The Editing of Infinite Jest

How a 750,000-word manuscript became the book

Infinite Jest did not arrive whole. It was a decade in the making, a tangle of false starts and abandoned drafts before it was a manuscript, and once it was a manuscript it went through a two-year editing war with its publisher that left it shorter than the object David Foster Wallace first handed over — though by how much, no two of the people involved remember the same way. This essay follows that process from the 1986 Yaddo application that first bore the title to the 1997 paperback that fixed the last errors — and it makes one running argument, which the composition record settles about as firmly as anything in the archive: the folklore has the story backwards. The legend is of a bloated draft the publisher cut down. The documented truth is of a book Wallace cut modestly and expanded enormously after 1993, whose difficulty and open ending were fought for scene by scene and kept on purpose. The published book is, in the editor's eventual phrase, "the Writer’s Cut" (iw-pietsch-2016-full).

The single richest document behind this account surfaced only recently: Michael Pietsch's actual December 22, 1994 second-read letter, with its four-page list of suggested cuts, in scans of Wallace's own marked-up copy from the Harry Ransom Center (pietsch-letter-dec-1994). Around it sit the biography D.T. Max built from the Ransom Center papers and Pietsch's files (dtmax-every-love-story), Steven Moore's page-by-page collation of the 1993 draft he owns (moore-first-draft-ij), the five days of tape David Lipsky recorded on the 1996 book tour (lipsky-becoming-yourself), Pietsch's own retrospectives from 2009 through 2026, and the memorial program that first printed Wallace's margin notes fighting the cuts (fivedials-memorial). Where those sources give different numbers — and on the cut count they give at least five — this essay reports each as it stands and does not average them.

The long gestation, 1986–1992

The title is older than almost everything else. Wallace's September 1986 Yaddo application already names a novel-in-progress called Infinite Jest, so the Hamlet-and-ghost frame and presumably the Incandenzas predate the end of his Arizona years (dtmax-every-love-story). Fragments scattered further still: in the fall of 1986 Gale Walden found draft pages under his bed that Wallace described as fiction about a terrorist organization in Canada, which Max identifies as the germ of the Marathe/Steeply dialogues, set on mountains like the ones the two of them hiked outside Tucson (same source). An early two-page draft titled "What Are You Exactly" has a boy named "David" — later Hal — visiting the "professional conversationalist" who is his father in disguise, the scene that would open the book for years before the Year of Glad frame displaced it. But it did not cohere. Wallace's own account, to Marshall Boswell: "I started IJ or somethin' like it several times. '86, '88, '89. None of it worked or was alive. And then in '91–'92 all of a sudden it did" (dtmax-every-love-story).

What made it live was, in part, what he had lived through. After a 1989 breakdown and four weeks at McLean, Wallace was discharged to a Brighton halfway house — the direct model for Ennet House — where he found that "less intellectual people" were often better at dealing with life and picked up the recovery stories that became the book's Boston voice (dtmax-2009-the-unfinished). By his own tour-era telling he spent "hundreds of hours" in three of Boston's twelve halfway houses, then "twisted everything way around" (lipsky-becoming-yourself). That lived substrate is the subject of addiction and recovery; what matters here is that the AA half of the novel could not have been written before 1989, which is one anchor under the reading that Infinite Jest is two books — a comic/tennis strand and a recovery strand — fused by a later pass.

The business machinery started earlier and smaller. Wallace's agent, Bonnie Nadell, was "a baby agent" at Fred Hill reading the slush when a query arrived from a 24-year-old MFA student in Tucson announcing "a diachronic chapter" — a word she had to look up — of what became The Broom of the System, the first book she ever sold (mcnally-making-of-ij-nadell-pietsch). By the spring of 1992 she had 250 pages and a proposal for the new novel, and Wallace's ask was frank: he warned "This is gonna take me a long time", needed money for health insurance, and asked "Could you get me, like, $75,000?" (same source). Gerald Howard at Norton could offer only $35,000; Michael Pietsch bought it for Little, Brown in June 1992 for $80,000, with a line Nadell still quoted thirty years later (same source) and Wallace himself put on record — he wanted the book "more than I want to breathe" (dtmax-every-love-story).

