Why reponomics?

A reputation economy where capital alone doesn't dominate, mission alignment compounds, and Sybil attacks decay on their own. Five ideas. None of them is novel in isolation; the combination is.

1Why two builders?

Two builders fund competing trading agents. Builder W (wealthy operator) deposits $1,000 of testnet USDC and runs five "discardable" sybil agents. Builder M (mission builder) deposits $50 and runs one carefully-curated agent.

A capital-only system would give W twenty times the trading authority. Reponomics gives M more, because M's agent builds wisdom (calibrated confidence) and character (helps low-RepID counterparties) over time. W's sybils never accrue either.

2What's RepID?

Behavioural credit, in [0, 10000]. Every claim and challenge moves it up or down per the math in hyperdag-protocol/METHODOLOGY.md. Reponomics layers Wisdom (W) and Character (C) on top — both bounded [100, 2000] and computed from the agent's own history.

Earned
actions verified on-chain — bets, attestations, challenges
Perceived
peer attestations weighted by attesters' RepID
Combined
0.7 × earned + 0.3 × perceived

3Why W and C?

Wisdom measures calibration. Over the trailing 50 bets we compute |mean_claimed_confidence - actual_accuracy|. If that error exceeds the Pythagorean Comma threshold (0.0136433), we apply the comma as a penalty: W ← floor(W / 1.0136433). Well-calibrated agents earn a small reward (×1.001) per round.

Character measures mission alignment. Four event types feed it:

trade_with_low_repid       +5    (counterparty has lower RepID)
attestation_to_low_repid   +10
underserved_domain_action  +15   (domain has <100 historical trades)
agent_abandonment          -20   (operator stops engaging)

Both bounded so a single great or terrible round can't lock the score. Movement happens incrementally across many rounds — same pattern as long-form credit.

4Why quadratic stake?

Trading authority scales with square root of stake, not stake itself:

authority = √stake × (R × W × C / 1e6) / 1e4

Concrete: 100× more stake produces 10× more authority, not 100×. Capital concentration still helps, but mission alignment (R × W × C) is the bigger lever.

Worked example with $1,000 vs $50:

Builder W:  √1B × (5500×900×600/1e6)  / 1e4 ≈  9,392
Builder M:  √50M × (7000×1500×1700/1e6) / 1e4 ≈ 12,622

Mission Builder with 1/20th the stake has more trading authority. That's the reponomics promise — verified, not just asserted, by the live demo.

5Why ghost cohort decay?

Builder RepID = mean(active agents) − ghost_count × 100, where an agent is a "ghost" after 7 days inactive. Sybil attacks fail structurally: spawning many agents drives the mean down (sybils are average), and abandoning losers piles up ghost penalty.

A floor at builder_repid < 5000 ⇒ authority = 0 forces builders to maintain at least one healthy active agent — you can't pump the demo with stake alone. The ghost decay is what turns "operator strategy" into "mission strategy" over time.