New York Attorney General Eric Schneiderman’s investigation of Exxon Mobil for climate sins has collapsed due to its own willful dishonesty. The posse of state AGs he pretended to assemble never really materialized. Now his few allies are melting away: Massachusetts has suspended its investigation. California apparently never opened one. The U.S. Virgin Islands has withdrawn its sweeping, widely criticized subpoena of research groups and think tanks. In an email exposed by a private lawsuit, one staffer of the Iowa AG’s office tells another that Mr. Schneiderman himself was “the wild card.” His initial claim, flounced to the world by outside campaigners under the hashtag “exxonknew,” fell apart under scrutiny. This was the idea that, through its own research in the 1970s, Exxon knew one thing about climate science but told the public something else. In an Aug. 19 interview with the New York Times, Mr. Schneiderman now admits this approach has come a cropper. He reveals that he’s no longer focusing on what Exxon knew/said but instead on how it goes about valuing its current oil reserves. In essence, Mr. Schneiderman here is hiding his retreat behind a recent passing fad in the blogosphere for discussing the likelihood that such reserves will become “stranded assets” under some imaginary future climate regime. His crusade was always paradoxical. The oil industry reliably ranks last in Gallup’s annual survey of public credibility. The $16 million that Exxon spent between 1998 and 2005 to support organizations that criticized speculative climate models is a minuscule fraction of the propaganda budgets of the U.S. Energy Department, NASA, NOAA, EPA, not to mention the United Nations’ climate panel, etc. etc. The episode ends happily, though, if Mr. Schneiderman’s hoped-for political career now goes into eclipse. But we haven’t finished unless we also mention the press’s role. The “Exxon knew” claim, recall, began with investigative reports by InsideClimate News and the Los Angeles Times, both suffering from the characteristic flaw of American journalism—diligently ascertaining and confirming the facts, then shoving them into an off-the-shelf narrative they don’t support. We have since learned that both the L.A. Times (via a collaboration with the Columbia School of Journalism) and InsideClimate News efforts were partly underwritten by a Rockefeller family charity while Rockefeller and other nonprofit groups were simultaneously stoking Mr. Schneiderman’s investigation. When caught with your hand in the cookie jar in this way, there’s only one thing to do, and last week the Columbia School of Journalism did it, awarding a prize to InsideClimate News. For this columnist, however, the deeper mystery was cleared up last year when I appeared on the NPR show “To the Point” to discuss the subject “Did Exxon Cover Up Climate Change?” (Google those phrases) with ICN’s “energy and climate” reporter Neela Banerjee. Ms. Banerjee has been collecting plaudits all year for her work. The work itself involved revisiting Exxon’s climate modeling efforts of the 1970s. Yet, at 16:28, see how thoroughly she bollixes up what a climate model is. She apparently believes the uncertainty in such models stems from uncertainty about how much CO2 in the future will be released. “The uncertainties that people talk about . . . are predicated on the policy choices we make,” namely the “inputs” of future CO2. No, they aren’t. The whole purpose of a climate model is to estimate warming from a given input of CO2. In its most recent report, issued in 2013, the U.N.’s Intergovernmental Panel on Climate Change assumes a doubling of atmospheric CO2 and predicts warming of 1.5 to 4.5 degrees Celsius—i.e., an uncertainty of output, not input. What’s more, this represents an increase in uncertainty over its 2007 report (when the range was 2.0 to 4.5 degrees). In fact, the IPCC’s new estimate is now identical to Exxon’s 1977 estimate and the 1979 estimate of the U.S. National Research Council. In other words, on the crucial question, the help we’re getting from climate models has not improved in 40 years and has been going backward of late. For bonus insight, ask yourself why we still rely on computer simulations at all, rather than empirical study of climate—even though we’ve been burning fossil fuels for 200 years and recording temperatures even longer. OK, many climate reporters have accepted a role as enforcers of orthodoxy, not questioners of it. But this colossal error not only falsifies the work of the IPCC over the past 28 years, it falsifies the entire climate modeling enterprise of the past half-century. But it also explains the non sequitur at the heart of the InsideClimate News and L.A. Times exposés as well as Mr. Schneiderman’s unraveling investigation. There simply never was any self-evident contradiction between Exxon’s private and public statements. In emphasizing the uncertainty inherent in climate models, Exxon was telling a truth whose only remarkable feature is that it continues to elude so many climate reporters.