Information Loss

Information loss is not a single event. A thought can leave attention before it is ever captured. A captured file can be deleted, corrupted, or lost behind a broken link. A preserved file can become unreadable years after it was saved. Documenting well means accounting for each of these failure modes, not only the first.

Never captured

A thought, an observation, a result not written down at the time is subject to the forgetting curves described in the previous post. A page-worth of detail becomes a paraphrase within a day and a vague impression within a month.

Captured, then destroyed

Drives fail. Files are accidentally overwritten. Cloud services are discontinued. An archive without a tested backup is one hardware event from gone.

Captured, then unreachable

Web references decay quickly: a 20-year longitudinal study of library and information science journals found accessibility dropping from 87% for citations 0 to 5 years old to 38% for citations older than 10 years, with permanent link rot tripling from 5% in 2012 to 15% in 2025 [1]. Across the broader science, technology, and medicine literature, one in five articles already contains at least one broken web reference [2]. The same failure mode applies on a personal scale to a file saved under a forgotten filename, a project stored on a service that was deprecated, or a password that was never written down.

Captured, then degraded

Many conversions are lossy. A JPEG re-saved is no longer the original JPEG. A spreadsheet exported to CSV loses formulas, cell types, and styling. A handwritten page photographed at low resolution preserves the appearance but not the searchable text. Each lossy transformation moves the record further from its source, and the loss is permanent.

In practice

Preserving information is an ongoing set of decisions, not a single act of saving. Choice of format determines how degradation accumulates. Choice of location determines how easily a file can be re-found and backed up. Choice of medium determines how much outlives a failed device or a discontinued service. The next posts in this series treat these choices one at a time.

References

  1. [1]Sadatmoosavi, A.; Khasseh, A.; Tajedini, O. (2026). Link rot in LIS literature: a 20-year study of web citation decay, recovery and preservation challenges. Aslib Journal of Information Management. doi:10.1108/AJIM-05-2025-0286
  2. [2]Klein, M.; Van de Sompel, H.; Sanderson, R.; Shankar, H.; Balakireva, L.; Zhou, K.; Tobin, R. (2014). Scholarly Context Not Found: One in Five Articles Suffers from Reference Rot. PLOS ONE, 9(12), e115253. doi:10.1371/journal.pone.0115253