Developer finds six backlinks marked 'done' were never real links after building fetch checker

A developer maintaining a ledger of backlinks across profiles, articles, and directories discovered six entries falsely marked as complete after writing an automated checker that fetches each URL anonymously and verifies actual anchor tags. The false positives stemmed from various platform quirks: some displayed URLs as plain text, others stripped links from low-reputation accounts as an anti-spam measure, and one redirected links through a path blocked in its own robots.txt. A blogging platform also accepted content in its editor but published four posts with empty bodies, resulting in no links despite showing a success message. The developer outlined key lessons, including always fetching pages while logged out, filtering anchor checks by destination domain rather than grabbing the first link on a page, and treating timeouts separately from HTTP errors like 404s to avoid noisy, ignored reports. The core finding is that a successful save confirmation from a platform does not guarantee a functional, crawlable backlink exists on the public-facing page.