The Modern Library’s own website still hosts the 100 Best Novels list, and it opens with a sentence that would embarrass a stricter institution: “what we call the literary canon is an ever-fluid collection of great books—books that gain their significance from readers engaging with their themes across the centuries.” This is the imprint that, in 1998, published a fixed, ranked list of one hundred novels and sold a display kit around it. The fluid-canon language and the ranked hundred sit on the same page, and the tension between them is the artifact worth examining.
The Modern Library was founded in 1917 by Boni and Liveright to provide inexpensive reprints of European modernist titles and a few contemporary Americans. In 1925 Horace Liveright sold it to Bennett Cerf, a twenty-seven-year-old vice-president who would next found Random House. The curator and the commercial parent share a lineage, and that lineage is not incidental to how the 1998 list was made or how the 1999 nonfiction list was corrected.
The fiction list and its disowned rankings
The 1998 list was a selection of the best English-language novels of the twentieth century, made by the Modern Library imprint from among 400 novels published by Random House, which owns Modern Library. Its stated purpose, as summarized in the standard account, was to “bring the Modern Library to public attention” and stimulate sales of its books. The board consisted of Daniel J. Boorstin, A. S. Byatt, Christopher Cerf, Shelby Foote, Vartan Gregorian, Edmund Morris, John Richardson, Arthur Schlesinger Jr., William Styron and Gore Vidal. All but Gregorian were published by Random House or an affiliate.
That overlap is the first structural fact. The pool of credible judges and the pool of published authors are largely the same pool. This is not a scandal of individual impropriety; it is a feature of publisher-run curation. When an imprint asks its own authors to rank its own catalog, the resulting list is a house document with a board’s names attached.
The second structural fact is stranger. David Streitfeld of the Washington Post discovered that many members of the board that had done the choosing had no idea how the Modern Library had ranked the novels they had chosen. The judges had selected titles; they had not produced the order. The ranking was generated somewhere the judges could not see, and the imprint did not disclose the method. A list that presents itself as collective judgment was, in its most visible dimension—the numbers—unowned by its own judges.
The nonfiction list as procedural correction
By the time the nonfiction list was being prepared, the imprint was addressing criticisms aimed at the fiction list: the age of the board members, the lack of women and minorities, and the seemingly arbitrary rankings. Random House publisher Ann Godoff conceded at the nonfiction unveiling, “We knew we had to improve the process.”
The corrections were specific. The Modern Library hired University of Chicago statistics professor Albert Madansky to do the rankings. The board was expanded with two younger authors (Caleb Carr and Jon Krakauer), two additional women (Carolyn See and Elaine Pagels) and a black author (Charles Johnson). The board met twice to thin the pool from 900 titles to 100, ranking on a zero-to-10 scale before Madansky ran the numbers through a computer program. The list was restricted to books written in English, and a rule that no author could have more than one book on the list eliminated Edmund Wilson’s To the Finland Station and William James’s Pragmatism. Another rule stipulated that board members could not vote for their own books, but five members wound up being represented nonetheless: Shelby Foote, Stephen Jay Gould, Edmund Morris, Pagels and Arthur Schlesinger Jr.
Modern Library board member Stephen Jay Gould had raised the point that the ranking methodology “wasn’t as sound as it could be,” according to publisher David Ebershoff. The statistician was the answer to that objection. The decimals prevented ties. The procedure was now documented, at least in outline, and the ranking had an author—Madansky—rather than an unexplained process.
What the correction did not change
The two lists are a controlled comparison. The fiction list’s ranking was disowned by its own judges; the nonfiction list substituted statistical aggregation for judgment. That substitution fixes the appearance of arbitrariness. It does not touch the structural fact that a publisher owned both the catalog and the curator.
Procedural rigor and commercial function are separable questions. Madansky’s decimals and the no-ties outcome answer “how were these ranked?” They do not answer “why is a publisher’s backlist the universe from which the canon is drawn?” The second question is the one the imprint’s own “ever-fluid canon” language implicitly raises and the ranked-100 format implicitly forecloses. A fluid canon and a fixed ranked hundred are different objects, and the merchandising apparatus treats the second as the sellable one.
The list as distribution instrument
The merchandising was not incidental to the list; it was the list’s operational form. A free merchandising kit including easel-back posters and bookmarks featuring the nonfiction and fiction lists, as well as a new nonfiction reading group guide, was mailed to booksellers. The nonfiction list was released in conjunction with BookExpo America at the opening party for the BEA show. The fiction list’s merchandising paid off: sales for Ulysses rose some five times more than typical yearly sales, and reorders increased across the board for the imprint’s fiction canon. The Modern Library’s subsequent program of returning to print some books that made the fiction list sold more than double the books’ initial 10,000-copy shipments.
These figures are the clearest evidence that the list functioned as a backlist engine. A canon that ships with an easel-back poster is a distribution instrument as much as an evaluative one. The reading group guide, the bookmarks, the BEA unveiling, the bookstore ballots—each is a mechanism for converting a ranking into orders. The list is not a statement about literature that happens to sell books; it is a sales instrument that happens to make a statement about literature.
The residual problem
The 1998–99 sequence is useful precisely because both versions—pre- and post-correction—are documented. The first list exposed the ranking as unowned by its own judges. The second substituted statistical procedure for judgment without changing the underlying commercial function. The procedural improvements answer the appearance of arbitrariness; they do not answer the structural question of who owns the list and what the list is for.
If a trade publisher today issued a ranked “best of the decade” list drawn only from its own catalog, with a board of its own authors, and sold a display kit around it, the procedural improvements—a statistician, a broader board, a no-self-vote rule—would not change the category of the object. The category is publisher-owned curation, and the 1998–99 Modern Library lists are its most fully documented instance.
The open question is not whether the lists were rigged. The sources document ignorance of the ranking method and criticism of the board’s composition, not intent. The open question is what a publisher-owned canon is for, and whether a ranked hundred can ever be the “ever-fluid collection” the imprint now says it believes in. That question belongs to critics and booksellers who can trace the next list’s merchandising kit, its board’s publication records, and its ranking method before the posters go up.
FAQ
Who served on the 1998 fiction board?
Daniel J. Boorstin, A. S. Byatt, Christopher Cerf, Shelby Foote, Vartan Gregorian, Edmund Morris, John Richardson, Arthur Schlesinger Jr., William Styron and Gore Vidal. All but Gregorian were published by Random House or an affiliate.
How was the fiction list ranked?
The retrieved sources do not document the method. What is documented is that many board members did not know how the Modern Library had ranked the novels they chose, as reported by David Streitfeld of the Washington Post.
What changed for the nonfiction list?
The imprint hired statistician Albert Madansky to do the rankings, expanded the board, thinned a pool of 900 titles to 100, restricted the list to books written in English, limited authors to one book each, and barred board members from voting for their own books. Five board members were represented nonetheless.
Did the lists sell books?
Yes. Ulysses sales rose some five times more than typical yearly sales, reorders increased across the board for the fiction canon, and reprints of listed books sold more than double their initial 10,000-copy shipments.