Yesterday, a memo describing the president’s legal justifications for drone attacks against U.S. citizens was obtained and published by NBC’s Michael Isikoff. The memo is a disturbing assertion of discretionary executive power that should concern and frighten all Americans. Unfortunately, the secretive use of drone attacks is one of the few areas of bi-partisan consensus in this highly divisive town, and the public still seems to resoundingly support current counter-terrorism policies.

Not being a foreign policy expert, I will not get into the broader questions of counter-terrorism policies. I agree, as I think most Americans would, that there are times in which the government can justifiably use lethal force against even its own citizens. As always, however, the devil is in the details, and here the details are encapsulated in the broad, discretionary language of the memo. Abstractly agreeing that there are times where a killing is justified does not answer who will determine when to use such force, what standards they are expected to uphold, and what possibilities of review exist for mistakes.

These standards—the “who,” the “how,” and the “possibility of review”—are at the core of the Western legal tradition. Putting process—that is, how something is determined—on equal level with substance—what is determined—is one of the Western legal tradition’s most important contributions. The goal of a legal system is not just to reach the correct result, but to reach that result via a just, open, and reviewable process. Fundamentally, these principles are concessions to our inevitable predilection for errors in thinking, judgment, and fact-gathering. The lynching of an obviously guilty child molester is problematic not just because of the disturbing result, but for how that result was determined.

Those are the principles that we should hold dear when analyzing the memo. Perhaps every drone attack has been the correct call (something we know isn’t true), and high-level officials certainly care about civilian casualties. Nevertheless, if we believe in the principles of the Western legal tradition, we shouldn’t okay with this power if it were in the hands of Mother Theresa.

When the memo is parsed out, the possibilities of error and misuse are obvious. In the most head-scratching line in the memo, the authors redefine the concept of “imminence”: “the condition that an operational leader presents an ‘imminent’ threat of violent attack against the United States does not require the United States to have clear evidence that a specific attack on U.S. persons will take place in the immediate future.”

This redefines “imminence” as a mere “possibility.” In both international law and at common law, “imminence” defines the situation where an individual or a nation can justifiably use self-defense. As Daniel Webster defined it in the Caroline Affair, it is a threat that is “instant, overwhelming, and leav[es] no choice of means, and no moment for deliberation.”

The memo defines an “operational leader” as someone who

is personally continually involved in planning terrorist attacks against the United States. Moreover, where the al-Qa’ida member in question has recently been involved in activities posing an imminent threat of violent attack against the United States, and there is no evidence suggesting that he has renounced or abandoned such activities, that member’s involvement in al-Qa’ida’s continuing terrorist campaign against the United States would support the conclusion that the member poses an imminent threat.

In the emphasized section, the looser definition of “imminent” is presumably being used. Thus, to be subject to a kill order, someone need only to have “recently” been involved in “activities” posing the mere possibility of a violent attack against the United States, a broad and expensive definition indeed. Moreover, the words “recently” and “activities” are incredibly vague. There are holes here you could easily drive a truck through—or a drone.

Obviously, the best rejoinder to my argument is that there are people out there who wish to do us serious harm, even possibly using nuclear weapons, and therefore our interests are particularly acute and thus our margin for error should be bigger. I agree that such people exist. I am not so optimistic, however, that drone strikes are being confined to even those broader individuals within the margin of error. Yet, even if they were, I would still have a problem with the discretionary and unreviewable decisions being made.

Drones pose a particularly acute problem from a public-choice standpoint. Currently, the president and other high-level officials suffer almost no costs for drone attack mistakes. Conversely, the costs to them, politically, personally, and in their legacies, of allowing a terrorist attack on U.S. soil are quite high. Many people will blame President Obama for any attack that occurs over the next four years. Intelligence reports after a future attack will inevitably point out when the U.S. could have acted but chose not to. Perhaps a drone had the eventual terrorist in its sights but the kill order was not given, maybe because innocents were in the area. Therefore, we can reasonably expect high-level officials to err on the side of overkill, and perhaps this is a defensible policy.

But we should not and cannot ignore the costs incurred by the civilian living under constant fear of drone attacks. A recent report, Living Under Drones, from Stanford and NYU law schools estimates that between 474 and 881 civilians have died in U.S. drone attacks, including 176 children. Moreover, the constant presence of drones over northwest Pakistan has caused “considerable and under-accounted-for harm to the daily lives of ordinary civilians,” including the undermining of “cultural and religious practices related to burial, and made family members afraid to attend funerals” out of fear that drones will attack large gatherings of people.

These costs to innocent civilians must always be considered. Yet, the memo disturbingly omits a crucial element of the Supreme Court’s due process test that would give these costs more weight. In Mathews v. Eldridge, the Court articulated the test for how much process is due a citizen who the government seeks to deprive of some vested right. That case dealt with depriving Mr. Eldridge of his social security benefits, but the test used by the Court is generally applied to all possible “deprivations,” including life. The test is three-pronged: 1) the nature of the private interest; 2) the risk of error in the procedures used and “probable value, if any, of additional or substitute procedural safeguards”; 3) the nature of the government interest. Astoundingly, even while citing Mathews, the memo omits the second factor entirely. As Lawfare’s Steve Vladeck writes:

There’s no discussion–none–of the risk of false positives under the existing procedures, or the potential cost of additional process. This turns the Mathews test on its head, for it suggests that the relevant question in any case is simply whether the balancing of the interests supports the already provided level of process–and not whether the error rate and/​or cost of more process is at all relevant to that determination. Not only has the Supreme Court never so understood the Mathews test, but such an approach would convert an already controversial metric for “measuring” due process into a completely standardless one–and completely obfuscate the underlying principle that the government has an obligation to provide as much process as can reasonably be expected under the circumstances.

Ultimately, this is the omitted factor that Americans should care about. There are certainly heightened interests when it comes to protecting Americans from violent attacks, and the government can act swiftly to counter a truly imminent threat, but it must take into account the risk of error and whether more procedures could help minimize those errors. And we must be clear: sometimes these “errors” are dead children.