What a digitized National Technical Reports Library must prove: a measurement on 50 public technical reports and a comment to NTIS docket NIST-2026-0166
About two in three of the National Technical Reports Library's nearly three million reports exist only on paper, microform, tape, CD or DVD, and at the National Archives' published scanning prices one pass costs $8 to $14 million at 100 pages per report. The choices made in that pass decide whether anyone pays twice. In a public comment on docket NIST-2026-0166, I report the file structure of 50 public federal technical reports, 4,310 pages in all, drawn at random from five years between 1965 and 2005. The largest 2% of pages hold 40% of the stored image bytes, and the largest page is 123 times the median. Of the image pages, 94.6% are one-bit, although the federal rule for new scans of permanent paper records lists gray or color at 8 or 16 bits. Seventeen of the 50 files mix image encoders. A federal-guideline gray master of the whole collection costs about $21,000 a year at a public cloud list price, and a second scan pass costs at least 375 times that. The evidence supports a pilot before production, with eight requirements and a 90-day plan. It does not show what a lossless master costs, which the pilot should measure. This text was filed as a public comment in response to the National Technical Information Service's Request for Information on the National Technical Reports Library, 91 FR 61835, 30 September 2026, FR Doc. 2026-20023, docket NIST-2026-0166, on 2026-10-05 (regulations.gov tracking number muw-4q33-m5or). The data, scripts and results are in Annex A, doi:10.5281/zenodo.23131223.
Authors
- Eugen Spivak (ORCID: https://orcid.org/0009-0005-7509-8609)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-10-05
- DOI
- https://doi.org/10.5281/zenodo.23199588
- Primary Topic
- Digital and Traditional Archives Management
- Type
- article
- Field-Weighted Citation Impact
- 0.00