In late 2013, Ma and his colleagues were eager to start leveraging real-time point-of-emissions data in China to prompt regulatory and civic action. IPE was developing a web-based platform on which it would re-publish the provincial EPB data streams. Having the data in one place would make it easier to compare the quality of information released by local governments, see where the most polluting enterprises were concentrated, and identify cities and provinces where enterprises were routinely over the limit. Moreover, by collecting and storing the data over time, IPE would enable researchers to conduct longitudinal studies. For example, it would become easier to find correlations between cancer rates and levels of chemicals in a given location.

IPE had plenty of experience building databases and online maps. But the task at hand was more complicated than anything it had done before, and difficulties arose. For one thing, the real-time data came day and night from over 30 EPBs; it was a technical challenge to manage the constant streams of data. Each EPB had its own publishing format, and some were churning out more information than others. A small technical team at IPE looked into ways to make the data consistent for the IPE site, and to translate the information into a user-friendly smart phone app.

Those were technical challenges; the political challenges were even trickier. The real-time data came from state-owned companies with a long history of being protected by both local and central officials. Ministries that regulated power generation, mining, manufacturing and China’s other huge industries were, in general, far more powerful than the MEP, and guarded their interests. Furthermore, SOE bosses were not susceptible to consumer pressure, like their counterparts at MNCs. They already resented having to report what they considered proprietary information. They would not be pleased when IPE made the data even more accessible to the public and media through its website and app.

Antagonizing an SOE or its government backers could yield negative consequences for an NGO, even when the law was on its side (and rules always held ambiguities, such as the “commercial secrets” exception in OGI). An NGO might have its registration revoked, its operations disrupted or its people harassed. If IPE published information showing that an SOE was grossly exceeding allowable emissions limits, the SOE might lodge a legal complaint, accusing IPE of spreading false reports and defaming the company. Even if the SOE’s case were flimsy, IPE’s resources could easily be consumed in preparing its defense, putting its other programs at risk. Its partner NGOs—most small, local operations—would be at even greater risk.

In the past, IPE had found it critical to keep screenshots of environmental citation records. When a “blacklisted” Chinese supplier questioned IPE’s information, the researchers could retrieve a screenshot as evidence. This was especially important when public records disappeared or were altered without explanation. But with real-time data flowing from 15,000 enterprises to 33 EPBs every hour and every day of the year, it would be impossible for IPE’s staff of 16 to take a screenshot each time an enterprise exceeded emissions limits. “Some of the companies break the rules almost every day,” says Ma. IPE could not capture every incident.

Making matters worse, SOEs might be correct, at least occasionally, in disputing the accuracy of the data. As they started collecting emissions data from the EPB websites to assess quality and develop the online tool, researchers noticed an SOE’s emissions would sometimes spike suddenly—well above legal limits—then revert to normal. Did this mean it had dumped illegally? Other times emissions would zero out. Had the factory halted operations briefly? Tampering was a possibility. Benign explanations were also feasible: a malfunction in the point-of-emissions monitoring equipment, or a glitch in the transfer of data from the SOE to the EPB, or from the EPB to IPE. By the time the researchers caught a blip in the data stream, it was often too late to double check, because the EBPs refreshed the figures hourly and many did not provide an archive of historical data. IPE recognized that errors were likely, particularly in the early days of monitoring.

Yet Ma felt that flawed data was better than no data, and that publishing the information would force all parties to concentrate on making it more accurate. He cites the example of urban air quality data. Before the government started real-time monitoring of PM2.5, people had only the US Embassy data, and therefore focused on the high levels of pollution in the capital city. Once monitoring was implemented across the country, however, it became apparent that many other cities were even worse off. “Beijing usually cannot get into the top 10 most polluting cities in China,” notes Ma. He continues:

Before, Beijing was under more scrutiny. Now, the real-time disclosure has leveled the playing field. I think on the monitoring data of major polluters, we need to make sure that it’s also under the same public scrutiny. I think it has a huge potential to help improve the quality of the monitoring data, which has long been the problem, actually.

Still, without screenshot evidence, and given that inaccuracies were likely, IPE expected SOEs to contest IPE’s information, and demand that unflattering numbers be scrubbed from its website. The researchers considered how to respond. Should IPE remove contested data? Or issue a caveat along with the data?

Even if erroneous numbers were caught, it would be nearly impossible to completely retract them. IPE’s app, when developed, would send hourly data directly to people’s smart phones. Based on the treatment of US Embassy readings, it was safe to assume that exceptionally high emission figures would be re-tweeted far and wide. IPE would not be able to retract numbers once they entered the ether.One option was to lower the bar for revising information on the IPE website. Previously, with the environmental citations, companies had to go through a third-party audit to have their records on IPE’s site changed. With the real-time data, the requirements could be simpler. For example, if an SOE were to get a document from an EBP stating that the original data were faulty, then IPE would change the information on its own site. But how could IPE be sure that an EBP was correcting the record only in cases where a technical error had occurred, and not due to political pressure?

While they considered how best to develop the online tool, Ma and his colleagues also prepared for the release of a new IPE report, Blue Sky Roadmap II: Real-Time Disclosure Begins , slated for early January. The report assessed the progress EPBs had made in implementing the real-time reporting rules. It lauded Shandong Province, for example, for establishing a user-friendly platform that color-coded data to indicate whether an SOE’s hourly emissions were within (blue) or above (red) the limit—and if above, by how many times. The report also showed that pollution was really a regional, not city- or province-level, problem. Beijing’s air remained smoggy, despite the exodus of heavy industry, because its neighbors were polluting. Says Gu Beibei:

In our report, we actually made a quick, simple analysis by taking the top eight emitters in three provinces: Hebei, Shandong and Beijing. And we found the gap is huge. In Hebei and Shandong, the top eight release 30 times and 37 times more than the top eight in Beijing.

But IPE also wanted to demonstrate to the public and media how to use the data: to reveal the misdeeds of individual enterprises. As examples, the report named specific SOEs that exceeded pollution emission standards, using the real-time data as evidence.

The Roadmap II report raised some of the same concerns as re-publishing the real-time emissions data. Would the SOEs cited in the report dispute the accuracy of the data? It so, how should IPE respond? Should IPE delay publication, in order to first contact each SOE mentioned in the report to get its side of the story? Should IPE postpone re-publishing any emissions data until kinks in the system were worked out? As the publication date drew near, the researchers considered the options. Clearly, there was a risk in naming SOEs—a risk that Ma and his colleagues wanted to minimize. On the other hand, the real-time data rules rewarded years of methodical effort on the part of IPE to force transparency and accountability in China. Ma and his team did not want to lessen the pressure now.