The submission and the mess, 1993

In May 1993 Wallace mailed the first 400,000 words — some 750 manuscript pages — in narrow margins and tiny type to hide the length. Pietsch "pulled out a calculator" and projected past 1,200 pages (dtmax-every-love-story). That December Wallace gave a photocopy to Steven Moore, one of only three people to read the 1993 draft, to suggest cuts; Moore's later collation is the closest public view of the book's composition. His verdict on the physical object was that the four-inch manuscript would give even a seasoned editor "the howling fantods" (moore-first-draft-ij). It was a working mess — patchwork fonts, nested pagination, footnotes still at the bottoms of pages, and some sections deliberately set in 8-point type to disguise the length from Pietsch, "who of course wasn't fooled for an instant" (same source).

The manuscript also had a reader inside the family, and her recollection is the rare firsthand view of the physical production. Wallace's sister Amy recalls that he wrote the novel longhand and typed it himself, and that he paid her to proofread the manuscript as he mailed it to her in enormous envelopes, in irregular bursts of production — the only time, she says, she was ever that aware of what he was working on, since he otherwise never announced a project; she suspects the errors she found were bones he threw her on purpose, and remembers the writing as harrowing, taking a very long time and a great deal out of him (paraphrased from audio; amy-wallace-havens-unedited-interview-2009). Her account is the firsthand version of an arrangement the biography records secondhand (dtmax-every-love-story).

Pietsch's first written response named the book's shape with a metaphor that has outlived his authority to have coined it. It was, he wrote, "a novel made up out of shards, almost as if the story were something broken that someone is picking up the pieces of" — and through those shards "there surfaces … an 'I' who may be the one trying to put everything together" (dtmax-every-love-story), an early, independent witness to the Hal-as-narrator question. But he pushed back hard on one thing: the Québécois superstructure. He doubted its "ornately bizarre-to-goofy" machinery and warned that "the dog is awfully shaggy already" (same source). This mattered, because it set up the central irony of the whole edit.

The great expansion, 1994

Wallace's answer to "the dog is awfully shaggy already" was to grow the shag. Through 1994 he expanded the "interAmerican huggermugger" rather than trimming it, believing the book needed it for his "seat at the table with Pynchon" (dtmax-every-love-story). Moore's collation is emphatic on how much of what readers take to be foundational is in fact late: through most of the 1993 manuscript the exhibition opponents are Ukrainians, not Québécois, and the opening geopolitical crisis is Sierra Leone, not southern Quebec — the whole O.N.A.N./Quebec apparatus is substantially a post-1993 addition, and so are all the Marathe/Steeply dialogues, most of the A.F.R. material, the videophony essay, and the Eschaton aftermath (moore-first-draft-ij). The political frame that anchors nationalism-onan was welded on last.

Two structural inventions of 1994 shaped the book we have. The first is the Year of Glad opening — Hal's collapsing admissions interview (pp. 3–17) — written in early 1994 to replace the professional-conversationalist scene that had opened the book for years. Max's reading of it is that the new frame "deliberately multiplied the explanations" for Hal's state (the DMZ, the marijuana withdrawal, having seen the Entertainment): "Hints for all three of these plot twists are in the text," and Wallace "was never going to let the reader settle on one" (dtmax-every-love-story). The novel's single most theory-generative structure is a late addition — a fact that hangs over every argument about the-ending.

The second invention is the endnotes. Wallace pitched them to Pietsch in April 1994 as a length-management device, vowing to "fight w/all 20 claws to preserve" them: they would "allow a discursive, authorial intrusive style" without turning the book into Finnegans Wake, would "mimic the information-flood and data-triage" of future American life, and would make the reader move physically back and forth through the book in a way that mirrored the story's own concerns. "I pray this is nothing like hypertext," he added (dtmax-2009-the-unfinished). Pietsch counter-offered footnotes and lost. The endnotes were also, from the first, an argument about design as much as length, and Wallace held the same line on the whole book's aclimactic shape: a January 1994 letter proposed an ending "not so much anticlimactic as aclimactic," citing the book's "annulation, paralysis, undecidability" (dtmax-every-love-story).

