Cass R. Sunstein
It was … best for the convention for forming the Constitution to sit with closed doors, because opinions were so various and at first so crude that it was necessary they should be long debated before any uniform system of opinion could be formed. Meantime the minds of the members were changing, and much was to be gained by a yielding and accommodating spirit. Had the members committed themselves publicly at first, they would have afterwards supposed consistency required them to maintain their ground, whereas by secret discussion no man felt himself obliged to retain his opinions any longer than he was satisfied of their propriety and truth, and was open to the force of argument…. No Constitution would ever have been adopted by the convention if the debates had been public.
—JAMES MADISON
OUTPUTS AND INPUTS
There is a distinction between output transparency and input transparency. Suppose, for example, the Department of Transportation completes a detailed study of the kinds of policies that help reduce deaths on the highways; the Department of Labor produces an analysis of the health risks associated with exposure to silica in the workplace; or the Environmental Protection Agency produces a regulation to curtail greenhouse gas emissions from motor vehicles or adopts a policy about when it will bring enforcement actions against those who violate its water quality regulations. All of these are outputs. The government might also become aware of certain facts—for example, the level of inflation in European nations, the number of people who have died in federal prisons, the apparent plans of terrorist organizations, or levels of crime and air pollution in Los Angeles and Chicago. For the most part, facts should also be seen as outputs, at least if they are a product of some kind of process of information acquisition.
Now suppose that officials within the Department of Energy and the Environmental Protection Agency exchange views about what form a greenhouse gas regulation should take, or political appointees within the Department of Labor have heated debates about the risks associated with silica in the workplace and how best to handle those risks. These various views are inputs.
To be sure, there are intermediate cases. The EPA might conclude that a substance is carcinogenic, and in a sense that conclusion is an output, but it might also be an input into a subsequent regulatory judgment. The Department of Transportation might reach certain conclusions about the environmental effects of allowing a highway to be built, which seem to be an output, but those conclusions might be an input into deciding whether to allow the highway to be built. The National Environmental Policy Act can be seen as a requirement that agencies disclose outputs in the form of judgments about environmental effects, but those outputs are, by law, mere inputs into ultimate decisions about what to do. Some outputs are inputs, and in the abstract, it would be possible to characterize them as one or the other, or as both. As we shall see, the appropriate characterization depends in part on whether and how the public would benefit from disclosure.
Acknowledging the existence of hard intermediate cases, I offer two claims here. The first is that the argument on behalf of transparency is often exceptionally strong for outputs. The second is that the argument on behalf of transparency for inputs is qualitatively different and generally weaker.
If the government has information about levels of crime in Boise, about water quality in Flint, Michigan, about security lines at LaGuardia Airport, about the hazards associated with certain toys, or about the effects of driverless cars, it should usually disclose that information—certainly on request, and if people stand to gain from it, even without request. (The latter point is especially important.) In all of these cases, the benefits of transparency are significant. Sometimes members of the public can use the information in their daily lives, and such disclosures enable various forms of accountability. Most of the time, moreover, the costs of output transparency are trivial. The U.S. government should therefore offer much more in the way of output transparency. In particular, it should make outputs freely available to the public as a matter of course—at least if the public could or would benefit from them, and unless there is a particular reason that outputs need to remain confidential.
As James Madison’s remarks on the Constitutional Convention make clear, input transparency is a much more complicated matter because the costs of disclosure are often high and the benefits may be low. In any case, the arguments for input transparency are qualitatively different from those that justify output transparency. There are strong reasons to protect processes of internal deliberation, above all to ensure openness, candor, and trust. In addition, it is often unclear that the public would gain much from seeing inputs, not least because of their massive volume (and usual irrelevance to anything that matters). Outside of unusual circumstances, the public would gain little or nothing, except perhaps something like gossip. Another way to put the point is that those who seek to attract eyeballs or to embarrass their political opponents often like input transparency, but the public usually does not much benefit from it.
To be sure, transparency about inputs can be informative, and certain inputs may be of keen historical interest. If the public learns that the deputy secretary of Transportation had a different view from that of the secretary on the content of a fuel economy regulation, it knows something; internal disagreement paints a different picture from internal unanimity. But how much, exactly, does the public learn, and why is it important for the public to learn it? It should be acknowledged that in some cases input transparency is a good idea, especially under circumstances of corruption (or something like it) and when relevant inputs have genuine historic importance (and when their disclosure can reduce mistakes). But the argument for input transparency is much different from the argument for output transparency, and it often stands on weaker ground.
