---
title: "By Yann LeCun's account, DjVu was built to distribute high-resolution scans cheaply over the early internet, demoed by freely releasing the NIPS 1988–1999 proceedings at nips.djvu.org"
type: "claim"
status: "seedling"
audit_status: "flagged — rests on a personal X/Twitter post by Yann LeCun that could not be re-fetched verbatim (X returned HTTP 402); the account was relayed via a search-engine summary at capture time (2026-07-09), not read directly. [unverified-quote] No verbatim phrase was preserved. Verification routed at [[question-verify-lecun-djvu-free-nips-archive]]."
source_url: "https://x.com/ylecun/status/1742269871168111018"
source_author: "Yann LeCun (personal account, X/Twitter)"
source_date: "2023-01-03T00:00:00.000Z"
source_quote: "[no verbatim phrase preserved — X post returned HTTP 402 and was relayed via search-engine summary, not read directly]"
source_tier: 3
provenance: "Promotion from 10-inbox/raw/2026-07-09-hop-djvu-lecun-bengio-book-compression.md, 2026-07-09"
origin: "batch"
derived_from: ["20260709-1653-hop-djvu-lecun-bengio-book-compression"]
date_created: "2026-07-09T00:00:00.000Z"
tags: ["djvu","lecun","at-t-labs","image-compression","free-knowledge","nips","history-of-ml","internet-archive"]
audits: ["2026-07-09 claude-fable-5"]
drafted_in: ["2026-07-13-we-never-unrolled-it","we-never-unrolled-it"]
---


In his own retelling, [[entity-yann-lecun|Yann LeCun]] describes the purpose of
[[claim-djvu-1998-paper-coauthored-by-lecun-and-bengio|DjVu]] as distributing
high-resolution document scans cheaply over the "newly expanding" internet of the
late 1990s — a bandwidth problem, in an era of slow dial-up, that a
better-compressing format could solve. As a demonstration, LeCun's account says
the team scanned and freely released the entire NIPS conference proceedings
archive — volumes 0–13, covering 1988–1999 — hosting it at **nips.djvu.org**
starting around 2000, where it became a genuine resource for the machine-learning
community rather than a mere technical demo.

This claim is **flagged `[unverified-quote]`**. The source is a single X/Twitter
post; it returned HTTP 402 when the capturing session tried to re-fetch it, so the
account reaches the vault only through a search-engine summary and no verbatim
phrasing was preserved. The free-books framing and the specific NIPS volume range
should be read as *what LeCun is reported to have said*, not as independently
confirmed history. Verification — re-reading the post verbatim or finding LeCun's
account in a citable venue, and confirming the NIPS 0–13 release — is routed at
[[question-verify-lecun-djvu-free-nips-archive]].

There is a second, self-undercutting detail worth recording: **nips.djvu.org
appears to be unreachable** (a DNS failure during the capture session), so the
free-access artifact this story is built around may itself have rotted — a
question also folded into that verification thread.

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.

> [!note] Seek's commentary:
> The theme rhymes with the vault's own
> [[claim-perceptrons-1969-standalone-scan-not-freely-available|no-free-1969-scan]]
> finding: a tool built expressly to make scanned books free forever is, a quarter
> century on, both deprecated as infrastructure (see
> [[claim-internet-archive-2016-stopped-generating-djvu]]) and possibly offline at
> its own demo URL. Free access to text turns out to demand continuous
> maintenance, not just a good format. — Seek