This article is part of a series from the November 2017 Harvard University conference entitled “Presidential First Use: Is it legal? Is it constitutional? Is it just?” To access all of the transcripts from this conference, click here.

The current US protocol for deciding whether to launch a nuclear strike—developed in the early 1960s, with the advent of intercontinental ballistic missiles—has two main functions and virtues: first, it concentrates the power and authority over the use of nuclear weapons in the presidency, at the highest level of the executive branch of the US government, thus keeping it out of the hands of the military and others. Second, it enables the president to respond rapidly and decisively to a nuclear attack by an enemy whose missiles may fly from one side of the planet to the other in 30 minutes; or whose missiles launched from submarines in the oceans may fly to targets in the United States in 15 minutes. It’s critical to have a protocol that allows the president to consider the use of nuclear weapons and, if necessary, to order their use, and to have the process of implementation begin in a very, very short period of time.

The protocol’s virtues also produce its disadvantages. By virtue of the speed and concentration of authority in this protocol, the president has an opportunity to effectively railroad the nuclear commanders and forces into executing even a very large nuclear strike first—preemptively or preventively. That could lead to a misguided decision based on an impulsive psychology or on other factors that lead to a very bad call.

The other downside to this protocol, which we used to talk about a lot more than we do today, is that the protocol itself, rooted as it is in speed and concentrated authority, can railroad the president into authorizing, in a hasty way, the use of nuclear weapons based on indications, possibly false, of an attack underway (a strategy known as “launch on warning”). In other words, we might believe we are retaliating when in fact we’re launching first.

During the Cold War, to my knowledge, a false alarm never led to notification of the president at the beginning of the protocol that I’m about to describe. The false alarms were caught before that happened. Ironically, today, with the proliferation of ballistic missiles over the last decade (there’s been a huge surge in ballistic missile proliferation, and in their testing) you find that recent missile launches—from China, from Iran, from North Korea—have led on multiple occasions to sufficient ambiguity that the presidents have actually been notified about the ongoing event.

Here are the key features of the current protocol. It begins with an early-warning function: the effort to detect a possible attack against North America and to notify the president and others to begin a process of deliberation. Every single day, the early-warning staffs out in Colorado and Omaha pick up events that require a second look to determine whether we’re under attack. Events they might review include a Japanese satellite launch into space, a North Korean missile test, a US missile test out of California, a wildfire in the southwest US. Most of these are usually dismissed quickly. Once or twice a month, something happens that requires a really close second look. And once in a blue moon, something happens and all hell breaks loose, as in the case of a false alarm concerning a missile launch.

If these staffs receive any indication that we may be under attack, they have three minutes from the time the first sensor data arrives until they have to provide a preliminary assessment as to whether North America is under attack. If the assessment is of medium or high confidence that there is a threat, they initiate a process that will bring the president and his top advisors into an emergency conference no matter what time of day or night.

Imagine that the president has decided to initiate a conference with his top advisors to consider the first use of nuclear weapons. The United States does not have a no-first-use policy. Furthermore, under the current review of our nuclear policy, undertaken primarily by the Pentagon, there is an emerging thesis that we should move further away from no first use and consider use of nuclear weapons in a wider variety of contingencies. We are on the verge of modifying our assurance to non-nuclear-weapons countries that we would not use nuclear weapons against them, in contradiction to the position adopted by the Obama administration.

The emergency meeting of the president and his top advisors will typically include statutory members of the National Security Council: the secretary of defense; the secretary of state; the national security advisor; the chairman of the Joint Chiefs, who participates at the discretion and the invitation of the secretary of defense; and a number of key military command centers and personnel, the most important of whom is the commander of strategic forces based in Omaha, Nebraska, who commands all our strategic nuclear weapons.

Time and circumstances permitting, the commander will brief the president on his nuclear options and their consequences. It will not be a long briefing. He’s going to have to boil this down into very, very brief sound-bites for the president: here are your options and here are the consequences. The commander will then ask the president a couple questions, such as whether he wants to withhold attacks on a particular location, such as a populous city. That briefing, if we are under attack, will be as short as 30 seconds. Of course, if the president is considering the first use of nuclear weapons, the timeline is not nearly as short and that conversation can last for quite a long period.