The full manuscript — some 750,000 words — arrived in late June 1994, brought to New York "handcuffed to a wrist" (dtmax-every-love-story). Wallace had been contracted to deliver on January 1 and delivered June 18; then came six months of silence during which, he remembered, "I was really mad" (lipsky-becoming-yourself).

The pivotal document: the December 1994 letter

When Pietsch did respond it was twice, and the second time is the document the whole legend orbits. In October 1994 he objected to the ending in a line the archive has come to lean on: of "all those dingleberries" in the air he complained, "and we don't get to find out who or how or why?" (dtmax-every-love-story). Then, on December 22, 1994, he wrote again — two pages of reaction on Little, Brown letterhead plus a four-page list of suggested cuts keyed to manuscript pages 496 through 1603 — and, crucially, capitulated on the ending he had just objected to. His second-read verdict on the close: "The ending of the novel, the horror of Gately's hitting bottom, is gorgeous and very very powerfully sad" (pietsch-letter-dec-1994).

The letter's tone is the first surprise. Every cut is framed as a question, not a demand. The covering rationale is entirely reader-facing: "I'm still hoping there are ways to make the novel much shorter, not because any one piece of it isn't wonderful but because the longer it is the more people will find excuses not to read it", and the attached pages offer "chapters and scenes that maybe can come out without killing the patient" (pietsch-letter-dec-1994). This is the posture behind the doctrine Pietsch would state plainly decades later: "An editor only suggests" (mcnally-making-of-ij-nadell-pietsch).

The second surprise is how little of the list executed. Cross-checked against the printed novel, most of what the December letter flagged survived: Steeply's father's MAS*H story, James Incandenza's father with the mattress, Idris Arslanian blindfolded, Ewell's money-stealers' club, the Beverly lobster memory, the boys in the tunnel, Molly Notkin's interview, the Fackelmann ending itself. Pietsch's green-ink "Yes" beside a suggestion did not even guarantee a cut — the mattress scene is marked "Yes" and runs in print across pp. 491–503 (pietsch-letter-dec-1994). The verified executions are surgical, not structural: the entire jamais-vu thread (the margin instruction reads "Eliminate all jamais vu"; the word occurs zero times in the finished book), Lyle's tale of the wise old bee ("643 Lyle's story of wise old bee: cut."), the myth of the hairy woman, and the vegetable-woman eavesdropping scene. The draining of the duck pond got something subtler than a cut. Pietsch called the pond-draining "the hardest scene in the novel to keep reading" and offered two remedies: drop it and "just say later that AFR found where Joelle was by kidnapping and interviewing her engineer", or leave it "direly condensed" — and the book does both at once. The published scene (pp. 620–626) adopts the editor's engineer-capture mechanism as plot, and the annual pond-draining survives inside it, compressed to a few lines of description — the clearest case on record of an editor's proposed replacement becoming published text, with the original scene revised down rather than cut (pietsch-letter-dec-1994).

The letter is also a rare record of an expert reader failing, on two full reads, to decode the machinery — and of Wallace licensing that difficulty rather than fixing it. Pietsch could not keep the separatist cast straight — its "whole supporting cast", he wrote, "remains indistinct" — could not place the samizdat's origin ("But I still don't get how the tape appeared in the first place, and why in Canadian hands." … "Have I missed something or is it meant to be obscure?"), and could only see the wraith's moved objects as an unexplained pattern. Wallace's answer, in Pietsch's paraphrase of the compact they reached, was that "This aspect of the novel remains shaggy, which is mostly okay" (pietsch-letter-dec-1994). The obscurity survived a professional reader's direct queries, which is the archive's strongest evidence that it is built in rather than botched — the argument worked out in full for the political plot in nationalism-onan and for the ending in the-ending.

And it settled the endnotes. Pietsch, who had wanted footnotes, conceded: "You're right that notes should go at the end." He went on: "A novel with lots of long notes at the bottoms of pages looks much more academic and daunting than one with a section at the end. Maybe we'll package it with a bookmark to help readers keep their place" (pietsch-letter-dec-1994) — the bookmark that shipped with the first edition, decided here in a Christmas-week letter.

