Editor’s note: This post also appears on Lawfare.

The problems in the prepublication review process that we identified in our Washington Post op-ed and in a subsequent post are very hard to document precisely. There are a handful of high-profile public cases involving challenges to the government’s classification decisions, most of which the government wins. And there are stray public complaints about the government taking too long to review, or exercising political judgment in deciding what to censor. But based on our experience and the experience of others we have spoken to, the vast majority of the errors and abuses in the system remain undisclosed. Most people subject to the prepublication review process have little incentive or opportunity to complain when something goes wrong. One of the reasons we wrote the op-ed was to shine light on and hopefully provoke discussion about the issue.

While we cannot document the scope of the problem, we have no doubt that the system of prepublication review results in the non-publication, or excessively delayed publication, of a great deal of potential writing related to the government that has nothing to do with classified information. Most of the limited public debate about the issue focuses on government orders not to publish information that authors think contains no classified information. But in our view a much larger problem is the chilling effect of the process, especially as a result of broad government constructions of the duty to disclose that leads authors simply not to write on certain topics at all. Two recent egregious cases of overly broad constructions of the duty to disclose imposed on acquaintances of Jack’s — cases that unfortunately we cannot here discuss — are what led us (in part) to write the op-ed. As we noted in the op-ed, the Defense Department even acted as though we had a duty to submit for preclearance review the Washington Post op-ed — a small indication of how absurdly overbroad the system has become, and how little the government takes account of First Amendment values in the prepublication process.

Since we published the op-ed, we have heard about similar experiences. A former senior intelligence official sent us the following powerful observations:

You hit the tip of an iceberg I have rammed into on too many occasions. … One particularly worrisome tendency is for the CIA to expand the scope of review to areas only tangentially related to the legitimate interest of protecting classified information and safeguarding sources and methods. For example, it is not uncommon for the [Publication Review Board] to send a manuscript to the NSC to review for sensitive “policy content” — classified or not. This risks the politicization of intelligence and encourages the application of subjective standards for authors depending on seniority, political affiliations and views, etc. My biggest concern is the agency’s apparent arrogance in wielding its unchallenged, absolute power of censorship. I have seen scant appreciation for setting boundaries between free speech and keeping “secrets” — just for the sake of keeping secrets, and for no other justification. In this context, if the agency doesn’t like a manuscript, there’s a good chance an excuse will be found to delay or redact it. If the substance is favorable from the agency’s perspective, an author might get preferential treatment. The subjectivity inherent in the prepublication process is stunning; Hollywood stars and directors are greeted with open arms by the agency, whereas retired officers tend to be treated suspiciously. The PRB clearly doesn’t like being challenged on the basis for its redactions; the result is an opaque, inscrutable, secretive process. Logic and reason offer no advantage to an author in arguing for reconsideration of redacted passages, because the agency knows that individuals have little recourse to challenge its authority. As a citizen, while mindful of the need to balance national security and freedom of speech, I have come to the conclusion that the CIA is incapable of being entrusted with such an unusual and Constitutionally questionable authority. I believe there should be a broad external review of current procedures, with an addition of a non-intelligence controlled oversight process to continuously monitor prepublication review. The power of censorship is simply too important to be entrusted to an institution that is prone to using it to protect its parochial interests.

One might reasonably ask whether the benefits of the prepublication review process are worth the manifest costs of the process. In assessing this issue, it is important to remember that prepublication review is just one way that the government prevents disclosure of classified information by former government employees. In some cases, the author also faces the powerful threat of criminal prosecution if he or she publishes certain classified information. (Even if conviction is unlikely, the mere threat of investigation or indictment is enough motivation for many.) The relevant questions are thus: What protection does prepublication review as currently practiced add to the criminal punishment threat, and are these gains worth the costs to freedom of speech and democratic government that we have outlined?

We have no doubt that prepublication review sometimes catches and prevents the publication of classified information that a conscientious author might have missed for any number of reasons. But these benefits must be qualified in at least two ways before one even considers costs. First, when authors write and submit material for review, it is vulnerable to a sophisticated hacker. We have been encouraged to submit our material for prepublication review via unclassified email. We write the documents on our personal or university-provided computers. Any classified information they might inadvertently contain is therefore already compromised the moment it leaves our fingers. Second, the current cumbersome, time-consuming, and seemingly arbitrary prepublication review process leads many people to simply avoid it altogether or at least substitute a personal standard of submission for the government’s standard.

To us it seems clear that the current system is dysfunctional and imposes many more costs than it yields in benefits. The challenges to fixing the system are significant, however. The government has a legitimate interest in preventing the disclosure of classified information. But putting the decision entirely in the hands of the government makes it likely too much will be withheld because the government’s incentives are skewed, and the line between classified information protection and self-serving policy interest is often blurry. Relying entirely on the author also carries risks: Those who want to publish might have an incentive to draw the line in an overly permissive direction, and may also have difficulty judging what information is classified and which is not. Both problems are exacerbated by rampant over-classification, which makes so many potential writings touch on or implicate classified information — and makes it all the harder to draw an appropriate line.

We believe that the standard for review that we propose — “writings that might reasonably contain or be derived from classified information” — accommodates the interests of both the reviewed and the reviewer, though the standard is still unclear and still susceptible to abuse by the government without the significant review and independent oversight that we propose.

We also think that the government must enforce its standards in a way that absorbs some of the risk of disclosing classified information. Right now the government enforces a super-cautious standard in prepublication review that accepts no risk of disclosure while imposing nearly all the costs on former government employees’ freedom of speech — and, with it, the public’s right to know. This standard assumes that preventing disclosure of classified information is the only relevant value. But national security, as important as it is, is not the only value; it is one among many, including the constitutional value of freedom of speech and the value of an informed public, both of which are essential to democratic governance. Especially since the government controls the classification stamp, and because the stamp’s extravagant use is the root cause of the problem, the government should be required to accept a standard for prepublication review that accommodates speech even if there is a limited risk that classified information will be disclosed to the public — a risk, after all, that can never be eliminated. Otherwise the value of secrecy will always prevail, which is especially bad since the government itself controls the scope of secrecy and has every incentive to expand it in order to tamp down criticism.