Did MIT Press's Parallel Distributed Processing (1986) really sell 6,000 copies on its first day?
Topic harvested from 2026-06-30-why-did-neural-network-research-stall-between-the-1969-perceptrons-book-and-the-mid-1980s-revival. The question asks for primary-source evidence behind the specific "6,000 copies, first day" figure for Rumelhart & McClelland's Parallel Distributed Processing (MIT Press/Bradford Books, 1986). This capture traces the figure to its earliest located textual home and follows how it was retold, rather than confirming the underlying sale actually happened that way — no primary publisher record (MIT Press sales ledger, royalty statement, or dated contemporaneous press account) surfaced after search.
Claim: The literal "sold six thousand copies the day it went onto the market" wording traces to an uncited aside in Dreyfus & Dreyfus's 1988 Daedalus essay, not to any publisher or author record
In "Making a Mind versus Modeling the Brain: Artificial Intelligence Back at a Branchpoint" (Daedalus 117(1), Winter 1988, pp. 15–43), connectionism critics Hubert Dreyfus and Stuart Dreyfus write, in the course of describing connectionism's resurgence:
"Rumelhart and McClelland's book Parallel Distributed Processing sold six thousand copies the day"
"it went onto the market, and thirty thousand are now in print."
(Split into two quote fields here only because a running page-header — "Making a Mind Versus Modeling the Brain 35" — falls between "the day" and "it went onto the market" in the PDF extraction; in the printed journal this is one unbroken sentence on pp. 34–35.)
This sentence carries no footnote or citation of its own in the essay — the footnote immediately before it (64) supports a different, preceding quotation from Rumelhart, Hinton, and McClelland, and the next footnote (65) supports a following quotation from Paul Smolensky. The sales-figure sentence sits between the two, unsourced. The Dreyfuses were philosophical critics of symbolic AI, not participants in PDP's publication — nothing in the essay indicates they had direct knowledge of MIT Press's sales data; the figure reads as an assertion offered for rhetorical effect (to illustrate connectionism's sudden popularity), not as reported fact with a traceable origin.
This is the exact phrasing — "the day it went onto the market" — that matches the topic's framing. It is a quantitative claim, and per the sourcing floor its source (an uncited aside in a secondary philosophical essay) does not independently clear Tier 1–2 for the number itself, even though the essay's authorship and venue are verifiable at Tier 1 for the fact that this is what the essay says. [unverified-quant — needs primary]: no MIT Press sales/royalty record, dated trade-press account, or statement by Rumelhart or McClelland confirming a specific first-day sales figure was located.
- source_url: https://www.amacad.org/sites/default/files/publication/downloads/Daedalus_Wi98_Artificial-Intelligence_Dreyfus.pdf
- source_sha: a4c98279cbae9f280bb9359258e76bf41d983414215fd61fae37e01e94f5fe9b
- source_author: Hubert L. Dreyfus and Stuart E. Dreyfus
- source_date: 1988
- source_title: "Making a Mind versus Modeling the Brain: Artificial Intelligence Back at a Branchpoint"
- source_venue: Daedalus (Journal of the American Academy of Arts and Sciences), vol. 117, no. 1 (Winter 1988), pp. 15–43; essay text at pp. 34–35. Open-access PDF hosted on the American Academy's own site (amacad.org) — the journal's own venue, not a third-party mirror.
- source_tier: 2 (named authors, reputable venue, but the sentence itself is an uncited secondary assertion about a fact outside the authors' own expertise — see floor discussion above)
- source_delight: The number the topic is chasing sits, completely uncited, between two footnoted quotations in a philosophy-of-AI polemic — not in anything MIT Press ever published about its own book.
Claim: A 2018 peer-reviewed history of the PDP volumes repeats the Dreyfus figure but reframes "6,000" as a sold-out first print run, not a single day's sales
Michelle Gibbons's history-of-science article on the PDP volumes' reception states:
"the volumes were resoundingly successful, selling out their first run of 6,000 copies before they even went to press. Two years later there were already an estimated 30,000 in print (Dreyfus & Dreyfus, 1988, pp."
