Skip to content

Empirical Research: Bit Rot & Media Decay Metrics

Data loss occurs via two distinct pathways: catastrophic hardware failure and silent data corruption (bit rot). This reference compiles empirical research metrics from drive manufacturers and enterprise storage studies.

Silent Bit Rot Rates by Media Type

Silent bit rot occurs when physical storage media undergoes microscopic magnetic flip, flash cell discharge, or background read-disturb events without raising a hardware I/O error.

Hardware Comparison Summary

  • Consumer SATA HDD: 1 Unrecoverable Read Error (URE) per 12.5 TB read. Silent corruption occurs at ~1 corrupt file per 20 TB read.
  • Enterprise SAS HDD: 1 URE per 125 TB read. Silent corruption occurs at ~1 corrupt file per 200 TB read.
  • Consumer NVMe SSD: High failure rates under unpowered or high-temperature conditions.
  • M-DISC Optical Media: Less than 1 URE per 1,000 PB read. Negligible degradation over 50+ years.

The 12.5 TB RAID Vulnerability: Because consumer SATA drives specify a URE of $10^{14}$ bits (~12.5 TB), reading through 12.5 TB of data during a traditional RAID rebuild carries a high statistical probability of encountering an unrecoverable read error. This is why copy-on-write filesystems with scrubbing (ZFS / Btrfs) are essential.

File Format Vulnerability Under Corruption

Experiments evaluating how common file formats handle 1-byte corruptions:

  • JPEG: Entire image past the corrupted byte shifts, changes color channels, or breaks completely.
  • PNG: CRC chunk validation fails; decompressors halt rendering.
  • MP4: Container header corruption renders the entire file unplayable without specialized recovery tools.
  • FFV1 (Matroska): Errors are isolated to a single frame slice; the remainder of the video plays cleanly with zero loss.