaluation Lens
Questions
How should S-curve position factor into how we evaluate stocks?