— explicitly citing Dreyfus & Dreyfus (1988), pp. 34–35, for both the 6,000 and 30,000 figures. Gibbons's wording is materially different from her own cited source: Dreyfus & Dreyfus say the book "sold six thousand copies the day it went onto the market" (a one-day sales claim), while Gibbons describes it as the first print run selling out before a second printing could be readied (a stock/inventory claim about the whole run, with no day-one timeframe stated). Both scholars point to the same underlying number (6,000) and the same two-years-later number (30,000 in print), but the "first day" framing is specific to the Dreyfus & Dreyfus wording and is not independently corroborated by Gibbons's retelling — if anything, her 2018 peer-reviewed version quietly drops the "first day" claim in favor of a more plausible "sold out the print run" claim.
This is itself a historiographical fact worth recording: the "first day" figure has been in circulation in the secondary literature since 1988, but the one later peer-reviewed source that engages with it directly does not repeat the "day it went onto the market" detail.
- source_url: https://courses.edx.org/assets/courseware/v1/1d43fffe0d4527c250963b7eb31e1ac9/asset-v1:UniversityofCambridge+2324EDX001+2T2023+type@asset+block/J_History_Behavioral_Science_-_2018_-_Gibbons.pdf
- source_sha: b3e93fb7183e236cd6c499dfc5811ae86bbb7c9aab477e4a3c573ccbe005fd54
- source_author: Michelle Gibbons
- source_date: 2018 (© 2018 Wiley Periodicals; print issue dated 2019)
- source_title: "Attaining landmark status: Rumelhart and McClelland's PDP Volumes and the Connectionist Paradigm"
- source_venue: Journal of the History of the Behavioral Sciences, 55(1):54–70 (Wiley). DOI 10.1002/jhbs.21946. The publisher's own page (onlinelibrary.wiley.com) returned HTTP 403 to archive_page; the text above was read via an unauthorized full-text mirror hosted on a University of Cambridge edX course-asset page (courses.edx.org). The Wiley/JHBS edition is the venue of record; the edX copy is the mirror actually read, named as such per vault convention for unauthorized mirrors.
- source_tier: 2 (named scholar, peer-reviewed journal, drawing here on a secondary citation rather than her own archival find)
- Collision note: this capture is a genuinely distinct angle from the vault's existing Dreyfus-1988 cluster (e.g. claim-dreyfus-1988-connectionism-was-vindicated-not-target, claim-olazaran-1993-cites-dreyfus-1988-for-chronology-not-combined-framing) — those notes draw on the same essay's vindication-of-Rosenblatt argument on the same page (34), not the sales-figure aside a few sentences later. No existing claim-note covers the sales figure itself.
Claim: No primary confirmation of a first-day sale was found; the closest first-person testimony concerns pre-publication doubt about reaching even 5,000 copies
In a 2007 oral-history interview conducted by Gibbons, James McClelland recalled a skeptical remark from an MIT Press/Bradford Books contact while the PDP Volumes were still in production:
"George Manley said oh, you know, you'll never sell 5000 copies of this. I think it was partly if we keep the price down then we'd have a chance."
This is McClelland's own recollection of his own experience and is Tier 1 for that — his testimony about what was said to him. But per the vault's oral-history sourcing convention, a recollection recorded in 2007 about a 1986 (or earlier) conversation is Tier 2–3 for the underlying fact it describes, and in any case this quote is about pre-publication skepticism toward a ~5,000-copy threshold, not a post-publication confirmation of a 6,000-copy, single-day sale. It is consistent with a print run in the 5,000–6,000 range being the publisher's expectation, but it neither confirms nor denies that the run sold out specifically within one day. [unverified-quant — needs primary] for any day-one framing; the broader "first run of 6,000" figure still rests only on the Dreyfus & Dreyfus 1988 aside and Gibbons's 2018 restatement of it, not on anything MIT Press itself is recorded as having said.
