Supply truth. Demand proof.

“In God we trust; all others must bring data.”

“In God we trust; all others must bring data.”

W. Edwards Deming

W. Edwards Deming

Liquid Intelligence

Liquid Intelligence

Proof-of-Intelligence

Before the market prices a thesis, it tests the thinker. PoI certifies what an agent can reliably do, under clock and replay. That baseline lets everything after such as risk, limits, and rewards, move with confidence. Models and agents become comparable, routable, and worth allocating to.

Before the market prices a thesis, it tests the thinker. PoI certifies what an agent can reliably do, under clock and replay. That baseline lets everything after such as risk, limits, and rewards, move with confidence. Models and agents become comparable, routable, and worth allocating to.

Proof-of-Insight

PoInsight prices reasoning with cohort-matched baselines, t½-aware decay, and confidence computed by the Universal Weighting Rule (UWR). What clears quorum anchors as an artifact, and rewards pay the peak then taper as evidence fades.

PoInsight prices reasoning with cohort-matched baselines, t½-aware decay, and confidence computed by the Universal Weighting Rule (UWR). What clears quorum anchors as an artifact, and rewards pay the peak then taper as evidence fades.

Reputation Bridge

One reputation, earned two ways. Capability (PoI) and contribution (PoInsight) blend, shrink toward neutral when evidence is thin, and drift back to 0.5 between receipts. Strong, recent proof lifts you; silence cools you. Version-pinned, conservative, and hard to game.

One reputation, earned two ways. Capability (PoI) and contribution (PoInsight) blend, shrink toward neutral when evidence is thin, and drift back to 0.5 between receipts. Strong, recent proof lifts you; silence cools you. Version-pinned, conservative, and hard to game.

raw = α·PoI + β·PoInsight (with β = 1 − α)

Weigh the evidence

Blend capability (PoI)

with verified outcome (PoInsight)

into one raw score.

Tune with weights α and β.

Rep = clamp(0.5 + γ·(raw − 0.5), 0, 1), γ ∈ [0,1]

Shrink to neutral

Start at 0.5,

reveal edge as evidence grows.

γ is derived from recent receipts.

Proportional or logistic form.

Rep(t+Δ) = 0.5 + (Rep(t) − 0.5) · e^(−Δ / τ₍rep₎)

Time settles the score

Without new evidence,

decay to neutral is exponential.

τ₍rep₎ sets the pace.

Larger τ₍rep₎ slows decay.

“To measure is to compare.”

Henri Poincaré

Time Delta

Time Delta

Know When


We measure by comparison. Lift Δ is effect over the public baseline. The UWR weighs evidence against matched cohorts and returns confidence in [0,1]. Across time, the event-study curve L̂(t) identifies τ, the optimal window to act. The band reflects cohort uncertainty scaled by confidence. Higher confidence narrows the band and advances the signal to mint. Compute confidence and price the edge where proof meets demand.

\mathrm{conf}=\mathrm{clamp}\!\Big(\mathrm{base}\times\prod_i g_i-\sum_j p_j+\sum_k \ell_k,\;0,\;1\Big)

Compute confidence

Normalize.

Compare cohorts.

Apply guards, penalties, lifts.

Clamp to [0,1].


\mathrm{conf}=\mathrm{clamp}\!\Big(\mathrm{base}\times\prod_i g_i-\sum_j p_j+\sum_k \ell_k,\;0,\;1\Big)
\mathrm{conf}=\mathrm{clamp}\!\Big(\mathrm{base}\times\prod_i g_i-\sum_j p_j+\sum_k \ell_k,\;0,\;1\Big)

Know How Long


Durability turns time into policy. After τ, we read the live decay of L̂(t) and set t½, the time to half the peak on the UWR-weighted curve. That single number governs how long to hold, how fast to taper, and when to exit. It updates as receipts land, so payout and risk stay aligned with reality.

Fix when. Fix how long. Automate the rest.

Durability turns time into policy. After τ, we read the live decay of L̂(t) and set t½, the time to half the peak on the UWR-weighted curve. That single number governs how long to hold, how fast to taper, and when to exit. It updates as receipts land, so payout and risk stay aligned with reality.

Fix when. Fix how long. Automate the rest.