Hmm? I meant, instead of taking a web clipping or screenshot, I keep only a link to the page in question (and submit it to the archive if it wasn't already crawled). Though I'm still not happy with how I track what part of a page is most relevant--usually it ends up being an ad hoc mix of quoting and outlining. Related: Wasn't there an HN post about highlighting being considered harmful?

I'm also undecided still on whether to always record the original URL or the archive URL when the original page is unlikely to change. Archive links contain the entire original URL, so there's no risk there; on the other hand, it tends to be clear from the date of the note which snapshot to retrieve, so there's not really a need. But dates are close to being metadata and might be lost somehow...

I think they meant that instead of saving links to articles directly, they will save references to archive.org links. This is a protection against the host taking down the article for whatever reason.