The Last Best Place on the Internet: A Human History of Wikipedia, by Richard Cooke (W.W. Norton & Company, 320 pp., $29.99)
Since its early 2000s launch, Wikipedia has cultivated a reputation among educated professionals as a dependable, if quirky, source of information vetted by the wisdom of consensus and community editors. When biases arose over the years, skeptics were told that these were necessary growing pains of a massive, open-source project serving the common good.
Finally, a reason to check your email.
Sign up for our free newsletter today.
This assumption of benign neutrality rests on the idea that Wikipedia is simply a website by and for curious people seeking to democratize knowledge much like the printing press. Richard Cooke’s new history of the organization, The Last Best Place on the Internet, is the latest work to indulge this naivete. Cooke rightly identifies Wikipedia’s critical role in shaping the modern information ecosystem, but he refuses to acknowledge the platform as a captured institution that will train artificial intelligence to treat progressive orthodoxy as the only legitimate viewpoint.
Wikipedia began in 2001 as an informal feeder project for Nupedia, an ambitious online encyclopedia founded the previous year by internet entrepreneur Jimmy Wales and philosopher Larry Sanger. Nupedia was conceived during the early dot-com boom to challenge the dominance in reference-text publishing of industry stalwarts such as Encyclopedia Britannica and Microsoft’s Encarta. Wales and Sanger believed that an unprecedented combination of open-content licensing, digital distribution, and an unpaid global labor pool could distinguish Nupedia and, in Cooke’s words, “remake the world.”
Yet Nupedia’s editorial process, which included seven rigid steps of peer review, editing, and fact-checking, stalled the project almost immediately. Rather than inspiring a grassroots campaign to expand accessible knowledge, the cumbersome workflow alienated contributors. Wales conceded that his own drafting of an article felt more like “homework.” In the site’s first five months, only a single article, on the musical concept of “atonality,” reached publication quality, and the minimal web traffic suggested a bleak future for the great experiment. Moreover, the dot-com crash depleted any capital that Wales’s parent company could have used to salvage the site by paying moderators.
Yet according to Cooke, it was Nupedia’s spectacular failure that allowed its founders to reimagine the project and conceive of the idea that eventually became Wikipedia. At a dinner in late 2000, computer programmer Ben Kovitz introduced Sanger to the “wiki” model (named after the Hawaiian word for “quick”), which eliminated traditional editorial gatekeeping by allowing anyone to draft or edit an article in real time. Realizing its capacity to bypass Nupedia’s bottlenecks, Sanger boldly emailed the project’s mailing list on January 10, 2001: “Let’s make a Wiki,” he proposed, encouraging skeptical academics to “humor me . . . Go there and add a little article. It will take all of five or ten minutes.” Thus was born Wikipedia.
By the end of 2001, Wikipedia had published thousands of articles across 16 languages. The site’s real-time coverage of September 11 cemented its advantage over legacy reference sources as an agile, living archive that could document history as it unfolded. In 2003, amid threats of Spanish Wikipedia’s defection if the site elected to run advertisements, Wales solidified Wikipedia’s status as a nonprofit, launching the Wikimedia Foundation, which still governs the platform today.
Yet Wikipedia’s anarchical structure almost imploded the project early on. Sanger left in March 2002 in protest over the lack of editorial discipline and refusal to police bad actors abusing the site. After several editorial scandals that threatened the platform’s credibility in the ensuing years, Wales, too, had to confront the possibility that an egalitarian ethos undermined the platform’s goal to be taken seriously.
To impose order, Wales established editing restrictions and committees to govern the site. By 2007, power over editing was functionally consolidated into the hands of veteran insiders who could navigate Wikipedia’s arcane bylaws and devote extended time to curation. The site’s early “inclusionist” ethos, which celebrated entries on esoteric subjects and cultural minutiae, yielded to the “deletionists,” who sought to gatekeep the types of articles considered for publication. The most consequential of these new mandates was the codification of strict sourcing guidelines that empowered tenured editors to determine which publications counted as reliable. In time, this shift would effectively outsource Wikipedia’s standard of truth to a legacy-press apparatus increasingly dominated by progressive orthodoxy.
Cooke views this bureaucratic maturation as a triumph. To the extent that he acknowledges bias on the platform, his concerns are merely demographic: he laments the site’s heavily male editor base, its Western-centric focus, and gender gaps in the platform’s profiles. Yet his authoring of a comprehensive history of Wikipedia in 2026 without mentioning its systemic blacklisting of conservative media or its purging of pro-Israel perspectives requires what amounts to an intellectual abdication.
Beginning with its 2017 decision to deprecate the U.K.’s Daily Mail as a “generally unreliable” source whose reference use ought to be “generally prohibited,” Wikipedia editors abandoned the case-by-case evaluation of individual citations in favor of a centralized classification regime that would designate news outlets as “reliable” or “unreliable” and purge claims derived from those deemed unfit. Fox News’s political reporting, along with the New York Post and the Daily Wire, were all stamped as “generally unreliable” sources, while The Nation and the New York Times maintained general reliability status. In perhaps the most egregious example of this double standard, administrators prevented the New York Post’s Hunter Biden disclosures from appearing as a recognized subject on the site only weeks before the 2020 election.
These biases extend beyond the confines of domestic American politics. In June 2024, following a sustained campaign by anti-Zionist editors coordinating in private, off-site Discord servers, Wikipedia downgraded the Anti-Defamation League (ADL)—one of America’s oldest Jewish civil rights organizations—to “generally unreliable” on the Israeli-Palestinian conflict and the definition of anti-Semitism itself. While mainstream Jewish outlets and organizations like the ADL and Jewish Virtual Library were labeled as unreliable sources, Qatar’s state-sponsored Al Jazeera, which routinely broadcasts Hamas propaganda, remains “generally reliable.”
Unsurprisingly, Wikipedia’s decisions have disseminated a distorted view of Jewish and Israeli history. Wikipedia’s page on Zionism once described the term as “the national movement of the Jewish people that supports the re-establishment of a Jewish homeland in the territory defined as the historic Land of Israel”; the definition today reads, “an ethnic nationalist movement that emerged in late 19th century Europe to establish and support a Jewish homeland through colonization in the region of Palestine.”
Cooke is correct in his thesis that Wikipedia’s modern dominance stems from its role as the bedrock for AI, yet his book falls flat by ignoring how the platform’s biases have corrupted that very data. In a future where AI likely shapes the average American’s outlook and becomes a crucial tool in policing, education, and government, the possibility that the information underlying these important functions is corrupted should concern all Americans.
While The Last Best Place on the Internet offers a fascinating history of Wikipedia and accurately assesses the encyclopedia’s current power, the book tells only half the story. In refusing to examine where Wikipedia has gone wrong, Cooke was either ignorant of the organization’s full history or complicit in its attempt to brand itself as the neutral public square. Honest critics and millions of users know otherwise.