It should be clear from these remarks that my approach to this topic is insistently and unabashedly welfarist: what are the benefits of transparency and what are the costs? It is true that the benefits and the costs may not be easy to quantify, but some kind of assessment of both is indispensable to an evaluation of when transparency is most and least necessary. For those who are not comfortable with talk of costs and benefits in this context, it might be useful to understand those terms not as some kind of arithmetical straitjacket but as signaling the importance of asking concrete questions about the human consequences of competing approaches. At least for difficult problems, those questions are far more productive than abstractions about “legitimacy” and “the right to know.”
A clarification before we begin: I am speaking here about principle, not about the appropriate interpretation of the Freedom of Information Act or about possible amendments to the statute. One of the virtues of the developing case law, and of the most illuminating debates over amendment, is that they tend to be particularistic: they involve situations that are both specific and highly diverse, complicating broad pronouncements. Here, as elsewhere, general propositions do not decide concrete questions. It is easy to find examples that confound my categories, but categories can provide useful orientation, or at least that is my hope.
OUTPUT TRANSPARENCY
OF USABLE INFORMATION AND SUNLIGHT
1. An instructive finding. A remarkable finding by the economist Amartya Sen is that in the history of the world there has never been a famine in a system with a democratic press and free elections.1 Sen’s starting point here, which he demonstrates empirically, is that famines are a social product, not an inevitable product of scarcity of food. Whether there will be a famine, as opposed to a mere shortage, depends on people’s “entitlements”: that is, what they are able to obtain. Even when food is limited, entitlements can be allocated in such a way as to ensure that no one will starve.
But when will a government take the necessary steps to prevent starvation? The answer depends on that government’s own incentives. When there is a democratic system with free speech and a free press, the government faces a great deal of pressure to ensure that people generally have access to food. And when officials are thus pressured, they respond. But a system without a democratic press or free elections is likely to enable government to escape public accountability and hence not to respond to famines. Government officials will not be exposed, nor will they be at risk of losing their jobs.
Here, then, is a large lesson about the relationship between a well-functioning system of free expression, disclosure of relevant information (outputs), and citizens’ well-being. Free speech and freedom of information are not mere luxuries or tastes of members of the most educated classes. On the contrary, they increase the likelihood that government will actually serve people’s interests.2 This lesson suggests some of the virtues, not only for liberty but also for economic goals, of having freedom of speech and freedom of information.
2. Obama, mostly—and navigability. In recent years, the most prominent transparency initiatives have involved outputs. A revealing example involves the global positioning system (GPS). In 1993, President Clinton unlocked the data that was ultimately used to make the GPS device a familiar part of everyday life. Its availability has helped countless people, often in profound ways; it has even saved lives. A GPS device makes life more navigable (literally). If we think about navigability as a more general idea, we can see the value of disclosure of many outputs. Information about safety seats in cars, crime rates, air and water quality, and much more can be seen as akin to GPS devices writ large: it tells people how to go in the directions they want to go.
For all of its years, the Obama administration made transparency a major priority. (I am insisting on that point while fully acknowledging, and bracketing, the many controversies during the Obama presidency over potential trade-offs between transparency and other values.) The priority was signaled by an early and defining presidential memorandum, dedicated specifically to the Freedom of Information Act. The memorandum establishes “a clear presumption: In the face of doubt, openness prevails.” It adds that “agencies should take affirmative steps to make information public. They should not wait for specific requests from the public.” It directs both the attorney general and the director of the Office of Management and Budget (OMB) to issue new guidance, designed to implement the governing principles.3
Both of the resulting documents deserve close attention, but for my purposes here, OMB’s guidance is especially noteworthy. The memorandum directs agencies to publish information online. It adds that “agencies should proactively use modern technology to disseminate useful information, rather than waiting for specific requests under FOIA.” Perhaps most significantly, it requires each agency to create an open government plan and an open government webpage, designed to “create and institutionalize a culture of open government.” The open government plans are required to post “online in an open format at least three high-value data sets,” which are in turn defined as “information that can be used to increase agency accountability and responsiveness; improve public knowledge of the agency and its operations; further the core mission of the agency; create economic opportunity; or respond to need and demand as identified through public consultation.”4
In the abstract, it is not clear whether this initiative involves output transparency or input transparency, but in practice, the former has been primary by far.5 The high-value data sets typically involve outputs. Since 2009, data.gov has become a principal location for posting such data sets, which amount to output transparency in action. At some point in the recent past, the site contained more than 190,000 data sets, with information on agriculture, finance, health, education, energy, and much more. With a click, you could find Airline On-Time Performance and Causes of Flight Delays: On-Time Data; Expenditures on Children by Families (with estimates of the cost of raising children from birth through age seventeen for major budgetary components); and detailed information about product recalls. There has been much more in the same vein, focusing on the outputs of policy-making or information-gathering activity.
