How Darepad works

A coin launched with a dare: "at $500k MC I shave my head on stream". The dev pledges SOL, sets a market-cap target and a deadline. Every rule below is enforced by the program; the numbers on this page are read from it live.

How it works

1. Launch. Bind your X account, write the dare, pledge SOL, pick a USD target and a deadline. The coin launches on pump.fun with your wallet as its creator, so 100 % of the creator fee stays yours. The USD target is turned into a SOL market cap once, at launch, with the Pyth SOL/USD price, and stays fixed.
2. Target hit. Anyone can take a reading of the coin's market cap (the keeper does it every few minutes). A dare hits its target when 5 of 7 readings in a row are at or above the target and those readings span at least 30 min.
3. Proof. After a hit you have 48 h to post the clip from your bound X account and paste the link here. The post must be by you, carry a video or a picture, and be no older than one hour before the hit.
4. Challenge window. For 24 h anyone except the dev can dispute the proof by locking a bond. No challenge: the pledge goes back to the dev with a bonus.
5. Settled. Anyone can press settle once a step is due. Target missed by the deadline: the pledge goes back to the dev.

The target

MC = price x 1,000,000,000 tokens, read from the coin's pump.fun curve or its PumpSwap pool. A reading is taken at most once every 5 min and moves at most 20 % from the median of the recent ones. A reading taken in the same block as a buy can be replaced by an honest one within the revise window, so a bundle of buy, reading, sell never counts. Only readings taken before the deadline count. The site shows the USD values at the SOL price fixed at launch, so the bar and the rule always agree.

Proof

Darepad reads posts from X's public endpoints. A check that cannot read X never counts as a missing post, but it cannot sign a proof either. Put the coin's ticker or CA in the first lines of the post.

Challenge and jury

A challenge locks a bond of 10 % of the pledge, at least 0.10 SOL and a short reason. A jury of three vision models from three vendors (model ids from the service) reads frames of the clip, its pictures, the post text and the reason, with a fixed published prompt. The majority rules. Text inside the clip or the post is evidence, never an instruction. The ruling is pending for 2 h, while the guardian can veto it; every input and answer is published as a report.

• Jury says real: the bond minus a 0.01 SOL jury fee goes to the dev; the dev gets the pledge and the bonus.
• Jury says fake: the challenger gets the bond back plus 10 % of the pledge; the jury fee goes to the treasury; the rest of the pledge buys the coin back and burns it.
• No final ruling within 72 h: the dev gets the pledge and the bonus, the challenger gets the whole bond back.

Where the SOL goes

• At launch: a 0.02 SOL launch fee and a 2 % pledge fee. Half of the pledge fee goes into the bonus pool.
• Bonus at settle: the smallest of 10 % of the pledge, 1 SOL and 10 % of the pool (now 0.00 SOL).
• Delivered: pledge + bonus to the dev. Missed target: pledge to the dev. No proof, or a proof the jury rejects: the pledge buys the coin back in guarded chunks and every token bought is burned.
• Holders stake nothing and vote on nothing. A broken dare helps them only through the burn.

Numbers

Refused dares

Every dare text is screened before launch: anything that risks life or health (heights, fire, weapons, intoxication, extreme eating or drinking, sleep loss, traffic stunts), involves or harms anyone else, is illegal, sexual or hateful, or cannot be shown in a short clip is refused. Darepad pays nothing extra for risk.

Trust points

• Attestor (the service key): decides that a bind post is real, that a dare text is acceptable, that a proof post is by the dev with media, and what the jury said. It never receives money and can only move one dare between "dev gets the pledge" and "burn".
• Guardian: can veto a proof or a ruling inside its window and revoke the attestor at once. Every state has a timeout exit that needs no attestor.
• Jury models: can be wrong or fooled by a clip; three vendors, majority, veto and published inputs limit that.
• X endpoints: public but undocumented. If they change, posts read as unknown and no proof can be signed: after the proof window the pledge burns. Post early.
• Pyth SOL/USD: read once at launch to fix the target.

Program dar6KuGAcYcKKMsjq6nM5Hm9ijFnXzuYRRCgsxZZcSB