MANY MINDED · THE COST CURVES · SUSTENANCE

What share of the world works in farming?

Agriculture share of world employment · % of employment · measured 1991–2025 · fitted live from 35 observations by the Many Minded cost-curve engine.

The last measured value is 25.8% (2025, OWID / ILO). Over the fitted window the series falls 1.0% a year (80% interval −1.8% to −0.3%) — a halving every 67.6 years (80%: 39.1–250.9). Carried forward on that fit, 2030: 24.5% (80%: 22.9%–26.1%).

25.8%latest measured (2025)−1.0%fitted, per year67.6yper halving35observations, 1991–2025measuredbasis

the curve

farm labor share — measured history and fitted projectionfarm labor share: measured 1991–2025 in % of employment, last value 25.8%, fitted at −1.0% a year, projected to 2033 inside an 80% interval. Logarithmic vertical axis.% of employment19911996200120062011201620212026203125.8% (2025)
Vertical axis: % of employment, log scale. Solid: measured observations. Dashed: the fitted central projection. Shaded: the 80% interval, which widens with horizon because shocks accumulate.

what this series measures

share of world employment in agriculture — the measured proxy behind the farm-autonomy gate: autonomy shows up as labor leaving the field.

what the fit says

quantityvaluehow it is computed
fitted rate−1.0%/yr (80%: −1.8% to −0.3%)Farmer–Lafond drift over the trailing window · n=11 points, 10 years
halving time67.6 years (80%: 39.1–250.9)implied by the fitted drift
curve checksteady (−1% → −2%/yr)first half of the record versus the second, judged against the direction that helps this metric (falling)

Not shown for this curve, because the machinery returns nothing: Wright's law (no cumulative-deployment series for this technology); a regime break (none clears the sup-F threshold).

the projection, with its interval

Log cost as a random walk with drift: the forecast variance grows with horizon (τ + τ²/m), which is why these bands widen instead of staying parallel. The middle column is the least useful number on this page; the interval is the claim.

yearcentral fit80% interval
202625.5%24.9% to 26.1%near horizon
202725.2%24.3% to 26.2%near horizon
202825.0%23.8% to 26.2%near horizon
202924.7%23.4% to 26.1%the band is already wide here
203024.5%22.9% to 26.1%the band is already wide here
203124.2%22.5% to 26.1%the band is already wide here
203224.0%22.1% to 26.0%far horizon — read the interval, not the middle
203323.7%21.7% to 26.0%far horizon — read the interval, not the middle

Projected to 2033, the horizon the whole engine uses.

questions this page answers

What share of the world works in farming?

25.8% as of 2025, the latest measured value in the series (OWID / ILO). The fitted trend has it falling 1.0% a year, with an 80% interval of −1.8% to −0.3%.

How fast is the share of the world working in farming falling?

−1.0% a year over the fitted window, an 80% interval of −1.8% to −0.3% — a halving every 67.6 years (80%: 39.1 to 250.9 years). Fitted from 11 observations spanning 10 years.

What will the share of the world working in farming be in 2030?

The central fit says 24.5%, inside an 80% interval of 22.9% to 26.1%. The interval is the forecast; the middle number is only its midpoint. Bands widen with horizon because shocks accumulate — a constant-width band would be overconfident.

Is the decline in the share of the world working in farming accelerating or slowing?

Splitting the record in half, the fitted rate went from −1% to −2% a year — steady. Regime breaks, not window choice, are what dominate this method's errors — which is why every projection here carries an interval.

Where does this data come from?

OWID / ILO. 35 observations spanning 1991–2025. Curated benchmark history, extended by a weekly authoritative fetch and by news figures fact-checked against their source before they may touch a fit. Both the observation ledger and the fitting code are public, and the engine publishes its own calibration score and its misses.

the raw numbers

Every observation behind the fit, unrounded by us and unsmoothed. This table is here on purpose: graphs make people underestimate exponential change, and the raw series beside the curve is the one correction shown to work.

year% of employmentchange
199142.8%
199242.6%−0.5%
199341.9%−1.6%
199441.3%−1.3%
199540.6%−1.9%
199639.9%−1.7%
199739.4%−1.1%
199839.5%+0.1%
199939.5%+0.1%
200039.4%−0.4%
200139.0%−0.9%
200238.8%−0.6%
200338.3%−1.3%
200437.2%−2.8%
200536.3%−2.4%
200635.2%−2.9%
200734.3%−2.6%
200833.7%−1.7%
200933.2%−1.5%
201032.6%−2.0%
201131.7%−2.7%
201230.9%−2.6%
201330.1%−2.6%
201429.2%−2.9%
201528.5%−2.3%
201628.0%−1.9%
201727.4%−2.0%
201826.8%−2.3%
201926.3%−2.0%
202027.2%+3.5%
202126.9%−1.2%
202226.9%+0.2%
202326.6%−1.3%
202426.1%−1.8%
202525.8%−1.2%

where this comes from

Other SUSTENANCE curves:

Recent briefs from SUSTENANCE, the daily record of what actually moved:

every SUSTENANCE brief →

Every tracked curve: utility solar, installed · battery pack price · battery cell price · solar module price · disk storage · human genome sequencing · transistors per chip · fastest supercomputer · DRAM memory · solid-state storage · onshore wind electricity · solar electricity (LCOE) · space launch, to LEO · offshore wind electricity · AI task horizon (METR) · frontier AI training compute · modular housing, turnkey · frontier-grade intelligence · AI video, finished minute · humanoid robot, capable unit · cheapest humanoid, list price · healthcare spend per person

← the whole engine: build clock, cascade, calibration · the archive · how we know