Lack of information is a continuing problem for environmental policy. In part, that’s unavoidable; we’ll never know enough about the world around us to be confident that we’re making the best choices. In part it is because potential regulatory targets control some needed information. And in significant part it’s because decisionmakers have a tendency to close their eyes to avoid confronting inconvenient facts (like the vole in the picture above, who doesn’t see the attacking owl).
Law can help fill information gaps by providing mandates and incentives for private parties to reveal information and for regulators to face the truth. Courts have a role to play, but it’s a delicate one. They can rigorously enforce legislative directives that agencies collect and consider information. But when it comes to deciding whether an agency has learned enough to justify its decisions, there’s a fine line between demanding too much and too little. Courts should not paralyze agencies by demanding an impossible level of knowledge. On the other hand, they should not encourage ignorance by automatically deferring to agency decisions any time the facts are uncertain.
The Ninth Circuit has struggled to find the middle point in recent years in a series of disputes over national forest management. The most recent entry in the series, Greater Yellowstone Coalition v. Lewis, goes too far in the direction of encouraging agency ignorance.
The dispute was about expansion of a JR Simplot phosphate mine on federal land in the Caribou National Forest. Because of selenium pollution it has generated, the existing mine is already a Superfund site. The Bureau of Land Management and Forest Service approved expansion. Greater Yellowstone Coalition argued that the agencies had failed to take a sufficiently hard look at the consequences of expansion under NEPA and had arbitrarily and capriciously concluded that the expansion would not exacerbate water quality violations, a result forbidden by the Clean Water Act and National Forest Management Act.
It’s difficult to evaluate these sorts of highly technical cases without knowing the underlying record. But just on the basis of the information in the opinions, the outcome doesn’t make sense. The majority decided that it wasn’t necessary to understand all the sources of selenium pollution at the existing site, because reductions at two sites would offset any pollution from expanded operations. That would make sense if it were clear how much pollution the new operations would produce and how successful remediation efforts would be. But that wasn’t clear. The leader of the Forest Service’s national groundwater program testified that the modeling used to evaluate the likely performance of a proposed cover design did not adequately take into account seasonal variations in water flows, and that improved modeling could be done with 4 to 10 days more work.
Judge Sidney Thomas, writing for the majority, characterized this as a case of disagreement among experts in which the agency decision must be upheld as long as all views had been considered and a rational basis offered for the choice. That’s an absolutely conventional choice, but one that effectively privileges agency ignorance. Senior Judge Betty Fletcher, who has been a leading voice for requiring agencies to learn when they can, offered a better approach in her dissent. She would not demand that agencies gather every scrap of potentially available information, but would insist that they at least explain why it is too costly to obtain information relevant to the decision.
CEQ’s NEPA guideline explicitly require as much. 40 C.F.R. 1502.22 deals with uncertainty. It requires that agencies obtain information which is “essential to a reasoned choice among alternatives” if the costs of doing so are “not exorbitant.” To satisfy that regulation, agencies should have to explain why they don’t think missing information is needed or obtainable. If needed information can’t be obtained, the agency must evaluate the range of possibilities.
The CEQ guideline also points the way to a more nuanced approach to determining whether lack of technical information makes a decision arbitrary or capricious under the APA. Where missing information complicates reasoned decisionmaking, courts should require that the agency directly address how that information would make a difference to the decision; what it would, in time and resources, to get it; and to what extent it would also benefit other or future decisions. If the agency makes a convincing case that the costs of getting the information outweigh its benefits, it should be allowed to proceed based on a reasoned evaluation of the available evidence. But if it cannot make that showing, it should be required to fill the information gap before proceeding.
The Greater Yellowstone majority gets it wrong on another recurring point. They allow expansion to go ahead based on an adaptive management approach, based on inclusion in the permit of a condition requiring future testing of the effectiveness of the cover once installed. The problem is that the court simply assumes that the agencies can and will require corrective action if testing shows the cover isn’t working as anticipated. That’s easier said than done. Problems may not be detectable until operations are complete, they may not be correctable, and if the agency fails to take enforcement action it may not be possible for citizens to do so. A closer look is needed before courts simply allow learning to be put off into the future.