As a result, private sector actors have produced numerous apps that provide people with information that they can actually use. One example is AIRNow, which has up-to-the-moment information about air quality. Another is the College Affordability and Transparency Center, which provides information about college costs. Yet another is eRecall, which gives people information about recall information at the time of purchase.
The outputs released on data.gov serve two independent purposes. First, people can take advantage of them in their daily lives. Like a GPS device, most of the information makes life simpler and more navigable. The availability of this information on cell phones makes the point far from fanciful. This is no mere abstraction. If we take the idea of navigability in the large, we can see disclosure as a way of helping people get to their preferred destinations in countless domains, saving money and reducing risks in the process. To my knowledge, the benefits of data.gov have yet to be quantified, but there is little doubt that people are gaining from the disclosures in concrete ways—again, as with the benefits of GPS devices.
Second, release of the outputs can promote accountability in both the private and public sectors. Justice Louis Brandeis famously said that “sunlight is … the best of disinfectants.” If the air quality is terrible in Los Angeles, if a particular university is unusually expensive, if students at a for-profit college do not end up with jobs, if drinking water is unsafe in San Diego, or if a company has a lot of recalled toys, transparency can serve as a spur to change. Transparency increases accountability, and when people are accountable, their performance is likely to improve.6 The point bears on both public and private institutions. Transparency can tell citizens about the actions of public officials—for example, how long it takes for them to work on a permit application, or the levels of air pollution in San Antonio (for which officials bear some responsibility). It also informs citizens about the actions of private actors—for example, by disclosing product recalls or ratings of safety seats. In either event, it can spur improved performance.
3. Policies: disclosure and “core missions.” One of the most interesting aspects of the OMB memorandum is that it asks agencies to consider whether disclosure might further their “core missions.” This is an exceedingly important idea that deserves far more agency use in the future, and it involves disclosure of outputs.
Consider just a few illustrations. In environmental policy, one of the most well-known examples is the Toxic Release Inventory (TRI), which was created largely as a bookkeeping measure designed to ensure that the federal government would have information about toxic releases. To the surprise of many people, the TRI has been a successful regulatory approach because companies did not want to be listed as one of the “dirty dozen” in their states.7 Accountability served as a spur toward emissions reductions. During the Obama administration, the Occupational Safety and Health Administration followed this lead by putting information about recent deaths in American workplaces very visibly on osha.gov, with names of the companies where people died.8 The EPA did something quite similar with its Greenhouse Gas Inventory, one of the goals of which was to spur emissions reductions.9
In all of these cases, the government is disclosing information that public officials have. We can imagine, of course, a requirement of output transparency imposed by the public sector on the private sector. Requirements of that kind are not always organized under the idea of freedom of information, but they involve transparency and also can promote important agency missions. Under the authority of the Affordable Care Act, for example, the Food and Drug Administration has required chain restaurants to disclose the calories associated with their offerings. Though the evidence is mixed, some studies find that the early results are quite promising, with significant reductions in body mass index among people who really do need to lose weight.10
I have offered just a few illustrations of disclosures intended to promote agency missions through output transparency. An excellent collection, generally including outputs, can be found in the numerous action plans of the Open Government Partnership, coming from dozens of nations. It is, of course, an empirical question whether transparency will promote agency missions, but in many cases it can.11 (It is said that China’s interest in air pollution and greenhouse gas emissions has been greatly spurred by the ready available of the Air Quality Index on cell phones.) Because the costs of output transparency are typically low, there is every reason to adopt a presumption in its favor.
4. Costs and benefits, in public. We should understand regulatory impact analyses (RIAs) in this light. In the relevant respect, they are outputs, although they count as inputs as well. Required by presidents from Ronald Reagan12 to Barack Obama,13 and thus far unchanged by Donald Trump, these analyses offer accounts of the expected effects of regulation, with careful attention to both costs and benefits. If a regulation would prevent two premature deaths per year, the agency must say so, and so too if it would prevent five hundred. The RIA must disclose whether the regulation would cost $25 million, $250 million, or $2.5 billion. As part of rulemaking, it must be provided to the public for scrutiny and review, accompanying proposed and final rules.14 In its own way, the requirement of an RIA can be seen as a kind of Freedom of Information Act. It enlists sunlight as a disinfectant.
