By Yann LeCun's own account, DjVu was built to distribute high-resolution scans cheaply over the early internet, demoed by freely releasing the NIPS proceedings at nips.djvu.org
In his own retelling, Yann LeCun describes the purpose of DjVu, a project he started at AT&T Labs in the mid-1990s, directly: "The purpose was to devise a new image compression format so that printed documents could be scanned at high resolution and distributed efficiently over the newly expanding Internet." He frames the whole account as "a story about free books." As a demonstration, his team scanned the NIPS conference proceedings — by his own account "the 13 volumes" — OCRed and indexed them, and put them on a free website in 2000, nips.djvu.org, after Morgan Kaufmann and MIT Press (the publishers of record) granted permission because they were not making revenue on past proceedings. A companion reply in the same thread credits a further, separate permission negotiation with Springer for NIPS Volume 0 (1987), originally published by the nonprofit American Institute of Physics before its rights passed to Springer.
This claim was flagged [unverified-quote] from 2026-07-09 to 2026-08-13:
the only source was this same X/Twitter post, which had returned HTTP 402 at
the original capture time and reached the vault only as a search-engine
summary, with no verbatim phrasing preserved. A 2026-08-13 verification
session re-fetched the post directly and preserved the quote above with
source_sha — the flag is now resolved. The same session corrected the
source_date: the post is timestamped 2 January 2024, not 2023-01-03 as
originally recorded (a full year off).
One discrepancy remains open, deliberately unresolved rather than smoothed over: LeCun states the demo covered "13 volumes," but a 2024 Wayback Machine crawl of the site's own table-of-contents page lists 14 numbered volumes, 0 through 13, spanning NIPS 1987 through NIPS 2000. No source states the count explicitly at both the 2000 launch and the 2024 crawl, so it is plausible the original release covered volumes 0–12 (1987–1999) and Volume 13 (NIPS 2000, published by MIT Press in 2001) was added afterward — but this is inference, not confirmed fact.
The freed NIPS archive also links this note to the vault's NETtalk cluster: the earliest volume, "NIPS 0 (1987)," contains a paper by Terry Sejnowski, lead author of the NETtalk paper (claim-nettalk-1987-complex-systems-distinct-from-1986-report). NETtalk itself appeared in Complex Systems, not NIPS, so it is not in the freed corpus — the connection runs through the author, not the document.
As of 2026-08-13, nips.djvu.org itself no longer resolves — see
claim-nips-djvu-org-domain-no-longer-resolves-2026 — though the archive's
mission persisted structurally: see
claim-neurips-absorbed-free-nips-proceedings-into-papers-nips-cc.
Source
“Let me tell a story about free books. In the mid 1990s, I started a project called DjVu at AT&T Labs. The purpose was to devise a new image compression format so that printed documents could be scanned at high resolution and distributed efficiently over the newly expanding Internet. [...] As a useful demonstration of the technology, I decided to scan and distribute the complete collection of proceedings of the Neural Information Processing conference (NIPS). I asked the publishers, Morgan Kaufman and MIT Press, for permission to do that. They agreed because they weren't making any revenue from past proceedings. We scanned the 13 volumes, OCRed and indexed all of the material, and put it up on a free website in 2000: nips.djvu.org This open-access repository turned out to be *extremely* useful to the machine learning research community. [...] Eventually, the NIPS conference stopped making printed proceedings and started hosting all the books on their website ( nips.cc ), including our scans.”