Should Journalists Use Lies to Fight Lies?

Zhao Yan has been in detention since September 2004. He was the research assistant and news assistant for the New York Times Beijing bureau, and the Chinese government accused him of leaking state secrets. The specific allegation concerns a Times story, published under Jim Yardley’s byline, that correctly reported Jiang Zemin was stepping down from his final military post several days before the official announcement. Zhao has been held without formal trial through most of 2005 and into this year. His case attracts periodic attention and then fades. The Chinese authorities are presumably content with that rhythm.

The question his case raises is not primarily a civil liberties question, though it is that. It is an epistemological one. When you are reporting from a country where official information is systematically false — where press releases are theater and statistics are political instruments — what methods are licit? The classic answer is to use human sources, documents, and triangulation against independent data. But that answer assumes independent data exist. In China’s case the gap between stated and actual is not marginal. China’s provincial GDP figures, famously, have consistently summed to more than the national total. In some years the discrepancy has been 10% or more. Every provincial party secretary has an incentive to report growth, so growth gets reported.

Google launched google.cn in January 2006 with self-censorship built in. Search queries for “Tiananmen,” Falun Gong, the Dalai Lama, and a list of other topics return results scrubbed of discomforting content. Sergey Brin and Larry Page were asked about this in front of a congressional hearing in February; their answer, roughly, was that a censored Google was better than no Google. Microsoft and Yahoo had made similar calculations earlier. Yahoo’s case was the most damaging: the company provided user account information to Chinese authorities that led to the imprisonment of journalist Shi Tao, who received a 10-year sentence for e-mailing notes from a Communist Party internal briefing to a foreign website. Reporters Without Borders ranked China 163rd out of 168 countries in its 2006 press freedom index.

The parallel with economic statistics is not metaphorical. It is structural. A researcher trying to estimate actual Chinese output, actual inflation, or actual unemployment faces a problem methodologically similar to the one a foreign correspondent faces when trying to report on an event the government wants suppressed. Both are working against a state apparatus that controls the primary data generation. Both must choose between (a) reporting the official figure with a disclaimer, (b) constructing an independent estimate from proxy data, or (c) refusing to engage with the number at all. The third option is honest but useless. The first is what most press coverage defaults to. The second is what researchers like Thomas Rawski at Pittsburgh have been attempting — Rawski’s 2001 paper questioning Chinese GDP growth figures during the Asian financial crisis used electricity consumption as a check and found a significant discrepancy.

There is a harder question underneath this, one that the journalism ethics literature handles poorly: if the ground truth is propaganda, does an accurate account of propaganda constitute accurate journalism? The British documentary maker who films a North Korean mass game and broadcasts it without commentary has not lied. He has also not told you the truth. The economist who reports Chinese official CPI figures has not fabricated data. He may, however, be transmitting fabrications while lending them his credibility.

Joseph Kahn and Jim Yardley at the Times have run a sustained effort to build Beijing reporting around on-record sources where possible and triangulated inference where not. The approach is labor-intensive and requires a network of contacts that takes years to build and can be destroyed in a week by a case like Zhao Yan’s. Each arrest or source intimidation is, from the government’s perspective, a data point about the cost of talking to foreign media. The effect on information markets in China is precisely what you would predict: supply of honest communication shifts inward.

None of this produces a clean answer to the title question. Fabricating sources or data to expose state fabrication is plainly wrong, for reasons that go beyond deontology: it poisons the well of the independent verification that is the only thing distinguishing good journalism from propaganda in the first place. But the comfortable alternative — report only what can be confirmed, caveat everything else — is not neutral. It hands the information environment to the party controlling official releases. The economic and political stakes of getting China right are large enough that the methodology question is not academic. The debate over whether China is best understood as a socialist or capitalist economy is partly a data problem: what you conclude depends heavily on which numbers you trust. Benjamin Friedman’s point about growth and civic virtue, which I have found myself citing repeatedly — it came up in full earlier in connection with the moral consequences of economic development — applies to information regimes too: a society cannot deliberate well about the future it wants if it cannot agree on a description of its present. What Chicago economists said about the information problems in the Iraq reconstruction applies, at a different scale and with different politics, to every foreign economic beat where official statistics are an instrument of governance rather than a tool of analysis.