Debating Cass Sunstein on Transparency

We need to find a language that allows us to differentiate honest, disinterested debate over policy options from special-interest lobbying and campaign-finance blackmail.


Professor Cass Sunstein is an eminent legal academic and former Obama appointee with a pretty amazing track record in writing about the right ideas at the right time.

I won’t list all his achievements here because that’s what Wikipedia is for. However, I will note that he warned the world about the Filter Bubble about a decade before it became a topic of widespread concern, and that he’s one of the key people who made “nudging” the hippest of hip public policy topics.

The reason that I’m drawing Civicist readers’ attention to Professor Sunstein today is that a couple of weeks ago he slipped out a paper that fills a vacuum in the transparency debate that’s been bugging me for several years. It might have been bugging some of you, too.

The vacuum relates to the weak language and ideas we have to describe the kind of information that governments should definitely make public, versus the kind that should legitimately be kept secret, at least temporarily.

The paper is entitled Output Transparency vs. Input Transparency and is marked with the refreshingly confessional warning “very preliminary draft.” I read this as an invitation to provide feedback, and if that’s not what Professor Sunstein wants then I can only encourage him to deploy less inviting caveats.

In what way does Sunstein fill a vacuum?

Most people who come to the field of transparency, myself included, don’t come at it from a totally neutral, disinterested perspective. Some parties, like anti-corruption activists, feel with certainty that more transparency is needed to uncover wrongdoing. Opposing them are others, like spymasters, who know with equal certainty that too much transparency will be a disaster for them and the countries they serve. Only a few academics who are interested in the effectiveness of transparency could be said to really approach the topic from a disinterested perspective.

Consequently, the debate over how much transparency we should have often comes down to fights over “more” or “less,” especially when individual policy reforms are on the table. From a politicking perspective this makes sense—the standard rules of activism suggest that if you want to win any fight that relies on public opinion you have to simplify and exaggerate. You paint your side as totally virtuous and correct, and your opponents as horned-and-hooved enemies of all that is good and right. To talk about compromises or the need for balance is simply to hasten the victory of the Enemy.

This polarization is unhelpful in any policy area, but it is especially problematic in the transparency arena, since it is pretty obvious that governments are always going to hold a mix of information that should definitely be made totally public, such as weather forecasts, and information that definitely shouldn’t be totally public, such as the locations of shelters for victims of domestic abuse. The tricky thing is where the line should sit between these two extremes.

What is so refreshing about Sunstein’s contribution, therefore, is simply that he has tried to write about where an ideal point of balance might lie, rather than making an overall case that we need more or less transparency. He’s interested in where the choices to hide or disclose get tricky, rather than where they are easy.

Sunstein’s Argument

The essence of Sunstein’s argument is that most government information can be categorized as belonging to one of two categories: inputs or outputs.

Outputs are created when a government performs some task, and as a consequence some information is generated—for example, measuring the pollution in a river, or recording the levels of crime in a particular city. Inputs, on the other hand, are types of information that are used to help make a government decision. These include, for example, emails from one agency to another, giving their views on a new law or regulation.

The heart of Sunstein’s argument is simple:

“There is good reason for a large increase in output transparency—and for caution about input transparency.”

His reasons for backing output transparency are pretty much the same that have been voiced by transparency advocates for a long time: better private and public services, and more competent and accountable government. I’m not going to go into much detail here since I’m guessing that most readers will be quite familiar with these arguments already.

His reasons to be cautious on input transparency are more interesting, especially given his time as a senior public servant in the Obama administration.

First up, he argues against disclosure of internal decision-making documents based on the idea that the public simply wouldn’t be interested in much of what is generated as inputs, and therefore the effort involved with disclosure simply isn’t worth it. In his words, “the sheer number and range of inputs is daunting, and it defies belief to think that the public would benefit from seeing all of them.” This is fundamentally a cost-benefit argument, based on a view that the costs most likely outweigh the benefits of implementation.