What Wallace fought for, and what he conceded

Wallace's side of the negotiation survives best in his margin notes, first printed at his 2008 memorial and better known from Pietsch's 2009 essay. They are a running comedy of resistance: on one Swiftian line, "which I will cut, you rotter"; on the Eschaton stretch, "my canines are bared on this one"; on a passage he would not justify at length, an offer of "5000 words of theoretico-structural arguments" (infinite-summer-pietsch-editing-ij); and, defending one scene Pietsch proposed cutting, the flat accounting "it introduces three characters, five themes and two settings" (fivedials-memorial). His code, per Max, was economical: a circled suggestion meant "total acquiescence" to the demand, a dash meant "bared teeth" (dtmax-every-love-story).

He won small things and lost small things, and the pattern is instructive. He won the Orin/Cardinals fly-in — "Just give me this one", of a page and a half he loved — and he won the real math proof in the Eschaton apparatus, telling Pietsch "this is real math" and that he wanted readers to "understand that I know real math" (mcnally-making-of-ij-nadell-pietsch). He conceded the LaMont Chu / Lyle fame exchange, "much cut down", and cut the denser tennis material before submission because "very few people find that stuff interesting" (lipsky-becoming-yourself). His favored evasions when he could not win outright were to condense rather than delete, or to relocate a beloved passage into the endnotes — the place where his favorite passages went to "make their last stand" (dtmax-every-love-story). When his sixteen-page February 1995 reply came back, its governing confession was almost sheepish: he had presumed "it was OK to make a reader read the book twice" (same source).

One moment nearly broke the good faith. When Pietsch mentioned pricing the book at thirty dollars, Wallace panicked that the cutting was commercial — "these cuts are for commercial reasons", around Christmas of 1994 — and had to be talked down by outside readers, Richard Powers and Steve Moore among them (lipsky-becoming-yourself); DeLillo's counsel, as Wallace relayed it, was that when "you feel incipient bladder tumors at the thought of cutting" something, don't cut it (dtmax-every-love-story). What he came out saying about Pietsch, though, was unstinting: "that book is partly him. A lot of the cuts are where he convinced me of the cuts" (lipsky-becoming-yourself).

The endnotes as compromise and as design

The endnotes deserve their own beat, because they were the point where a length problem and a design idea became the same thing. Roughly half of them are post-1993 additions (moore-first-draft-ij), and the system doubled as the refuge for material Pietsch wanted cut (dtmax-every-love-story). Even in the December 1994 letter Pietsch was still pressing on them — "I still hope you'll look for many, many notes you can cut" (pietsch-letter-dec-1994) — and the notes nonetheless grew. Wallace's account of what they were for is unambiguous — the claim that the book reads differently with and without them is his own, quoted and weighed in what-wallace-said — and on the tour tape he made the stakes plain: "There’s just a whole lot of plot stuff that isn’t clear, if you don’t read the endnotes." Asked whether the book was meant to be hard or easy: "The book was designed to be both, I guess" (lipsky-becoming-yourself). The apparatus that most tries a reader's patience is thus one of the most deliberately engineered features of the book — the structural argument taken up by annularity.

What the cuts tell us

The most persistent piece of Infinite Jest folklore is that the publisher cut 300 to 400 pages from a monstrous draft. Weighed against the documentary record, that story is a myth, and the correction is graded strongly-supported: Wallace cut comparatively little and added enormously after 1993. The cut count itself dissolves into a family of figures that measure different things against different states of the manuscript, and no source reconciles them. Moore, working from the 1993 draft, counts only about 40 pages cut against 200-plus added, and finds the legend that the publisher "cut 300–400" pages a puzzle (moore-first-draft-ij). Pietsch, across three retellings, gives "around 250 manuscript pages" accepted (2009), "around 150 pages were cut … call it 75,000 words" (2016), and "I probably suggested cutting about 250 pages" (2026) — and in the same 2026 breath names the source of the big number: "He said at one point that I cut 500 pages. That’s a great exaggeration" (infinite-summer-pietsch-editing-ij, iw-pietsch-2016-full, mcnally-making-of-ij-nadell-pietsch). That charge lands on Wallace himself. On the 1996 tape he put it at "about five hundred pages longer. Of which four hundred unambiguously needed to go, and the other hundred was painful" (lipsky-becoming-yourself) — one stop in a run of author-side figures that wandered with every retelling, traced in full in what-wallace-said; to David Markson he inflated it further, calling the book "cut by 600 pages" (dtmax-every-love-story); Nadell, for her part, remembers a "600,000-word manuscript when David was done" (mcnally-making-of-ij-nadell-pietsch). The figures do not settle. What settles is the direction of the work and the author's ownership of it — a popularity/authority inversion worth keeping, in which the modest, scope-preserving edit is the consensus of everyone who was in the room and the big-cut legend is what the wider culture believes.

