As the federal government embarks on a project to modernize electronic rulemaking, agencies are looking for effective ways to deal with mass comments. Some agencies have been hit with millions of nearly identical ones, and with fake comments coming from aliases or even from outside of the United States. As part of Federal Drive with Tom Temin‘s series on rulemaking modernization, we review some of the basics of notice-and-comment with the director of research at the Administrative Conference of the United States, Reeve Bull.
Insight by Carahsoft: This exclusive e-book demonstrates just how far agencies have come and where they still need to go to take fully advantage of DevSecOps to drive modern capabilities to their customers.
Tom Temin: Mr. Bull, good to have you on.
Reeve Bull: Wonderful to be on.
Tom Temin: Notice and comment is one of the basic processes for so many federal agencies and often not understood. Give us a sense of the genesis of it and why it’s important.
Reeve Bull: Notice and comment was really one of the major innovations of the Administrative Procedure Act, which was enacted in 1946 at the end of the New Deal period and prior to the Administrative Procedure Act, a lot of the work that agencies did was done through adjudicative processes where agencies would decide individual cases a lot like the courts do, and rulemaking was envisioned as being more of a legislative process where agencies could promulgate rules that apply to a wide variety of different activities and notice and comment was designed as a way for the agencies to basically gather information from the public. The idea was that the agencies don’t necessarily have all of the relevant information themselves. So through a notice and comment process, agencies could gather that first information throughout society. And the idea was that would help them make better rules that by using this information, using the data they received from the outside, they can improve their rulemaking process.
Tom Temin: And that’s pretty much is how it works today. What are the legal requirements for dealing with comments that agencies have?
Reeve Bull: Excellent question. The relevant provision of the APA is section 553 C, and its exact language is that the agencies have to consider to the relevant matter presented. And then earlier in this section of the APA, it refers to data, views or arguments. So basically the way the courts have interpreted this language, and there’s a large body of case law that’s emerged over the last several decades. But the basic idea is that agencies are legally required to consider any relevant information contained in the comments. So, for instance, let’s say the EPA is looking to enact a new rule that affects companies in a particular way. And the companies could submit information to the EPA and say this rule will affect this in the following ways. And then the EPA is legally required to consider that information if it’s relevant. And if it doesn’t do so, if if entity submits relevant. If your information that the agency doesn’t consider, then they confined lawsuit once, once the rule is promulgated, challenging the rule as arbitrary, and if the court agrees on the rule can actually be struck down on that basis.
Tom Temin: But can acknowledging a comment and saying yes, thanks for that information, but it’s gonna not affect what we decide is a final rule. Does that count as consideration?
Reeve Bull: Absolutely, absolutely. So the agency is not by any means required to take the course of action that’s being recommended by a company or by an NGO or a trade group or any other entity that’s submitting information. All they have to do is show that they considered it, that they took it into account and if they ultimately decided this isn’t relevant, or perhaps it is but we still think that the better course of action is what we originally proposing. As long as they give a reasoned explanation, then the court will uphold what the agency did.
Tom Temin: But the general principle is they have to respond, agencies?
Reeve Bull: Exactly.
Tom Temin: This idea of mass comments, it’s nothing new. I mean, people used to flood the mailboxes with postcards, but now it’s kind of accelerated in the electronic age. And what are some best practices to respond to, you think to, mass comments?
Reeve Bull: That’s an excellent question. And it’s one that, actually we at the administrative conference have done a little bit of work on. So back in 2011, we issued a series of recommendations, partly on the mass comment issue and from our perspective, the best practice is this. You know, you’re absolutely right agencies, you know, really, since probably the EPA was promulgated, but for a long time we’ve seen postcards, letters and often times they were so called mass comment campaigns where an entity, environmental group or some other NGO or trade group encouraged their members to submit comments to the agencies. And then people would submit virtually identical comments. And, of course, when the internet arose and e-rulemaking emerged, you know, early in the 21st century, the problem became magnified. Because, you know now it’s a whole lot easier to submit these comments. You just go online to regulations dot gov and submit a comment. And so the numbers have gone up, and it’s created some challenges for agency. So what we recommend it is that the agency’s if they get, you know, let’s say, 50,000 comments that are either identical or nearly identical. It’s not uncommon that a group will ask its members submit a comment and we’ll give them draft language, but will encourage them to, you know, change a few words here and there. So the comment maybe 95% identical, but not 100% identical. So what we recommend is that agencies can use software to actually identify when the comments are completely identical or nearly identical, and that they should certainly acknowledge that they’ve received these comments. They might even acknowledge the number of comments that were received as part of this process. But as far as the information that’s contained in the comment is concerned, they all contain the same information because they’re identical or nearly identical. So in that scenario, if it got 50,000 comments that say the same or nearly the same thing, they really only need to address it once. They simply need to acknowledge the information that’s in the comment and say, we got X number that said this, but they’re not required to consider it to a greater extent than they otherwise would be simply because they got 1000 comments as opposed to just one comment.
