The Illusion of the Immutable Snapshot: Why Saving Everything Isn't Preservation
Common advice in digital preservation and web archiving often boils down to a single, seemingly noble directive: save everything. Capture more, crawl deeper, hoard the bits. The fear of loss—the dreaded link rot, the disappearing forum post, the vanished government page—drives a preservation ethos modeled on the librarian's ideal of a complete collection. But what if this drive to capture and freeze is, in a profound way, betraying the very nature of the thing we seek to preserve?
The web is not a library. It is a live performance. A webpage is not a static document but a temporary constellation of code, style, assets, and backend calls, rendered uniquely in a specific browser at a specific moment. The ‘Save As’ button, the web crawler, the WARC file—they all create a facsimile, a taxidermied version of something that was alive. We console ourselves that we have preserved the ‘content,’ but in doing so, we often sterilize the context. We trade the dynamic, often messy, experience for a frozen, brittle snapshot we can control.
The Performance That Cannot Be Archived
Think of a mid-2000s Flash-based artist’s portfolio. An archivist can save the .swf file, but without the specific version of the Flash plugin, the right screen resolution, and even the sluggish download speed it was designed for, the experience is lost. The ‘art’ was in that specific performance on a machine of its era. Our snapshot gives us a binary we can no longer execute, a script for a play whose theater has been demolished. In aiming to save the object, we failed to preserve the event.
This illusion extends to data. We champion open data dumps—massive, immutable CSV files released quarterly. We call this preservation and transparency. But data is also a performance. It is queried, filtered, visualized, and debated in real-time. A static snapshot of a dataset freezes it in time, but it also severs the living connection to the database that generated it, to the API that allowed for dynamic inquiry, to the community that gathered around its updating stream. We preserve the fossil but extinguish the ecosystem.
The counterintuitive argument, then, is this: sometimes, the most faithful act of preservation is not the exhaustive capture, but the deliberate, documented loss. It is the acknowledgment that some phenomena are inherently ephemeral. Our resources might be better spent not on trying to stuff every digital moment into a WARC file, but on developing richer forms of documentation—not just of the ‘what,’ but of the ‘how’ and the ‘why’ and the ‘feel.’ Capturing the user’s testimony, recording the screen interaction, detailing the technical dependencies. This is more anthropology than librarianship.
Preservation should not only be about creating perfect, immutable copies. It should also be about building better obituaries. Instead of pretending we can stop time, we could focus on teaching time’s passage. We can preserve the tools, the emulators, and the stories that allow future generations to understand what it *was like*, not just present them with a broken link or a silent, un-renderable file we proudly label ‘preserved.’ The goal shifts from preventing death to conducting a meaningful autopsy. In accepting that we cannot save the performance, we are freed to truly honor it.
Notes & further reading
A few pages I came back to while writing this:
- one area's overview
- The Data Cathedral: How John Graunt Counted London's Dead
- a local resource
- The Archive of One Weary Email: On the Preservation of Personal Exhaustion
- a place-by-place guide
- The Last Echo of the Server Fan: On Hardware as Witness
- a regional guide
- a nearby resource
- a helpful reference
- a practical rundown
- a useful directory
- a local resource
- a useful directory