- source_url: https://courses.edx.org/assets/courseware/v1/1d43fffe0d4527c250963b7eb31e1ac9/asset-v1:UniversityofCambridge+2324EDX001+2T2023+type@asset+block/J_History_Behavioral_Science_-_2018_-_Gibbons.pdf (same mirror as above; quote is McClelland's words as transcribed and quoted by Gibbons)
- source_sha: b3e93fb7183e236cd6c499dfc5811ae86bbb7c9aab477e4a3c573ccbe005fd54
- source_author: James L. McClelland, interviewed by Michelle Gibbons (2007); quoted in Gibbons (2018)
- source_date: interview conducted 2007; published/quoted 2018
- source_title: oral-history interview with James McClelland, Carnegie Mellon, March 29, 2007, archived at the Center for the History of Psychology, University of Akron; quoted in "Attaining landmark status..." (see Claim 2 for full citation)
- source_venue: oral-history interview quoted within Journal of the History of the Behavioral Sciences 55(1):54–70
- source_tier: 1 for McClelland's own testimony about his own experience; 2–3 for the underlying facts described, per the vault's oral-history sourcing convention
Overall verdict on the core question
[unverified — could not confirm or deny after search]. The specific "6,000 copies on the first day" framing traces to one uncited sentence in a 1988 philosophical essay by critics of connectionism, not to a publisher or author record. A 2018 peer-reviewed history citing that same essay repeats the 6,000 figure but reframes it as a sold-out print run rather than a one-day sale — suggesting later scholarship itself treats the literal "first day" claim with some caution. No MIT Press sales record, dated trade report, or first-person author/publisher statement naming a specific first-day number was found after search.
Further leads
- MIT Press itself priced the two-volume set at $45 and Rumelhart/McClelland did their own typesetting/copyediting to keep costs down — publisher Harry Stanton (Bradford Books imprint) is credited by McClelland as pivotal to the volumes' circulation and press coverage (Gibbons 2018, pp. 62–64). Worth a dedicated capture on Stanton/Bradford Books' role if the vault wants the publishing-dynamics angle.
- Gibbons (2018) reports a University of Minnesota Center for Cognitive Sciences committee ranked the PDP Volumes 5th among the 100 most influential 20th-century cognitive-science works — original list link now dead, only a mirror cited (web.mnstate.edu/schwartz/cogsci100.htm).
- Nature's review called the PDP Volumes' publication "one of the publishing events of 1986" ("PDP in paperback," 1987, p. 592, per Gibbons) — a possible lead to a contemporaneous 1986–87 trade-press account that might carry an actual sales figure with its own sourcing, not yet checked directly.
- Schneider (1987) reported connectionism talks at the Cognitive Science Society meeting rose from 17% (1984) to 23% (1985) to 31% (1986) of the program — a quantitative claim not yet checked against Schneider's original text; flagged but not pursued here (out of scope for this capture).
Entity candidates
- David Rumelhart — person — co-editor/senior author of the PDP Volumes; the book the sales figure is about; his own 1998 oral-history recollections (quoted secondhand in Gibbons) describe pricing decisions but not sales figures directly.
- James McClelland — person — co-editor/author of the PDP Volumes; primary living source for oral-history testimony about the book's production and reception (2007, 2017 interviews with Gibbons).
- Hubert Dreyfus — person — co-author of the 1988 Daedalus essay that is the earliest located source of the literal "first day" sales claim; longtime philosophical critic of symbolic AI, not a PDP participant.
- Stuart Dreyfus — person — co-author of the same 1988 essay; also independently relevant to the vault's backpropagation-origins cluster via his 1962 chain-rule work (see claim-dreyfus-1988-connectionism-was-vindicated-not-target).
- Michelle Gibbons — person — author of the 2018 Journal of the History of the Behavioral Sciences article that restates and reframes the sales figure; conducted the cited 2007/2017 McClelland oral histories.
- Harry Stanton — person — Bradford Books/MIT Press editor credited by McClelland as central to the PDP Volumes' publication and press coverage; a concept/entity not yet in the vault.
- George Manley — person — MIT Press/Bradford Books contact who reportedly doubted the volumes would sell even 5,000 copies, per McClelland's recollection; minor but names a specific pre-publication skepticism data point.