Analysis
Redaction Analysis Center
Comprehensive analysis of 519,436 DOJ Epstein documents. 34.1% of all redactions were improperly applied — the underlying text was not actually removed from the PDF.
1,808,857
Total Redactions
616,221
Recoverable
34.1% of total
218,960
Docs with Recoverable Text
42.2% of all docs
376,559
Docs with Any Redaction
34.1%Recoverable
616,221 bad redactions — text recoverable under the black boxes
1,192,636 proper redactions — text actually removed
“Bad” redactions use PDF overlays that visually hide text but don't remove it from the file. The underlying text can be recovered by analyzing the PDF text layer beneath the redaction overlay.
Redactions by Dataset
DS1
67,114 bad
DS9
13,320 bad
DS10
162,494 bad
DS11
369,915 bad
DS12
3,378 bad
Redaction analysis data from Epstein Research Data by Rye Howard-Stone. See also: Recovered Text · DOJ Document Audit