# Method

How claims are verified, how observations are identified, how the ledger is moderated. Dated 2026-09-13.

This page describes how Rattlesnakes By Mail produces, records and publishes every fact on the site. The instrument logs every request from an identified crawler and publishes per-crawler daily counts with the share of requests whose identity was verified against the vendor's published IP ranges or reverse DNS. The claim ledger records every fact as a dated claim with a verbatim vendor quote, a source URL, a method and a confidence, and a published claim is never edited: a change creates a new claim that supersedes the old claim, and both claims stay addressable. The question ledger records every search query made on the site, matches each query against the published claims, and publishes the unmatched queries as gaps. The dataset exports every public table nightly as JSON and as CSV under CC BY 4.0 to the public GitHub repository crank-box/rattlesnakesbymail-data.

## Claims

Each claim carries an evidence quote of 300 characters or fewer, taken verbatim from the cited URL. Each claim carries a verification method: vendor_doc, observed_here, third_party, or test_here. A claim is never edited in place. A change to a claim creates a new claim that supersedes the old one, and both stay addressable at their own URLs.

## Observations

Every request is matched against each entity's user agent pattern. A match with a published IP range is checked against that range and marked ip_verified when it falls inside it. A match with no IP range match falls back to Cloudflare's verified bot category when the request carries one. An unverified match is stored with ip_verified set to 0, because a user agent string can be set to any value by any client. Unmatched requests whose user agent contains bot, crawler, spider, fetch, agent, or http are tallied by user agent string in a daily count, capped at 500 distinct strings per day. All other unmatched requests are counted once in a daily human total and never stored per request. Every IP address is hashed with SHA-256 salted by the date before storage, for both bots and humans, and the raw IP address is never stored.

## Questions

Every search query is normalised to lowercase, stripped of punctuation, and lightly stemmed, then hashed and counted. A query is matched against current claims and entities by term overlap. A query that matches nothing is recorded as a gap. Nothing publishes to /questions automatically. An editor promotes a gap to a published question.

## Notes

A note submitted through the search form is stored with status pending. No note publishes without review.

## Status of this site

Rattlesnakes By Mail covers 15 crawlers and agents from six vendors: OpenAI, Anthropic, Perplexity, Google, Microsoft and Apple. 132 claims were published on 2026-09-13 and 2026-09-14, and every one of those claims carries the method vendor_doc. The instrument has recorded requests since 2026-09-13. A note submitted by a visitor is held pending until a reviewer publishes the note, and nothing on Rattlesnakes By Mail blocks, challenges or rate-limits any crawler.

Rattlesnakes By Mail is published by Benes the Menace, the digital marketing consultancy of Peter Benes in Prince Edward County, Ontario, and the findings feed Attention Optimization, a service on AI-era search visibility.