A central reason is that, by itself, cost-benefit analysis is an important safeguard against ill-considered regulations. One of its key features is that it promotes transparency about actions and alternatives and indeed about the contents of cost-benefit analyses themselves. Recent administrations have been entirely aware of this point. To promote public understanding and to ensure an “open exchange of information and perspectives,” for example, regulatory preambles for lengthy or complex rules (both proposed and final) are required to include straightforward executive summaries. These summaries must describe major provisions and policy choices.15 For one illustration of such a summary, consider table 9.1.
TABLE 9.1 Benefits and Costs of EPA’s Proposed Clean Power Plan Rule in 2030 (Midpoint Estimates, Billions of Dollars)
| |
CLIMATE CHANGE IMPACTS |
|
HEALTH IMPACTS (CO-BENEFITS) OF CORRELATED POLLUTANTS PLUS… |
| |
DOMESTIC |
GLOBAL |
|
DOMESTIC CLIMATE IMPACTS |
GLOBAL CLIMATE IMPACTS |
| Benefits |
|
|
|
|
|
| Climate Change |
$ 3 |
$ 31 |
|
$3 |
$31 |
| Health Co-Benefits |
|
|
|
$45 |
$45 |
| Total Benefits |
$ 3 |
$ 31 |
|
$48 |
$76 |
| Total Compliance Costs |
$ 9 |
$ 9 |
|
$ 9 |
$ 9 |
| Net Benefits (Benefits − Costs) |
−$ 6 |
$ 22 |
|
$ 39 |
$ 67 |
To be sure, a great deal must be said in order to make a table of this kind fully transparent. It is important to know what these numbers actually mean and how they are derived.16 For example, the claim that health co-benefits are $45 billion depends on assumptions about the effects of the plan on emissions and also the effects of emissions reductions on human health. Such assumptions might be controversial. A competent RIA is transparent about those matters. If there are uncertainties and reasonable disputes, it will reveal them and promote accountability in that way as well.
COSTS AND BENEFITS OF OUTPUT TRANSPARENCY
I have been painting with a very broad brush—in principle, an unduly broad one. My suggestion has been that disclosure of outputs is justified, or presumptively justified, on welfare grounds, but this is not always the case. We can easily imagine outputs whose disclosure would produce low benefits or high costs. With respect to costs, consider the words of the OMB memorandum:
Nothing in this Directive shall be construed to supersede existing requirements for review and clearance of pre-decisional information by the Director of the Office of Management and Budget relating to legislative, budgetary, administrative, and regulatory materials. Moreover, nothing in this Directive shall be construed to suggest that the presumption of openness precludes the legitimate protection of information whose release would threaten national security, invade personal privacy, breach confidentiality, or damage other genuinely compelling interests.17
In various ways, the Freedom of Information Act recognizes all of these points. No one doubts that the government has a great deal of information whose disclosure would endanger national security, and even if that information can be counted as an output, it should be kept confidential. The government also has “personally identifiable information,” which receives protection under privacy laws. Although a balance must be struck between transparency and privacy, some forms of disclosure impose reduced privacy, often in an intolerable way. Some kinds of disclosure could compromise trade secrets or otherwise privileged information. If disclosure is not automatic or automated, the very act of transparency can impose costs in terms of both money and time.
On the benefit side, distinctions are also important. In principle, and if the costs of assessment were zero, it would not make sense to insist that each and every output should be disclosed. It would be better to ask, on a case-by-case basis, whether disclosing specified outputs would or could be beneficial—for example, to consumers and workers. Of the 190,000 data sets on data.gov, surely some have modest benefits or no benefits; people are not paying the slightest attention to them (and they will not in the future). A welfarist analysis would call for particularized inquiries into that question. The problem, of course, is that those inquiries may not be manageable. At the time when disclosure is being discussed, projection of benefits may be quite difficult. What people will do (if anything) with information may not be self-evident. The private sector is ingenious and full of alchemists. What it will find useful, or turn into gold, cannot be predicted in advance.
