Emergency help, paid peer to peer on Bitcoin

Bitaid lets a person in danger summon capable people nearby and pay them, with no dispatcher, no token and no company in the middle. Every payout the case can end in is signed before anyone has a reason to dispute, so whoever decides the outcome can pick one but never take the money. It now runs end to end on Bitcoin's public test network, over Tor, between a PC, a phone and a tablet.

November 2026 · @fiatstandard

Read

The problem

$5

Violence is decided in the minutes between an attack and the response to it. Self-defence is always there, but one person against several is outmatched; what turns the outcome is a second and third capable person arriving within minutes.

Today insurance pays after the harm is done. Rescue in the moment is almost nowhere for sale, and nothing lets you pay whoever capable can get there first. Private security bills hours, not results, so it guards places for those who can afford idle time. The police are a monopoly: no prices, no rival, no customer who can leave, so they cannot learn what protection is worth or to whom. Bitaid pays for the problem solved, not for the hours spent waiting for one.

The insurance lever

Give me a place to stand, and I will move the earth.Archimedes

A Bitaid offer is escrowed in advance and costs nothing until it is used: the money waits as unspent bitcoin, paying no wages for idle hours. And because emergencies rarely strike the same people at once, an insurer can stand one stake behind many offers.

A thousand people paying two dollars a month can each summon help with an offer of $8,000: leverage of some three hundred to one. That leverage rests on emergencies being rare and spread out, and on the insurer holding reserves of its own for the bad years; the premium prices both. Underwriting it lets bitcoin capital earn bitcoin from real risk, without lending it out. Bounties that size draw the most capable responders, and fund the teams, drones and medics that compete to arrive first.

1,000 people paying $2 a month
$8,000
the offer each of them can summon
Escrowed offers cost nothing until used, and emergencies rarely strike the same people at once: one pooled stake stands behind many offers.

Then the insurer's own interest takes over. With enough capital committed, it pays to keep its own guards and patrols near its clients: they deter the emergencies its offers would otherwise pay for, and when one happens they are first on the scene, earning back the bounty it escrowed. Fewer claims mean lower premiums, which bring more clients, whose premiums fund more patrols. That flywheel turns insurers into security providers that compete on price and on how fast they arrive, paid for results rather than idle hours.

Bitaid does not declare a right to security or ask everyone to agree on one; it lowers what it costs to keep reinforcements on call, yet raises the price paid for saving life and property, and lets the market answer.

How a case runs

  1. Before any emergency, the caller (or their insurer) escrows an offer in bitcoin. It sits in a script nobody can spend alone; if no emergency ever comes, it returns to its owner.
  2. The alert goes to bonded responders in the surrounding area. Each one who accepts posts a bond of their own.
  3. Within about a minute, the parties' devices sign every payout the case can end in: on the public test network, 39 possible payouts for three responders. Nothing new can be added later.
  4. Everyone records. Whoever claims pay must prove what they did on their own footage, so every scene is filmed from several angles. Each recording is hashed and anchored to Bitcoin.
  5. The people who were there vote, blind and item by item: who attended, who helped, what happened. If they agree, that settles it.
  6. If they disagree, pseudonymous arbitrators drawn at random answer narrow factual questions about the footage. Each staked bond is at risk if the answer is wrong. A party can buy a larger panel, and a sample of verdicts is reviewed again by three fresh panels.
  7. An attestation picks one of the pre-signed payouts, and it is broadcast. The money moves straight from the escrow to the people who earned it.

Why cheating does not pay

  • Those who decide never hold the money. They choose among payouts signed in advance, each paying only the parties. No outcome pays them. A wrong ruling can still pick the wrong payout, which is what the bonds, reviews and restitution are for.
  • Every role stakes a bond larger than what cheating could win. An arbitrator who votes against the final ruling forfeits part of it.
  • Faking the scene is the hard part. Footage hashed the moment it is taken cannot be edited afterwards, and a staged scene must hold up across several independent cameras and the people who were there.
  • Weight is earned, not bought. A responder's vote counts by the paid work done in that area; an arbitrator's seat by the cases already judged. Both season over calendar time no one can buy.
  • Any case can be reopened. Three fresh panels that overturn a verdict seize the whole case stake of those who voted for it, and repay the wronged.
  • Nothing to capture. No token, no company, no administrator, no governance vote. Ordering and finality come from Bitcoin.

