The archive visits a page when it visits. For a page that matters from today on, that's not enough: a change next month is only caught if the archive happens to pass by. Watch the page, and Yore checks it itself once a day.
Watch a page
Open the page from its plan and choose the Going forward tab, then Watch this page. Every watched page is checked once a day, at 04:30 UTC. Check now runs a check at once, at most once a minute per page. Stop watching ends it.
A check is one request from Yore to the site, under Yore's own name. Yore doesn't pretend to be a browser, doesn't sign in, and follows ten redirects at most.
What a check finds
- Captured. The first time the page answers. Yore keeps the bytes under their SHA-256, as a capture of its own with the time of the check, and reads their text.
- Unchanged. The page's content reads as it did at the last check. The check still counts as a capture: it proves the text stood on that day.
- Changed. The content reads differently. The capture becomes a version in the page's timeline, to compare with any other.
- Removed. The site answered 404, 410 or 451. That opens a removal. Three such checks, each at least 24 hours after the one before, confirm it. If the page answers again, the removal is resolved.
- Blocked. The site refused the request, with a 403, a 429 or a server error, or answered with something that doesn't read as the page: a redirect to the front page, or less than a fifth of the text the page last had.
- No answer. The site didn't respond.
Every check is listed under Going forward with its time, the status, and the hash of what was captured. The list only grows: a check is never rewritten.
Blocked says nothing about the page
A refusal isn't the page. Yore records it and never hashes it, because a block page compared with a block page would read as the page unchanged, and that's a claim Yore has no ground for.
When a site blocks a daily check, Yore reads the archive's index for that page again, so whatever the archive captured since is known and a removal the archive saw still shows.
Yore doesn't get around bot protection. A page behind it is not captured going forward: it shows as blocked, with the archive's captures beside it.
A second witness
A capture of Yore's own is Yore's word. So when a check captures a page for the first time, finds it changed or finds it removed, Yore also asks the Internet Archive to capture the page, through Save Page Now. The check notes that it asked. The archive's capture then appears among the page's captures the next time the index is read, under the archive's own timestamp.
Yore asks only where it holds an Internet Archive account's keys, and the archive may refuse or be blocked by the site itself. A witness that didn't come never fails the check.