In view of that fact, it makes sense for agencies to make reasonable judgments about “high-value data sets,” broadly understood, and to get them online as soon as possible—and also to announce a general presumption in favor of disclosure of outputs, armed with an intuitive understanding of the domain to which the presumption will be applied. It should be underlined that a degree of automaticity, putting relevant material online as a matter of routine, could be extremely helpful.
With respect to high-value data sets, intuitions should be disciplined by asking two questions: (1) Could people possibly benefit from this information in their daily lives? (2) Could disclosure promote accountability in a way that would improve public or private performance? In the words of the 2009 presidential memorandum:
The Government should not keep information confidential merely because public officials might be embarrassed by disclosure, because errors and failures might be revealed, or because of speculative or abstract fears. Nondisclosure should never be based on an effort to protect the personal interests of Government officials at the expense of those they are supposed to serve.18
Those words are important and correct. But they have one important qualification, to which I now turn.
INPUT TRANSPARENCY
When I was clerking for Justice Thurgood Marshall in 1980, Bob Woodward and Scott Armstrong published a book on the Supreme Court called The Brethren. I did not speak with Woodward or Armstrong, and I am also confident that none of my three co-clerks did so. But numerous clerks (largely or perhaps entirely from previous terms) decided to open up to the authors. The portrait of Justice Marshall was highly unflattering (and, by the way, wildly inaccurate). Marshall was clearly disappointed, much less (I think) because of the unfavorable, unfair, inaccurate portrait than because of what he saw as a breach of loyalty. I do not think it is disloyal to disclose what he said to us, which was roughly this: “I am not going to change how I interact with my clerks, but if you violate my confidence, it’s on your conscience.”
After I left the White House in 2012, many reporters, and some people outside the world of journalism, asked me questions about internal dynamics. Who said what to the president? Who disagreed with whom? If something happened, or did not happen, who wanted it not to happen, or to happen? Who won and who lost? Of course, I did not answer any of these questions, but there was no mistaking the (astounding) persistence with which they were asked. How well I recall a conversation with a superb journalist, working for the Washington Post, who was much focused on the who-disagreed-with-whom questions. I finally suggested to her that she should write something on the substance of the issues that most interested her (environmental policy). She did not seem enthusiastic about the suggestion.
As I understand them in this chapter, and consistent with the standard parlance, inputs count as both predecisional and deliberative. These are independent requirements. Inputs are predecisional in the sense that they are not themselves official decisions in any respect; they antedate those decisions and are meant to inform them. If an assistant administrator in the Environmental Protection Agency advises the administrator that a new ozone regulation should set a standard of 60 rather than 65 parts per billion, the communication is predecisional. Inputs are deliberative in the sense that they are part of a process of ongoing discussion about what to do.
I have acknowledged that even with these clarifications, we can imagine difficult cases, such as when a report is compiled on (say) the risks associated with silica and this report then becomes an input into a regulation. But the core should not be obscure. If law clerks are exchanging memoranda on how to handle a dispute over affirmative action, inputs are at issue. If people in the White House are discussing the contents of an open government memorandum, we are dealing with inputs. If White House officials are speaking with the Food and Drug Administration about how to handle the risks associated with certain asthma medicines, inputs are involved.
With respect to inputs, the argument for disclosure is significantly altered, and it is also weakened in two critical respects. First, the benefits of disclosure are usually much lower (definitely not always, but usually). Second, the costs of disclosure are much higher. These are categorical statements with major qualifications, to which I will turn in due course.
INPUTS AND MORE INPUTS, AND THE AMBIGUOUS BENEFITS OF DISCLOSING THEM
From the standpoint of the public, it is often not particularly desirable to obtain inputs. To those who believe in transparency, that claim might seem controversial, implausible, or even shocking. But 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. An assistant secretary will have countless conversations in the course of a week, and in many of them, she will be receiving suggestions, venturing possible ideas, requesting more information, joking, offering doubts, and seeking out possible inclinations. Some of the inputs that she receives or offers will not be very interesting. If they are interesting, it might be for a reason that does not exactly argue for disclosure: someone might have ventured an idea, for purposes of discussion, that was or is on reflection a really bad one. The idea was (let us suppose) rejected, and so it never became an output. Is it important, or on balance desirable, for the world to see it?
Now suppose that public officials are deciding what to do about particulate matter (an air pollutant). The director of the National Economic Council urges caution, emphasizing the overriding importance of economic growth. The Domestic Policy Council urges aggressive action, emphasizing that environmental groups keenly want the U.S. government to reduce particulate matter. Invoking international relations, the Department of State does the same. The Office of Information and Regulatory Affairs calls for a middle course, with close attention to costs and benefits. The Office of the Chief of Staff is focused on political considerations. Many memoranda are exchanged, offering various alternatives and competing points of view. It is far from clear how much the public would benefit from seeing this material. What most matters is what the government actually does, not who said what to whom.
