The Next Fifteen Years

A forecast built from first principles
Section future / 07-indicators / diffusion / README.md

B - Diffusion Indicators#


Baselines as of mid-2026. Cadence: quarterly to annual.

These test the central timeline claim: that from ~2028 the binding constraint is institutional absorption, not model capability (Part IV). They are harder to read than the substrate indicators because most of the available data is self-reported and therefore biased in the direction Game 4 identifies.

Rule for this family: weight instrumented and revealed-preference measures above surveys, always.

The twelve indicators split on what they watch. The labor and institutions group tracks the people side - who gets hired, how skill is certified, what the tax base does, and where the welfare channel moves. The economic ledger group tracks whether the technology shows up in the numbers firms and statisticians keep - productivity, margins, pricing models, and the price level itself. The capability frontier group tracks the leading edge - how long agents run unsupervised, whether the verifiable/unverifiable gap holds, whether the research loop compresses, and whether any of it reaches atoms.

#IndicatorPageTests
B1Entry-level : senior posting ratioLaborGame 4, Uncertainty 3
B2Measured productivityLedger2028–2032 J-curve
B3Firm-level margin dispersionLedgerGame 3 surplus capture
B4Agent reliabilityFrontier2026–2028 transition
B5Outcome-priced contractingLedgerCapital revenue case
B6Assessment reversionLaborGame 5
B7Two-economy price splitLedgerPrices
B8Verifiable vs unverifiable gapFrontierUncertainty 5
B9Research cycle time (RSI panel)FrontierUncertainty 1
B10Labor share of tax baseLaborFiscal squeeze
B11Meaning and relationship channelsLaborMeaning
B12Physical diffusionFrontierRobotics, Part V row 5

Read the family jointly, not row by row. The highest-information patterns are cross-group: B1 moving with B10 and B7 is the AI-relevant fiscal signature; B4 moving without B12 is the cognitive/physical fork confirming; B5 moving before B2 is revealed belief running ahead of measured fact, which is exactly the order every prior general-purpose technology showed.

Contamination hierarchy#

PreferDistrust alone
Outcome-priced contracts (B5), margin dispersion (B3), component CPI (B7)Vendor NPS, "AI adoption" surveys
Profession-level posting ratios (B1)Blended "knowledge work" employment
Teleop→autonomy + delivered $/hour (B12)Unit sales, demo videos
Production deployment depth (B8)Benchmark gap closes

Surveys are allowed as secondary series when they name the same object as an instrumented measure and move with it. Alone they are marketing.

A×B joint (reminder)#

B stall + A slack → institutions. B stall + A tight → substrate. Do not re-score Part IV from B alone. → substrate


Related: Game 3 · Game 4 · Part IV - Timelines · Part IX

View markdown source

select · Enter open · Esc close