Where every number on this site comes from

What gets collected, and how

Three passes run over the same list of boards, and they do not overwrite each other.

1. The seed. The census starts from outoutbid.lol’s public directory export, used under CC BY 4.0. That gives an address, a category, an entry price and, for most boards, a figure the board publishes about itself. Everything taken from it is labelled third-party sourced on the board’s page — never presented as this site’s own reading.
2. The sweep. Every address is then fetched from here directly, with a two-attempt retry and an http fallback before anything is written off. What comes back gives the status, the page title, the meta description, the payment provider loaded in the page source, and any money printed on the page. These are labelled first-party or automated observed.
3. The registry. Domains without a registration date are looked up in RDAP — the public record that replaced WHOIS — through each registry’s own service, using IANA’s bootstrap file. That took the coverage from 43% to 97% of the census. These are labelled registry verified.
4. The rendered read. A plain fetch sees nothing on a board that builds its page in the browser, and a third of this wave does. Every board that answers is then opened in a real browser, which reads the page as a visitor would, follows at most two of its own rules or pricing pages when no payment provider was visible, and photographs the boards with a complete file. Where the browser saw more than the fetch did, the board’s page says so and gives both character counts.

The seven source types

Every value on every board page carries one of these, with a timestamp and a confidence mark. Where two sources disagree about the same field, both stay on the page. The newer one does not silently replace the older one — that disagreement is usually the most interesting thing on the page.

What the money figures are, and are not

Every money figure on this site is what a board publishes about itself. It is not audited revenue, it has not been through anyone’s books, and a board that inflates its own counter is reported here at its inflated figure. That is why the field is called reported total everywhere and never verified revenue.

These four things are different, and this site keeps them apart:

• Publicly reported total — a number printed on the board’s own page.
• Observed board total — a floor summed from the rows a page happens to show.
• Owner-confirmed revenue — a figure whoever runs the board has stood behind, after a domain check.
• Payment-verified revenue — nobody has this, including this site. No board on file carries it.

Null is never zero. 136 boards publish no readable figure at all. They are excluded from totals and medians rather than counted as having taken nothing — treating them as zeros would understate the wave and libel the boards.

Totals only go up. Most of these boards let a bid stand forever, so a cumulative total says nothing about whether anyone is still bidding this week. A board reporting $6,000 may have taken all of it on day one and nothing since.

Currency

Boards price themselves in dollars, euros, rupees, reais, SOL, ETH and satoshis. Each figure is kept in the currency the board published it in — the raw string appears on every page. A rough fixed-rate conversion exists only so lists can be ordered and totals summed, and it is deliberately not shown as a converted figure anywhere:

1 USD = $1 · 1 USDC = $1 · 1 EUR = $1.08 · 1 GBP = $1.27 · 1 INR = $0.012 · 1 BRL = $0.19 · 1 SOL = $190 · 1 ETH = $3400 · 1 BTC = $62000 · 1 SATS = $0.00062

These rates are fixed, not live. Any board priced in crypto should be read as approximate on the sorting axis and exact only in its own currency.

How a board is called live or dead

Status is measured here rather than relayed. Each address is requested with a normal browser user-agent, following at most four redirects, with every hop re-checked, and a hard cap on how much of the response is read. A board is only written off after two failed attempts and an http fallback.

How rules get classified

A board’s own words — tagline, page title, meta description and the first several thousand characters of visible copy, taken from the rendered page where one exists — are matched against a published list of phrases per mechanic. A match records the phrase that triggered it, and that phrase is printed in the board’s evidence ledger, so a classification can be argued with rather than trusted.

Where nothing matches, a board is filed under single highest bid — the default shape of this wave — and the field is marked estimated, at low confidence. Around a third of the census sits in that bucket. That is a guess, and it is labelled as one on every page it appears.

The five editorial marks

Each is a formula over the data, scored 0–5. None of them is an opinion about whether a board is good.

• Originality — 5 × (1 − similarity to the nearest other board in the corpus), where similarity is the overlap of distinctive words across every board’s copy. A board whose page reads like a copy of another board scores low by construction.
• Transparency — one point each for publishing a total, a top bid, a bidder count, an entry price, and at least two readable rules.
• Activity — two points for answering, one for being in the top quartile of clicks out of this site, one for reporting more than $100, one for showing a leaderboard at all.
• Survival — days since registration, capped, and only counted while the board still answers.
• Data confidence — how many independent kinds of source stand behind the board’s fields.

They are deliberately not combined into one score. A single number would hide exactly the disagreement this site exists to show.

Why some pages are not indexed

A page with a name, a URL and nothing else helps nobody. Boards are sorted into three tiers, and only the first is offered to search engines:

• 120 full files — a registration date, a readable page, a rule, money or an entry price, and a complete evidence ledger. Indexed and in the sitemap.
• 324 thin files — listed, browsable, linked, and marked noindex until the record fills in.
• 6 held back — parked domains, boards that redirect somewhere else entirely, and pages that are not pay-to-rank at all.

The archive, and why it exists

Every sweep is stored as a dated snapshot and nothing is ever overwritten. That is the whole point: a board's own counter is a single number on a page, replaced in place each time it changes. What a board said it had taken last Tuesday is not recoverable from the board. It exists here, or it does not exist.

The change tape is the difference between the last two snapshots. The curves and the uptime strips on each board's page are every snapshot replayed. Identical consecutive readings are collapsed — a flat line of the same figure says nothing — but the first and last of any run are kept, so the dates stay honest. There are 3 sweeps on file so far, the first from 27 Aug 2026.

This is also why boards that die keep their pages. A dead board's last reading is the only record that it ever had rules, a price and a top bid.

Fetching addresses safely

Anyone can submit a URL, so nothing is fetched blindly. Only http and https are allowed; credentials in URLs are rejected; hostnames are resolved before the request and refused if they point at loopback, link-local, private or cloud-metadata addresses; every redirect hop is re-checked the same way; redirects are capped; and response bodies are truncated.

The rendered pass does execute a board’s JavaScript — that is the whole point of it — so it runs in a throwaway headless browser, off the machine that serves this site, with every address re-checked through the same guard before it is opened, dialogs auto-dismissed, and nothing kept but text and a picture.

Screenshots

Boards with a complete file are photographed at 1200×750 when they are rendered, and the picture is dated on the page. It is a record of what the board looked like on that day, kept for the same reason the rest of the archive is: most of these addresses will not be there in a year.

Getting it wrong

Some of this is certainly wrong. Rules are inferred from marketing copy, categories come from someone else, and money is self-reported by parties with an obvious interest. Corrections are free, they do not require an account, and if you run a board, claiming it is the fastest way to fix your own file.