Tom Temin: Yeah, that’s really a key point. It’s not a voter or a plebiscite. And if it’s the same information 50,000 times in effect, they’ve only received the information once. You find that agencies are sometimes maybe lured into thinking, well, there’s a plebiscite here, maybe we shouldn’t do this? Which is not really their responsibility.
Reeve Bull: No, it’s a very interesting point. And, you know, you’ve seen that there is this temptation, and I think part of it is probably and maybe this is a good thing that you know democratic norms are so ingrained in our culture. We’re so used to thinking of our interactions with government is being voting, that I think there’s definitely a perception on the part of the public and probably to some extent on the part of you know, some agency officials as well that if they’re getting tens of thousands, hundreds of thousands of comments that say something and particularly if the comments point overwhelmingly in one direction or the other. If 99% of the comments favor the policy, or 99% of the comments oppose the policy, I think there is a tendency to think that that’s relevant. And the agency should, you know, take that into account. The problem was that from my perspective, and I think it’s both a legal issue and then just sort of a basic social science issue or policy issue, is that the comments are not necessarily representative of public opinion. When people submit a comment often times, the people who will take the trouble to do so are people who have strong views one way or the other. Even in those cases where the agency might get hundreds of thousands of comments and where 99% of them might say one thing or the other. That doesn’t necessarily mean that 99% of people in the public at large feel that way. The agency really wanted to figure out what public opinion was, They could use an opinion poll or perhaps create an advisory committee with the demographically representative group of citizens. There are a number of other way the agency, I think, more accurately can determine public opinion. If indeed, they want to take public opinion into account in a rule, rather than tabulating the number of comments.
Tom Temin: And with respect to fake comments, something else the GSA is trying to learn to deal with. And they’re coming in large numbers. Can a fake comment also have information that’s relevant even though it came from a source maybe outside the United States or under some alias?
Reeve Bull: From my perspective, yes. And there’s a tricky legal issue, perhaps here in that if you look at the language of the APA, it refers to a persons submitting comments. So a lot of these called fake comments are generated by a computer algorithms or so called bots. And in the case of a bot comment, one could ask, is that a person submitting it? Or is it just a computer submitting it? And in that case, is the agency required to take into account, And I think you could argue it both ways. I mean, a person actually created the algorithm, so in some ways it is, you know, a person submitting the comment. But then there’s also the separate question that you raised and this is true I think for both bot generated comments as well as fake comments where somebody may write a comment and then put a false identity at the bottom of the comment and submit the information of the agency. And I think that both of those cases it’s certainly theoretically possible that either the computer actually generates information that’s relevant to the agency, And I think, particularly as computer algorithms evolve and become more sophisticated, it becomes more and more likely that a computer could actually generate relevant information and then, similarly, with the fraudulent comments, the comment may be fake, but it may contain relevant information. So it’s an open question, but I think a strong argument could be made that the agency should still take that information into account. Of course, you get into additional challenges or tricky issues where some extent the identity of the commenter affects the amount of credibility or wait that it received. So if I submit a comment and claim that I’m a doctor and you know that I have expertise in this area and that’s not, in fact true, then you know any factual claims that I make are subject question, given that I’ve misrepresented who I am or what my expertise is. So I think it goes to the credibility of the comment or the weight that the agency should give it. But if in fact, it is relevant information and the only thing that’s inaccurate is the name of the bottom of the comment, then I think, you know, arguably, the agency should still go ahead and take that information into account.
Tom Temin: So, in other words, as they review that e-rule making process, there’s really no cut and dry answers on any of this, is there?
Reeve Bull: That’s correct. That’s correct. We’re sort of in a brave new world here to some extent, and I think it’s only likely that that more and more challenging as we proceed, particularly as computer algorithms become more and more sophisticated. We did a program here at the administrative conference about a year and a half ago on this very issue and at that time, the perspective of a lot of the commenters was that this really isn’t that big of a problem. You know, if the agency gets a bot comment or a fraudulent comment, it’s pretty easy to identify that it’s illegitimate. And the agency can discount it accordingly. But it sounds like in the last year and and a half even the technologies have become a lot more sophisticated. There was a recently a research paper written on so called deep fake comments, where basically a person designed an algorithm to write a comments that were virtually impossible to distinguish from legitimate comments. So I think this is only likely to become more and more important as time proceeds in there. A lot of challenging
Tom Temin: Reeve Bull is research director at the Administrative Conference of the United States. Thanks so much for joining me.
Reeve Bull: Absolutely thank you.