How much accuracy can a market buy from contributors who value their privacy?

What it found

A market pays each contributor the log-score gain of a public Gaussian belief for releasing a statistic plus Gaussian-mechanism noise, so the expected payment is exactly the information released, ½λln(1+τ/a). The best noise is a closed-form quadratic root, and participation is possible if and only if κa < 1 with κ = cΔ²/λ: precision is capped at λ/(cΔ²) however many contributors arrive, approached geometrically at rate 2σ²/(2σ²+κ), and total payout is at most ½λln(1/(κa₀)). Early arrivals add too little noise (1.5–5×; 1–22% welfare loss against a planner), and admitting small data holders first lifts final precision from 28.0 to 34.2. Checked against grid search and 10⁶-run Monte Carlo. Stylised.

A figure from its own output

Paper, code and results

• README and method
• White paper
• Experiments and results
• Code
• Tests

More in Paying contributors

• Paying for improvements, not for overfittingPaying for model improvements against a hidden holdout: telescoping budget, a payment-feedback overfitting attack (paid 0.15 while true loss worsens 0.37), and the Ladder fix with its measured cost.
• What does submitting copies earn under Shapley payments?Shapley and leave-one-out payments under replication: Shapley pays every extra copy even when worthless, saturating at stand-alone value ((n+2)/2× honest), naive evaluators hand over the pot, LOO pays copies 0 but distributes 1/(n+1), dedup underpays lookalikes by 45%.
• How should Shapley payments be sampled and audited?Auditing sampled Shapley payments: exact estimator variances in a saturating game, permutation sampling is exactly efficient but stratified sampling has 6–8× lower variance (zero for equal contributors) at up to 7% budget error, antithetic reversal gives 1.6× lower RMSE, Hoeffding sample counts are 4–5× conservative.
• How precise should a training-contest leaderboard be?Winner-take-all training contests scored on a noisy holdout: effort V a_n √m/s with a_2 = a_3 = 1/(2√π) exactly, a precision ceiling V/σ² ≤ λ_max(n) beyond which no pure equilibrium exists (below the zero-rent limit), and the resulting optimal holdout size.
• What does it cost to buy decentralised compute?Buying k units of decentralised compute by reverse auction: exact VCG payment k(k+1)/(n+1) is exactly 2× winners' cost, optimal reserve is exactly v/2, first-price equilibrium matches by revenue equivalence.
• What does it cost to make workers race?Paying the first k of n finishers makes speed a contest: π'(1)=(1−k/n)(H_n−H_{n−k}), equilibrium speed R(n−k)(H_n−H_{n−k})/(nc), exact round time nc/((n−k)R); paying all finishers buys no speed, a budget favours winner-take-all until an accuracy floor k_min binds, and recruiting more workers barely speeds the round.

All 8 Paying contributors projects →

Other research topics

Proper scoring & elicitation28Effort & peer prediction9Verification games & slashing19Detection & tolerances14Markets as routers13Prediction-market design9Paying contributors8Decentralised training systems23Digital twins & sim-to-real164LLM chip design loops40Robot manufacturing loops53Ambient AI assistant46Self-replicating manufacturing34Simulation-first design64Engineered clothing45Task-specific AI accelerators40Algorithmic chip-design loops42Open-source purpose-built robots44The software-engineering loop43Adversarial human imitation41Open-source residential construction41Autonomous multi-angle capture34Sensing, imaging and 3D31Energy supply and dense energy storage39Robot societies among people31Agentic co-design of low-cost robots27Laundry-hanging robots25Machine learning on qualitative data23Robots on the building site27Civilizations & group psychology1History & human sciences1

Produced with AI assistance (Claude). Results are from the code in this repository and are simulated or stylised unless the abstract says otherwise; nothing here implies affiliation with or endorsement by any lab, researcher or company cited. MIT licensed.