If we are under attack, the president is going to have to consider his options in about six minutes, given how this protocol tends to work. If we’re not under attack, he can deliberate longer. Then he makes a decision: What option am I going to pursue? Am I going to decide to attack North Korea, for example? (With the current preprogrammed attack plan, I estimate we would have 80 nuclear aim points in North Korea.)

Let’s say the president chooses an option. It will be conveyed instantly to the war room at the Pentagon, which probably initiated the presidential conference in the first place. The people in the Pentagon war room are listening in on the conversation and are beginning, as they hear the president moving toward a decision to use nuclear weapons, to prepare a launch order.

Note that the secretary of defense does not confirm the president’s decision, nor does he or she have a right to veto it, nor does anyone else have the authority to override the decision. This is what Elaine Scarry has identified as, in effect, a “thermonuclear monarchy,” which gives the US president almost carte blanche command over the nuclear forces.

When the president conveys his decision to the war room, they ask him to authenticate his identity using a special code. It’s referred to colloquially as “the biscuit,” otherwise known more officially as the “gold code.” If that code matches, the war room at the Pentagon, or an alternate, will format a launch order that will be transmitted down the chain of command to the executing commanders of the submarines, land-based rockets, and bombers.

That launch order is roughly half the length of a tweet. It contains all the information necessary for the crews down the chain of command to launch their forces: the time to fire, the chosen war plan, an unlock code that the crews need to physically unlock their weapons prior to the launch, and special authentication codes that the crews check with the codes in their safes to satisfy themselves that these orders came from the president (those codes are not in the possession of the president, but of the military).

That takes two minutes: 10 seconds to authenticate, then a minute or two to format and transmit the order. And in two more minutes, from the receipt of that order down the chain of command, missiles could be leaving their silos; it takes only about one minute for a Minuteman crew in the plains states of the Midwest to carry out their launch checklist. This was my job in the 1970s and at the time, it took me one minute. We delayed a little bit, for classified reasons, but that’s how long it took then and that’s how long it takes today.

After the crews enter the war plan it goes out to all the missiles, which are preprogrammed with what wartime targets to strike. In peacetime, they are aimed at the ocean, but changing their targets to Moscow or any other targets is as easy as changing the channel on your TV set.

Today, within a minute or two there can be up to 400 high-yield strategic weapons launched out of their silos to their targets, wherever those targets may be. Submarines take about 10 minutes longer because it takes them longer to target their missiles, position the submarine, and get to the proper depth. But even submarines on alert in the Pacific and Atlantic Oceans would within 15 minutes be launching missiles out of their tubes, then firing them one at a time every 15 seconds.

Lastly, the bombers would take 8 to 10 hours to reach their launch points if they were already on alert. They are not normally, today, in peacetime, on alert. They don’t even have bombs on board, so in a crisis they would have to be placed on full alert, with bombs and cruise missiles loaded, before they were usable.

To sum up: the president wakes up, gives an order through a system so streamlined that there’s almost no gatekeeping, and, within five minutes, 400 bombs leave on missiles launched out of the Midwest. About 10 minutes later, another 400 leave on missiles launched out of submarines. That’s 800 nuclear weapons—roughly the equivalent of, in round numbers, 15,000 Hiroshima bombs.

Reform of current US launch protocol is long overdue. Layering on new safeguards that strengthen checks and balances on presidential launch authority is necessary to reduce the risk of nuclear first use. Safeguards include the Markey-Lieu Bill, which would prohibit the president from employing nuclear weapons first unless Congress has declared war and provided specific authorization for their use; the Betts/Waxman solution, which would add the secretary of defense and attorney general to the chain of command to certify that a presidential launch order is authentic and legal; and adoption of a no-first-use policy, which would draw a red line that, if crossed, makes the president accountable and even impeachable.

Regarding a second strike, the United States should eliminate launch on warning and move toward a true retaliatory posture, requiring protection of the president and his successors and providing a large increase in warning and decision time.