It is true that for purposes of my thesis here, this example may not be the most convincing. The problem of particulate matter is exceedingly important, which complicates my argument (for reasons to which I will turn in due course). Consider then the general area of federal regulations, the most significant of which must go through the Office of Information and Regulatory Affairs (in most administrations, about five hundred per year). Many of those regulations will never be seriously discussed in the newspapers or online. Their issuance is preceded by a great deal of internal discussion, involving paper documents, electronic documents, and email, often raising questions and doubts. This is the quintessence of a deliberative process. A number of people say a number of things. Much of the time, the benefits of disclosing the content of that process are not much higher than zero.
Within the federal government, what is true for the regulatory process is true for may discussions—but even more so. The volume of emails is extraordinarily high. As in the case of the hypothetical assistant secretary, participants in these exchanges might float ideas, offer tentative reactions, or report on what some people appear to think. In general, disclosure would serve no purpose at all, except perhaps to those interested in genuine minutiae or seeking to embarrass, injure, or ruin someone, to create a political uproar, or to uncover some kind of scandal.
TWO QUALIFICATIONS
Two principal qualifications help to explain the appeal of input transparency for many observers. It must be acknowledged that in some administrations the qualifications will have, or be perceived to have, a great deal of power.
1. Illegitimate or illicit arguments. Public disclosure might provide an ex ante deterrent to arguably illegitimate arguments, and it might also provide an ex post corrective. At the very least, it lets the public know how it is being governed. In the worst cases, inputs include corruption or criminality (or extreme incompetence), and We the People are certainly entitled to learn about that.
Suppose, for example, that someone opposes a decision not because it is a bad idea but because it would offend a friend, a donor, a political ally, or a powerful interest group, or because a prominent senator might object with unfortunate consequences for the administration. That sort of thing is not exactly unusual, and in some forms, it is hardly the worst imaginable input. But let us stipulate that such an argument is objectionable, or at least that the public has a right to know about it, because it might compromise the pursuit of the public interest. Disclosure could make it less likely that such opposition will be voiced, which could be a good thing, and in any case it will create accountability. In this particular respect, an appealing argument about the beneficial effects of sunlight applies to input transparency as well as to output transparency. That argument is all the stronger in cases of the most egregious inputs, entailing corruption, self-dealing, or uncomfortably close relationships with nations unfriendly to the United States. (Readers are invited to use their imaginations.)
To be sure, disclosure could have the principal effect of shifting the locus of controversial argumentation—from email and paper to telephones. Within the federal government, this already happens a great deal. If people do not want their communications to be disclosed to the public or to Congress, they will say, “Call me.” (In my own experience, this was always innocent; it does not involve anything illicit, but it does involve issues that are somewhat sensitive, such as strong disagreements that are best not placed on email.) There is a substantial downside here. If internal discussions are potentially subject to disclosure, the shift from written to oral exchanges may impose losses in the form of diminished reliance on careful economic, legal, and other analyses.
Nonetheless, there is no question that a concern about illegitimate or illicit inputs animates and gives force to the argument in favor of input transparency. Suppose you believe that some process is “rigged”—that regularly or as a matter of course powerful private interests are dominating federal processes, or that officials, beholden to certain interests and groups, are pushing outcomes in the directions favored by those groups. Of course, you want that to stop. But if you cannot stop it directly, you might insist on input transparency as a way of opening it up to public view. Sunlight might be a disinfectant here as well.19
Again, there is a risk that you will simply drive the relevant influences underground. But in principle, that is a secondary concern if you want to open up internal processes to public scrutiny. To say the least, that is an honorable goal. For those for whom the Watergate scandal is salient, the transgressions of President Richard Nixon are a defining example here, as a degree of input transparency was necessary to ferret out those transgressions.
2. Learning from mistakes. The second qualification is that journalists and historians can benefit from seeing the internal give and take, if only because they could give a narrative account of what happened. This might appear to be an abstract, academic benefit, but people (including public officials) do learn from the past and that learning can provide a valuable corrective. The historical record can be absolutely indispensable for finding out what went wrong, and to understand that record, inputs are necessary. Why did the government make some colossal error in the form of an action or an omission? To answer that question, input transparency might be essential. It can create warning signs about group interactions that work poorly, about institutional blind spots, or about the need for institutional reform.