Two firmer facts fall out of the record. The first is that the December 1994 list, read at scene grain, confirms the reversal from the primary side: the editor's actual maximum ask was per-scene surgery, and most of it was declined (pietsch-letter-dec-1994). The second came out of the twentieth-anniversary preparations. For a possible "writer’s cut" edition the Wallace Trust had the Ransom Center papers reviewed for the excised sections — and there were none. The marked-up draft had been returned to Wallace at publication, and "he doesn’t appear to have kept it." Pietsch's conclusion is the doctrine this essay opened on: "The cuts David made were all intentional–the version we have is the version he wanted, not something imposed on him. It is the Writer’s Cut" (iw-pietsch-2016-full).

Two related composition findings sit at the same strongly-supported tier. The post-1993 stratum — the Year of Glad frame, the whole Marathe/Steeply and A.F.R. apparatus, the mold-eating incident, half the endnotes — is documented by three independent witnesses and disputed by none. And the ending was always the ending: Moore reports that "Wallace knew early on how he wanted it to end" (moore-first-draft-ij), and the aclimactic close was argued for explicitly through the editing war — from the January 1994 "aclimactic" letter through the October objection and the December capitulation to the February 1995 defense that the book tells us "what's happening to Gately by end, about 50% of what's happened to Hal," and little but hints about Orin (dtmax-every-love-story). Twenty years on Pietsch confirmed it from the editor's chair: the two of them "discussed the fact that the story doesn’t narrow to a dramatic conclusion, and that some readers would find its ending frustrating. He was content with this" (iw-pietsch-2016-full). The missing year between p. 981 and p. 3 is a withholding, not an omission — the argument carried in full by the-ending.

One caution the record keeps against itself: the manuscript did carry outline notes for the missing year. After Gately's final line, Moore found sketched plans — "Hal and Helen," "Pemulis and DMZ", "Gately and Joelle" — and an image of a victim whose captors "sewed his eyes open" and "Superglued the back of his head to the wall" (moore-first-draft-ij). But these are notes for chapters never drafted across three further years of work; the book contemplated the year and chose not to write it. What was planned and abandoned is evidence of the withholding, not a defeat of it.

The last rounds, publication, and the corrections tail

The editing did not stop with the big December negotiation. In the spring of 1995 Pietsch, by Wallace's account, sent a second cut letter — mostly footnotes this time — of which Wallace "ended up taking about half" (lipsky-becoming-yourself). One of the book's few pieces of late-added resolution came at the editor's request: Pietsch asked for "some indication of the fate of Orin", and on June 11, 1995 Wallace faxed the tumbler-and-roaches scene now at pp. 971–972, with a note calling it an insertion "about which I'm not exactly qualmless" (dtmax-every-love-story) — a scene that mirrors the fate Orin had inflicted on the roaches, the mirroring being the pitch. A quieter, externally forced revision came in May 1995, when Mary Karr warned Pietsch that Ennet House scenes drew on confidential recovery talk; Wallace changed names and details and added the evasive denial on the copyright page, a whole layer of revision invisible to Moore's earlier collation (same source).

Two smaller decisions shaped the finished object. The working title "A Failed Entertainment" — Wallace's own name for the book's designed failure as entertainment, a claim quoted and weighed in what-wallace-said — was killed by Pietsch, who refused to put it on a book "people were supposed to buy" because "it's not" (dtmax-every-love-story). And Little, Brown's dare-marketing (teaser postcards promising a "masterpiece") appalled its author, who begged them to stop — "I'm 33 years old; I don't have a 'masterpiece'" (same source). The book was published in February 1996. The text scholars should cite is the 1997 paperback, which corrected errors Moore and others had caught — Hal's age reset from "thirteen" to "eleven" among them (moore-first-draft-ij).

