The irony is that Pruitt's proposed rule for Strengthening Transparency in Regulatory Science comes cloaked in the language of open data and replicability reform while developing a tool that will be quite effective in suppressing the use of any scientific results that the EPA Administrator finds inconvenient. I'd seen the various editorials raising the alarm about what seemed to be in the works, but it wasn't until last Tuesday that the proposal was finally posted on the EPA website. Yesterday (April 30) it was officially published. It's quite impressive. If one wasn't aware of the tangled history leading up to this proposal one might be fooled into thinking that it is indeed simply building on the trend towards open data policies that journals and funding agencies have been developing for several years now. The document repeatedly references open data policies at leading journals like Nature and Science and PLoS One and the Proceedings of the National Academy of Sciences as models for the kind of policy it's proposing. But where the clear intent and effect of those policies is indeed to make science more transparent, the intent and effect of the EPA's rule is to give the Administrator an efficient mechanism for ignoring scientific studies that won't provide support for the kind of regulatory rollback that Pruitt is clearly engaged in.
The rule requires that studies dealing with dose response data (i.e., studies that address the effects of particular levels of pollutants) can only be used as evidence for significant regulatory decisions if all of the data is made available in a fashion that facilitates independent review and analysis of the data. Open data advocates might be expected to cheer. Isn't this the kind of policy that so many of us have been working towards? But wait -- those journal policies are designed to be prospective and to encourage researchers to plan for data sharing as they develop their protocols. No one has suggested that work that has already been done is somehow less valid because it was published at a time when data sharing wasn't the norm. No one has suggested that we can't make use of solid peer reviewed literature in making decisions and developing policies. Not even the most radical advocate of open data and the need for replicability has advocated ignoring well established science because it hasn't been, or can't be independently verified. Not until now.
The privacy exemptions are exceptionally clever. Section 30.9 gives the EPA Administrator the authority to grant exemptions if he or she believes that it isn't "feasible" to make the data publicly available or to conduct an independent peer review. Which, of course, gives the Administrator the authority to refuse to grant such exemptions. It also lays out a clear roadmap for industry in developing research supporting the conclusions which industry prefers, the data for which the Administrator will have the authority to grant an exemption for.
The rule itself is quite short -- five columns in the Federal Register version, following the fourteen plus columns of the preamble, which lays out the justification for the rule and makes the case that it is simply building on the "scientific community's moves toward increased data sharing" (as the accompanying press release puts it).
I have to admire it from a creative writing standpoint. It's quite breathtaking in its brazenness. For years Senator Lamar Smith has been inveighing against EPA's "secret science," introducing legislation intended to achieve this result. Steve Milloy, tireless opponent of "junk science" has been its strongest advocate. Fortunately, the legislative approach has gone nowhere. But now it's about to be done, without needing to bother with the messiness of Congress. Thirty days for public comment is all that's required.
Open data advocates should be outraged. The scientific community is pushing back. But I have no confidence that it will be enough.