SuperLedger · Lumero DAG Explorer

Every ledger answers “is this safe yet?” with a folk rule — six confirmations, two epochs. Nobody can say what that’s worth in money, and the same rule covers a $4 coffee and a $40M settlement. This replaces the rule with a price.

Simulated throughout — not live network data. Model parameters mirror phase0/simulate.py.

Lumero transaction explorer

Search a transaction to inspect its evidence and its assurance certificate.

Account lattice · simulated evidence

account chain quorum reached conflicting spend

Inspected transaction

Bonded capital

—

four underwriters

At-risk exposure

—

Utilization

—

of bonded capital

Marginal premium

—

per coverage window

Indemnity paid

no claims

Should the merchant hand over the goods?

Underwriters commit capital against this payment being reversed. Coverage climbs. The merchant’s rule is a number, not a wait.

WAIT

coverage below threshold

Order value

—

goods at risk

Coverage now

—

Coverage vs the accept threshold

coverage committed threshold = order × margin

This is the whole idea. Accept when coverage ≥ order value × safety margin. A $40 stand clears the moment coverage passes $40; a $2M house waits until coverage passes $2M — and now you know why it waited, in dollars rather than block counts.

Where does the coverage come from?

Each underwriter quotes off its own utilization. A request fills from the cheapest ask upward — an auction, not a queue. That’s what stops any one of them setting the price.

Underwriter	Bond	Its quote	Committed here	At-risk / bond

The hard cap. No underwriter may hold more at-risk exposure than its own bond. That one admission check is what makes a correlated fault survivable — panel 5. It also means coverage can run out: the book fills the coffee every time and may leave the whale wire short.

Concentration cap 45% of a policy here. The design says 20%, but with four underwriters that is unfillable — a deliberate demo divergence.

Payment assurance queue

Every policy in flight. Click one to inspect it above.

Id	Order	Covered	Status	Premium

Why does it cost what it costs?

Almost none of the premium pays for risk. It’s rent on capital locked for the length of the coverage window.

Capital rent

—

yield forgone while locked

Expected loss

—

hazard × at-risk share

Premium

—

per window

To cover $40M

—

one window

Rent ÷ loss

—

which term dominates

Coverage window 10 min

The window is the price. Doubling it doubles the cost and buys nothing — a fault can only reach back minutes. Phase 0’s original 72-hour default was wrong by two orders of magnitude; drag to 72h and watch the $40M figure.

What happens when a payment is reversed?

Insurance normally dies on claims adjudication. Here it can’t — the loss event proves itself.

Conflict proof

1Both transactions carry valid signatures

2Both consume the same output

3The certificate covered tx A, unexpired

4tx B reached quorum

≈200 bytes. Anyone can submit it. No adjudicator, no dispute window, no oracle.

Settlement

Seized from	Amount
No claim yet.

The part nobody else has. Preconfirmations promise inclusion and punish by burning the bond. Smart-contract cover needs a human to assess the claim. A coin spent twice is self-proving, so payout is automatic and goes to whoever actually lost money.

Does the book survive everything failing at once?

A BFT safety fault doesn’t reverse one payment — it reverses every payment in the window simultaneously. The worst possible shape for an insurance book, and the question Phase 0 existed to answer.

Enforce the at-risk invariant — no underwriter may commit more than its own bond.
Uncheck it, let the book load, then inject a fault. That contrast is the entire result.

Claims due

—

Paid

—

Shortfall

—

Underwriters solvent

—

Verdict

—

Owed vs bond, per underwriter

bond available owed when everything claims at once

The result. With the invariant on, every underwriter owes at most what it holds — everyone pays in full, shortfall zero, at any leverage. With it off, one event wipes the book out. The fix is a write-time admission check, not a consensus change.

Utilization and premium over time

As the book fills, the marginal premium climbs — steeply past the 65% kink.

utilization marginal premium (scaled)

Simulation controls

Merchant safety margin 110% Background demand 1.0×

Deterministic: the same seed replays identically. Seed — — included in the shareable link.