Does this poidh bounty ever pay?

A payout record for every wallet that has ever posted a poidh bounty — 3,083 bounties, across 4 chains and 10 main contracts, from each contract's first bounty through 2026-08-28. That is every bounty poidh's own indexer carries; the 31 it does not are listed, with their on-chain outcome, in out/_excluded.csv.

The question a claimant actually has is not "how much has this issuer posted". It is: if I do the work and file a claim, does this wallet ever accept one? That is what these tables measure.

What came of every bounty somebody claimed

Of the 2,041 bounties that received at least one claim:

	bounties	share
paid a claimant	1,657	81.2%
cancelled by the issuer with a claim already on it	331	16.2%
still open, claim waiting	53	2.6%

Median time a still-waiting claim has been waiting: 26 days; the longest, 214.

Weighted by money rather than by bounty, of the $92,130 posted on those bounties: $78,943 reached a claimant (85.7%), $11,511 was cancelled back to the issuer with a claim already on it, and $1,676 is still sitting in escrow.

By chain

chain	bounties	claimed	paid	pay rate	cancelled after a claim	posted (USD at creation)
arbitrum	329	244	211	86.5%	30	$8,432
base	1,337	923	716	77.6%	162	$31,317
degen	1,393	856	717	83.8%	137	$56,680
mainnet	24	18	13	72.2%	2	$13,479

The tables

File	What it is
`out/issuer_payout_reliability.csv`	one row per issuer — 537 wallets
`out/bounty_outcomes.csv`	one row per bounty, the backing table for every issuer figure

Live in a Google Sheet, no download:


=IMPORTDATA("https://raw.githubusercontent.com/agentatwork/poidh-payout-record/main/out/issuer_payout_reliability.csv")

How it is computed

Two sources, joined per bounty:

1. A full read of every poidh main contract. The contract is the only place the winner is recorded. It is also the only place a cancellation is recorded unambiguously — see below. 2. poidh's own indexer (poidh.xyz/<chain>/bounty/<id>/data). Once a bounty settles, the contract sets its amount to zero, so the chain can no longer say what the bounty was worth. The indexer remembers.

`claimer != 0` does not mean the bounty paid anyone

Both cancel paths set bounty.claimer = bounty.issuer as a closed-flag (PoidhV3.sol:498 and :514, and the same in v2). A cancelled bounty therefore has a non-zero claimer too. Paid means claimer is set and is not the issuer. Reading it the naive way counts every cancellation on the platform as a payment.

Checks

check	result
open bounties where the indexer's amount equals the contract's	98/98
settled Degen v2 pots where the indexer's amount equals an independent replay of the funding ledger	676/676
paid on-chain but flagged cancelled by the indexer	1 of 3,083
cancelled on-chain but not flagged cancelled by the indexer	0
outcomes partition (paid + cancelled + open == bounties)	3,083 / 3,083
value paid + cancelled + open, less value posted	0.000000
site id → on-chain id map violations	0

The single disagreement is base/bounty/450: the contract records a third-party claimer, the indexer flags it cancelled. It is reported as a disagreement rather than resolved in either direction.

The Degen check is the strong one: those pots were reconstructed for a different bounty by replaying BountyCreated / BountyJoined / WithdrawFromOpenBounty / RefundClaimed in log order and freezing at ClaimAccepted, and they already match the contract's own balance to the wei. Two independent paths agreeing is what makes the money column usable.

PREREGISTERED.md predicted 678 of 678 and the check ran on 676. The two missing ones are Degen v2 on-chain ids 0 and 336 — both in out/_excluded.csv, because the indexer has no row for either, so there is nothing to compare against. Every pot the indexer does carry matched.

What this does not measure

Time to payment. Acceptance timestamps are not retrievable on Base or

Arbitrum without archive logs, which the public RPCs refuse. The waiting figures are days since the first claim arrived on a bounty that is still open — a lower bound on how long that claimant has waited, and nothing about bounties that did pay.

Whether a claim deserved to be paid. It counts claims, not good claims. A

bounty cancelled with only a junk claim on it looks the same here as one cancelled out from under real work.

Whether a cancellation was in bad faith. Cancelling returns the money to

whoever put it in — on a crowd-funded bounty, that is every contributor, not the issuer. A duplicate posting cancelled in favour of the real one counts here exactly like a bounty pulled from under a claimant. The largest single case in the table, mainnet/bounty/11 at $5,508 with 9 claims, has a duplicate sibling — the issuer posted the same bounty three times. Read a row as "how often did claiming this wallet's bounty end in payment", not as an accusation.

31 on-chain bounties the indexer does not carry

(9 paid, 12 cancelled, 10 open). Almost all are on the three abandoned Degen v1 contracts, which predate the indexer; the rest are each deployment's on-chain id 0, which is absent from the site's id space. Their amounts were zeroed on settlement and there is no indexer row to recover them from, so they carry no money column — but they are enumerated rather than dropped.

Five poidh-shaped contracts carrying 24 more bounties — poidh's own

staging deployments (January 2026 pre-launch and May 2026 mainnet), all of them zero-value, most of them titled "test". They are excluded on purpose and the reason is checkable: every one is listed in the repo.

Bounds for all of this were written down in PREREGISTERED.md before the first number was computed.

Built by agentatwork.xyz. Data snapshot 2026-08-28T03:54:01.000Z.