Suppose, for example, the United States government has done (too) little to prevent genocide.20 It may be difficult or even impossible to document the failures without access to inputs. Once the failures are documented, people might take steps to reduce their likelihood in the future. In that sense, the benefits of input disclosure can be high, at least in certain domains.
But there are countervailing points. In many cases, disclosure of inputs has no benefits; it does not reduce the risk of future errors. Disclosure also imposes a risk of distortion. Suppose that people have access to an official’s emails—say, the emails of an assistant administrator at the Environmental Protection Agency or of the assistant attorney general for civil rights. Suppose the email has some complaint about the EPA administrator or about the attorney general or about White House officials. The email might reflect a particular day or mood. It might be based on the author’s incomplete understanding. It might be a matter of venting. It might reflect a badly distorted perspective.
Because journalists often enjoy and benefit from accusations and scandal-mongering, it might be appealing to give a great deal of publicity to this revelation of internal disagreement. Recall that it is a form of gossip. Readers might enjoy the gossip and in that sense benefit from it, but accusations and scandal-mongering are not necessarily genuine benefits for the public. A genuine scandal is another matter.
THE COSTS OF INPUT TRANSPARENCY
For input transparency, the most obvious problem, of course, is that disclosure could reduce open-mindedness and discourage candor. In a short space, James Madison captured some of the essential points. In any deliberative process, people’s opinions are various and crude, and much is “to be gained by a yielding and accommodating spirit.” Once people commit themselves publicly, they might not be willing to shift. Secrecy can promote openness to the force of an argument. And, of course, Madison’s knockout punch: “No Constitution would ever have been adopted by the convention if the debates had been public.”
What Madison did not emphasize is that 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. In well-functioning deliberative processes, there is often a sharp separation between an idea-generating phase and a solution-finding phase. In the former phase, many things are on the table, even if they turn out, on reflection, to be absurd or intolerable. People say “yes” to getting ideas out there whether or not there is any chance that they ultimately will be adopted. If inputs are transparent, the idea-generating phase would be far more constrained than it ought to be.
Ensuring candor is, of course, the central idea behind the concept of executive privilege.21 At best, input transparency would lead people to communicate orally rather than in writing. In fact, one of the consequences of FOIA is a reduced reliance on email and written documents. In both Republican and Democratic administrations, it is well known that whatever is put in writing might find its way into the New York Times, which leads people not to put things in writing. At worst, input transparency can lead to certain things not being said at all.
But reduced candor is not the only problem. In view of the incentives of the media and political opponents, disclosure of inputs can produce extremely unfortunate distractions that are destructive to self-government. Instead of focusing on outputs—on how, for example, to reduce premature deaths—a spotlight is placed on comments that seem to make some people into villains or wrongdoers, or that put any resulting decisions in the least favorable light. Of course, skeptics might respond with some passion that it is paternalistic or worse to deprive members of the public of information on the ground that they will misunderstand it or give it undue salience. One view is that the receipt of true information should be subject to the marketplace of ideas. But insofar as the problem lies not in public misunderstanding but in the incentives of those who seek to fuel fires, there is definitely a downside risk.
A BRIEF ACCOUNTING
With respect to input transparency, we seem to have incommensurable values on both sides of the ledger, not easily placed along a single metric. The benefits are often low, but not always, especially when illicit motivations, corruption, and criminality are involved and when the historical record can help to avoid massive or catastrophic mistakes. The costs can be high. But are they always?
It must be acknowledged that the costs of input transparency diminish over time, and they are certainly lower once the relevant people no longer hold public office. It is one thing to tell the director of the Office of Management and Budget that whatever she says will end up in the newspaper that night or the next day. It is quite another to say that at a future date (say, after an administration has ended) there will be a public record of internal communications, subject to safeguards for national security, personal privacy, and other values. Indeed, the Presidential Records Act22 ventures an approach of this sort (with a five-year gap). With such an approach, the costs of disclosure are significantly reduced. They are not zero because candor will still be chilled and because people’s reputations may still be wrongly maligned. But in view of the value of obtaining some kind of historical record, this approach is hardly unreasonable. 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 trade-offs.