I find this argument by far the weakest he makes in his paper, and primarily of interest because it’s a salient reminder about how people who grew up in the pre-digital world carry with them very different assumptions about the cost of information storage, categorization, and retrieval, compared with digital natives. It’s not really worth getting too worried about, though, because his following arguments are much stronger.

His next argument is that rules that require disclosure are already pushing more and more information into telephone calls and face to face meetings, as people seek methods of conversing secretly. He argues that this is bad for decision-making because conveying all information in verbal form “impose[s] losses, in the form of diminished reliance on careful economic, legal, and other analyses.” He’s definitely right that this has already happened, and that a lot of people no longer convey sensitive ideas with the written word. However, I can’t help but think that Sunstein might be slightly defeatist about the power of legislation here. Not many public servants would have friendly off-the-record chats if they thought the FBI might send them to federal prison for violating disclosure requirements about the content of meetings. They don’t currently think this, and so phone calls flow freely.

Third, and where I empathize the most, is his worry about the chilling effect (my term, not his) of transparency: the pressure that people inside government feel to not express how they really feel if they believe that what they say will be disclosed and will come back to haunt them later on. In Sunstein’s words, “input transparency can lead people not to say what they think. It can reduce candor and the free play of ideas. In that sense, it can ensure that groups will have less information than they need.”

I know that some transparency advocates think the chilling effect is an overstated problem, but I’m a long-time transparency advocate and there have been times when I have felt this pressure acutely. In particular, I have experienced various moments when advising civil servants and ministers where it was very important to communicate that Person X was a Real Problem. However, the fear that Person X would find out and then make my life a misery was tangible, and I self-censored my advice as a result. I can only imagine how strong that pressure must become if the thing you want to criticize isn’t a slightly inept public servant but a homicidal foreign power.

Sunstein for his part doesn’t exactly need my modest experiences to back up his argument. He cites James Madison, who reportedly claimed, “no Constitution would ever have been adopted by the convention if the debates had been public.”

Where Next?

Towards the end of the paper, Sunstein makes it clear that he has not written a policy paper to be implemented in detail, but a think piece to kick off a debate. He writes, “My aim has not been to reach a definitive conclusion about concrete practices and proposals, but to outline general concerns to help identify the appropriate tradeoffs.”

I think this has never been more needed. The whole transparency debate has been very tactical in recent years, fighting one street at a time over individual policy changes and shifts of position and emphasis. There is a need for some more general principles that can inform the battle plan ahead.

However, as much as I respect him for kicking this off, I don’t feel that Sunstein has got the balance quite right, drawing the line between “inputs” and “outputs.” This is because he seems to give too little weight to what appears, at least from these foreign shores, to be the U.S.’s greatest current governing flaw.

This flaw is that the U.S. governance system seems unusually poor at balancing the needs of special interests against the wider public interest. So, for example, several national governments effectively don’t require people with simple tax arrangements to calculate their own tax returns, but lobbying ensures that Americans don’t get to save time and money like this. Other countries manage to build highly popular train lines to support their expanding populations, but America can’t quite manage to drive them through. I’m sure that my readers could fill in many more examples without even troubling Google search. Put together, this means that setting a transparency norm that assumes that basically all lobbying input has a right to remain secret seems like a curious choice.

So I agree that a debate over the boundaries between public and private is very much needed, and I agree that some of the downsides of transparency that Sunstein mentions are real. But the general input/output typology doesn’t quite seem to get at the heart of the difference we need to find. We need to find a language that allows us to differentiate honest, disinterested debate over policy options from special-interest lobbying and campaign-finance blackmail. The former clearly deserves some protection, the latter not so much. We also need some better thinking and better words to describe the nature of harm that justifies secrecy classifications, versus the kind of harm that is purely political in nature.

The civic tech community has a lively side-interest in transparency, and a good tech-driven lens to help understand what is possible and impossible today, and in the near future. I hope that it will come together to debate with Professor Sunstein, and hopefully cobble together a new consensus about where the balance between government secrecy and transparency should lie in the information age.