What the published book owes the process

Two of the people closest to the making left the clearest verdicts on what the process produced, and both were delivered at Wallace's 2008 memorial. Franzen described the most intense interaction of the friendship as "sitting alone in my armchair, night after night, for ten days, and reading" the "manuscript of Infinite Jest", which he called the book in which Wallace had "arranged himself and the world the way he wanted them arranged" — and he named its design, a thousand pages whose humor "became less and less and less funny, section by section," until the title might as well have been "Infinite Sadness" (fivedials-memorial). Nadell, at the same memorial, testified that Wallace's habit of the unresolved close was old and principled: The Broom of the System "ends without the last word", and when she tried to make him wrap it up conventionally he explained "using Wittgensteinian philosophy" why it had to stay (fivedials-memorial).

That is finally what the editing of Infinite Jest shows. The difficulty is not residue; it is design that a professional reader pressure-tested and could not talk out of the book. The endnotes, the withheld ending, the shaggy separatist plot, the deliberately multiplied causes of Hal's collapse — each survived a direct editorial query and stayed. The cutting that did happen was mostly condensation and relocation, at scene grain, agreed to by an author who kept adding back more than he removed. Wallace's own mission statement for the book — he wanted to "restructure worlds and make living people feel stuff" — is the right frame for the whole two-year negotiation (dtmax-2009-the-unfinished); Pietsch thought the line important enough to close his memorial tribute on it (fivedials-memorial). The version that does both is, by every account of the people who watched it happen, the one he wanted.

Sources

• pietsch-letter-dec-1994 — the primary document: Michael Pietsch's December 22, 1994 second-read letter and four-page cut list, in scans of Wallace's marked-up copy; the record of what survived and what was cut, the endnote capitulation, the licensed obscurity, the Gately-ending praise.
• moore-first-draft-ij — Steven Moore's page-by-page collation of the 1993 draft he owns: what was cut, what was renamed, and how much was added after 1993; the ~40-cut/200-plus-added count; the surviving ending and its outline notes.
• dtmax-every-love-story — the definitive month-by-month composition biography (Ransom Center papers + Pietsch's files): the 1986 germ, the 1994 expansion, the October objection and February defense, the markup code, the Orin insertion, the killed subtitle.
• dtmax-2009-the-unfinished — the New Yorker profile that seeded the biography: the endnote pitch in Wallace's own words, "more than I want to breathe", the mission statement.
• lipsky-becoming-yourself — five days of book-tour tape (March 1996): the author's own cut arithmetic, the two-letter structure, the "$30" commercial scare, "A Failed Entertainment", the endnotes-by-design statement, Pietsch as "partly him."
• infinite-summer-pietsch-editing-ij — Pietsch's 2009 retrospective: the every-section "Can the book possibly live" test, "around 250 manuscript pages," and the margin notes ("you rotter," "canines are bared", the theoretico-structural offer).
• iw-pietsch-2016-full — Pietsch's 2016 interview: the "designated knucklehead" role, "the story doesn’t narrow to a dramatic conclusion … He was content with this", the ~150-pages/75,000-words figure, the Ransom Center negative, "It is the Writer’s Cut."
• mcnally-making-of-ij-nadell-pietsch — the 2026 Nadell/Pietsch oral history: the acquisition and the $75,000 ask, "An editor only suggests", the Wallace-said-500 attribution, the refused math proof, Nadell's 600,000-word figure.
• fivedials-memorial — the 2008 memorial program: the earliest print of the cut-war margin notes, the new "three characters, five themes and two settings" defense, Franzen's ten-night read and funny-to-sad design claim, Nadell's Broom-non-ending testimony.
• amy-wallace-havens-unedited-interview-2009 — paraphrased from audio; the sister's firsthand production testimony: longhand composition, self-typing, and the paid proofreading of the manuscript mailed in enormous envelopes.