Despite the promise of a digital object’s lasting lifetime, in practice digital objects degrade faster than books or other physical information containers.
This can occur when:
- Storage media bit rots
- The equipment to read the storage media becomes obsolete, breaks down, and is no longer manufactured
- The structure keeping the data (e.g., filesystems, databases) becomes obsolete
- The file format or character encoding becomes obsolete
- The software required to access the data is no longer available
This degradation can happen on the order of a few years. The Iomega Jaz drive was only sold for 7 years.
Compare with books, important documents, and artworks, which often last for decades without any particular preservation efforts (e.g., stored in a box in the attic). Some works have survived millennia. (Will that Photoshop document be openable in two thousand years?)
- Edward M. Corrado and Heather Moulaison Sandy. Digital Preservation for Libraries, Archives, & Museums. 2017.
Despite the promise that a digital object can be perfectly copied, in practice digital copies experience generation loss.
Platforms lossily recompress images and video on upload. Metadata is stripped or altered. Content is trimmed and cropped, audio added, replaced, or muted, overlays, watermarks, intros, and outtros are added.
People often shift text to images or video (think a screenshot of the Notes app), I think for several reasons:
- To cross-post to platforms that only support those formats
- To work around platform limitations (character count limits)
- To preserve formatting like rich text, color, fonts, or mixed or complex media
- To preserve metadata, structure, or context (keep multiple posts with their associated usernames together, to preserve the structure a conversation)
- For attribution or provenance (a screenshot of a post says “I am quoting this” and clearly separates it from the reposter, a screenshot may be seen as a way to preserve content in case the original is deleted11 I’m not saying this is good provenance, but I’ve definitely seen people use screenshots of text for this purpose.)
- To obscure the content from automated moderation (a screenshot won’t trip up word filters22 This might be less effective as more extensive image analysis happens, but for now this remains effective in a lot of places.)
- To use image management tools (images are often deeply integrated in a way text isn’t, photo management has tags, albums, ways to quickly visually scan)
These shifted objects become the objects that are further copied, so even text can suffer generation loss.