What already runs

A working prototype in Rust runs every role: callers, responders, archivists, the threshold-signing oracle pool and arbitrators. Each check below is a script that passes or fails on a real chain, with the logs published.

Measured on the public signet, over Tor, phone to tablet (all figures):

  • 2 to 6 seconds from the alert to the first responder accepting, bond checked on chain; three within 44 seconds.
  • About a second for an accepted responder to hold the exact location, sealed to it alone.
  • Seconds to sign every payout the case can end in, across three machines; within the hour the incident is funded and co-signed by the pool, while help is already on its way.
  1. 61passed, 0 failed
    Nov 20, 2026, Bitcoin public signet, over Tor, PC + phone + tablet
    Every arbitrator staked before being paid; the exact location sealed to each accepted responder; an outsider's forged records ignored; fines paid to the arbitrator who dissented and was proved right
  2. 61passed, 0 failed
    Nov 19, 2026, Bitcoin public signet, over Tor, PC + phone + tablet
    A disputed case through two tiers with every deadline anchored on chain; a node that saw nothing rebuilt the same record from the chain and the archive alone
  3. 81passed, 1 failed
    Nov 9, 2026, Bitcoin public signet, over Tor, PC + phone + tablet
    Every deadline counted in blocks. A captured panel's verdict stood unappealed; three fresh review panels overturned it, and the two arbitrators who backed it lost their whole stakes to the wronged
  4. 15passed, 0 failed
    Nov 8, 2026, Liquid testnet
    The escrows: funding, a pool-co-signed incident, pre-signed payouts, a wrong attestation refused by the network, a bond forfeiture, the timelocked fallback
  5. 74passed, 0 failed
    Nov 7, 2026, Bitcoin public signet, over Tor, PC + phone + tablet
    The staged review: three fresh panels overturn a verdict; the wrong arbitrators lose their stakes, which repay the wronged responder and the dissenter who was right
  6. Undisturbed
    Nov 7, 2026, Same run, plus a hostile stranger
    Floods, garbage, oversized and replayed messages, fake block heights, connection storms aimed at the network while the case ran. The case finished, and every node refused the fake height
  7. 71passed, 0 failed
    Nov 6, 2026, Bitcoin public signet, over Tor
    A full disputed case: a split vote, two arbitration tiers, an appeal, missed deadlines replaced, penalties, all payouts on chain
  8. 84passed, 0 failed
    Oct 29, 2026, Private signet, over Tor
    The same case on a private test network
  9. Reproducible with one command
    Oct 2026, Economic simulations
    Eight experiments, thousands of cases each: honest work profits; lazy, bought and staged verdicts lose under the adopted rules

What is not done yet

  • Every node so far is ours. The cheaters in these tests cheat because a script tells them to. Independent operators, choosing their own strategies, are the next milestone.
  • One fee coin short. In the latest run a responder ran out of coins to pay a payout's fee; another party broadcast it and every pot settled. The fix waits for the next run.
  • No outside review yet of the cryptography (FROST, MuSig2, the scripts) or of the incentives.
  • Test coins only. The prototype must not hold real funds.

How to help

  • Run a node in the coming test pilot, and try to break it. Attacks that work are the most useful thing anyone can send.
  • Review the cryptography: the threshold oracle (FROST with ROAST), the pre-signed payout sets (MuSig2) and the Taproot scripts.
  • Attack the incentives: the simulations are open; change an assumption and see what breaks.
  • Pilot it: insurers, security firms and communities where help arrives late or not at all.

Issues and pull requests are welcome on either repository.

Read more

  • Whitepaper: the argument, in full
  • Specification: scripts, messages, parameters and the record of every test
  • Prototype: the Rust code, the test logs and the simulations