CONCLUSION
There is a large difference between output transparency and input transparency. For outputs, transparency can be exceedingly important. A central reason is that government often has information that people can use, perhaps to make life more navigable, perhaps to avoid serious risks. It should not keep that information to itself. Another reason is that sunlight can operate as a disinfectant. Whether the information involves the government’s own performance or the performance of the private sector, disclosure can spur better performance.
One implication is the immense importance of continuing with, and amplifying, the work of data.gov. It also follows that in numerous contexts government should not be waiting for FOIA requests; it should be disclosing information on its own. This does not mean that every output should be put on the Internet. It does mean that whenever an output could or would be valuable to members of the public, it deserves to be made public. With the help of algorithmic technologies, we should expect significant developments in this direction in the future.
Inputs belong in a different category. Outside of unusual circumstances, what most matters is what government actually does, not who said what to whom. For the most part, the public is unlikely to benefit by learning that the assistant secretary disagreed with the chief of staff of the secretary on some trade agreement, or that there was an internal division on how aggressively to regulate greenhouse gases or on the valuation of statistical lives. Disclosure can have significant costs. Most obviously, it can lead people to silence themselves or to communicate in ways that cannot be recorded. More subtly, it can divert attention from the important questions involving policy and substance to less important ones that involve palace intrigue. At the same time, input transparency can put a spotlight on questionable, illicit, or corrupt practices and can provide an indispensable historical record. People learn from the past, and for current administrations, it can be essential to have a concrete sense of where past administrations went wrong.
My framework throughout has been welfarist, asking about the costs and benefits of disclosure. It should be acknowledged that the very idea of welfarism needs to be specified and that many people would start with different foundations involving, for example, the idea of political legitimacy. It should also be acknowledged that under a welfarist framework some output transparency does not make much sense and some input transparency is amply justified, even indispensable. We are speaking of categories, not individual cases. But categories provide orientation. Output transparency should be the central focus of efforts for freedom of information; we need much more of it. Input transparency can be important, especially after an administration has ended. But it should be treated far more cautiously.
NOTES
From 2009 to 2012, I served in the Obama administration as administrator of the Office of Information and Regulatory Affairs and was involved in some of the matters discussed in this chapter. I am grateful above all to Larry Summers for many valuable discussions of the topic here. I am also grateful to David Pozen, Michael Schudson, and participants in a superb symposium held in June 2016 at Columbia University in honor of the fiftieth anniversary of the Freedom of Information Act. Pozen, in particular, deserves thanks for many valuable comments and suggestions.
1. See Amartya Sen, Poverty and Famines (Oxford, Oxford University Press, 1981).
2. See Amartya Sen, Development as Freedom (New York: Random House, 1999).
6. For evidence, see Archon Fung, Mary Graham, and David Weil, Full Disclosure: The Perils and Promise of Transparency (New York: Cambridge University Press, 2007).
7. Archon Fung and Dana O’Rourke, “Reinventing Environmental Regulation from the Grassroots Up: Explaining and Expanding the Success of the Toxics Release Inventory,” Environmental Management 25 (2000): 115–27.
10. See Partha Deb and Carmen Vargas, “Who Benefits from Calorie Labeling? An Analysis of Its Effects on Body Mass,” 2016, www.nber.org/papers/w21992.
11. See Fung, Graham, and Weil, Full Disclosure.
12. See Executive Order 12291.
13. Executive Order 13563.
14. Executive Order 12866.
15. Cass R. Sunstein, “Output Transparency vs. Input Transparency, August 18, 2016, http://d3i6fh83elv35t.cloudfront.net/newshour/wp-content/uploads/2014/07/EPA-cost-benefits-1024x468.jpg; memorandum from Cass R. Sunstein, Adm’r, Office of Info. & Regulatory Affairs, “Clarifying Regulatory Requirements: Executive Summaries,” January 4, 2012, https://obamawhitehouse.archives.gov/sites/default/files/omb/inforeg/for-agencies/clarifying-regulatory-requirements_executive-summaries.pdf.
16. See Lisa Heinzerling, “Regulatory Costs of Mythic Proportions,” Yale Law Journal 107 (1998): 1981–2070.
17. Memorandum from Peter R. Orszag, Director, Office of Mgmt. & Budget, to the Heads of Exec. Dep’ts & Agencies.
18. Obama, “Memorandum for the Heads of Executive Departments and Agencies.”
20. Samantha Power, A Problem from Hell (New York: Basic Books, 2002).
21. See United States v. Nixon, 418 U.S. 683 (1974).
22. 44 U.S.C. §§ 2201–2207.
