North Korea, frequently the butt of jokes and memes for being backwards, is preparing to test an intercontinental ballistic missile (ICBM). Such a missile, unlike the medium-range variety the country tested off its east coast on Monday, is one that could possibly deliver a warhead to the American mainland. The possibility is no joke, and it is going to be one of several painful discussions that Presidents Donald Trump and Xi Jinping need to have at their upcoming summit in Florida. According to Thae Yong Ho, a recent North Korean defector who had served in the country’s embassy in the U.K., Kim Jong Un wants a test in 2017 or early 2018.

This test will likely fail. Launching an ICBM successfully is hard, and there are many points where failure can occur. The regime spent much of 2016 conducting component tests and rolling out propaganda images and video to promote its capability. While North Korea is notorious for faking images of its weapons programs, there were several images that stood out as probably true.

Officials in the U.S. and abroad have become increasingly wary of North Korea’s space launch program at Sohae on the country’s west coast. The rocket launched there as the Unha-3/Kwangmyongsong in 2012 and 2016 is in many ways similar to an ICBM. Like a space launch vehicle, an ICBM can consist of two or more component parts called stages, each of which carries fuel and an engine. The stages burn and drop off into the ocean one by one, until the payload—whether a satellite or a warhead—is in space. However, unlike a satellite that will orbit the earth, an ICBM warhead needs to reenter the atmosphere and land at a designated target.

It’s unlikely that the Unha-3 will ever be used as an ICBM. Instead, North Korea has paraded what appear to be two variant road-mobile ICBMs that should have the U.S. worried. To date, these have not yet been tested, so it is not clear they “work,” but road-mobile missiles can be constantly shuttled around the country, and hidden in warehouses, tunnels, and bunkers. This makes them hard to track from satellites, thereby increasing the likelihood they would survive a strike on them. The semi-good news is that North Korea only has six untested ICBM launchers, so no matter how many missiles or warheads North Korea makes, it is constrained. In addition, the trucks needed for road-mobile ICBMs are huge, and must travel in convoys that include fuel and oxidizer trucks as well as a large number of personnel. Many of the facilities and hiding places where North Korea puts these trucks are already known, including where the trucks were adapted and a prominent ICBM missile warehouse.

The KN-08 ICBM has been paraded in Pyongyang since 2012. At first, open-source analysts found its crude workmanship unconvincing, but by 2016 North Korea demonstrated that not only is it a real ICBM, but its engine is using a more energetic fuel than previously thought. The translucent pinky-purple color of the flames seen in photos of the April 2016 engine test at Sohae are a hallmark of a fuel-oxidizer combination energetic enough to make a target not only out of the West coast of the U.S., but also Washington, DC.

The KN-14, paraded in October 2015, is a variant. It is less frequently photographed, but if it has the same engine, it can probably put at least Chicago at risk, and may even be more accurate than the KN-08.

Still, North Korea has never tested an ICBM, so there is a fair amount of educated guessing going on in and outside of several governments about when, where, and how North Korea might make its debut. The easiest way to find out would be if North Korea issued a notice to airmen, known as a NOTAM. The idea behind a NOTAM is to warn sea and aircraft of potential hazards during the flight time of the rocket. The North Koreans have done this before for their space launches, but not for their missile launches. North Korea, if you read The Atlantic, I strongly recommend that you issue one in advance of an ICBM test. By giving advanced warning, you are making it clear that you are starting a test and not restarting a war.

But North Korea is not a very trusting state, so its leaders may not issue a NOTAM for fear the U.S. or South Korea might strike the missile before it launches. There’s been a lot of talk in the U.S. about doing just this, though it is another quick way to restart a war.

Without a NOTAM to tip off observers that a launch is imminent, the next way to know what’s happening is to turn to satellite imagery. Those of us in the open-source world use commercial satellite imagery companies like Planet to get near-daily updates. Where does one start to look? One possibility is that the North Koreans will not use a transporter erector launcher (TEL) truck of the kind visible in parades, because they are scarce and it would be too damaging to the ICBM program to lose one in what will likely be a failed first test. That makes Sohae a possibility. The satellite launch site there conveniently already has everything one would want for a test—a gantry tower for stabilizing the missile, fuel, oxidizer, engineers, and communications that make it easy for Kim Jong Un to observe from afar in case of failure or preemptive strike.

Alternatively, the North Koreans could set up a temporary gantry tower or use a TEL somewhere else in the country. They would need to build or use an existing flat, probably paved, area with good sight lines for communicating their data. This considerably widens our search area, but my bet is that like a space launch, an ICBM test would launch from the west coast, heading south into the ocean. That way, if it fails, it won’t land on South Korea and restart a war—nor, if it fails, will it land on Japan and start a war. Similarly, I don’t think the North Koreans are looking start a war with China or Russia anytime soon by aiming west or north. Large airstrips and highways in the west make good candidates for launching sites, and we’ve already seen some shorter-range missile tests and failures from these locations. At this point, there may be rumors of activity leaked from government sources, or open-source researchers may detect the location in advance due to construction and vehicle activity at the site.

Once the missile is launched, most open-source analysts will turn to Twitter, where there are other analysts as well as the accounts of the U.S. Strategic Command, Pacific Command, the North American Aerospace Defense Command, and a variety of Korean and Japanese publications willing to circulate statements and leaks from their governments. The U.S. has space-based overhead persistent infrared (OPIR) sensors which can detect a missile launch (and failure) from nearly anywhere. South Korea and Japan have ground-based radar, which means they only detect missile launches as the missile rises over the horizon. Information from all three usually becomes available nearly simultaneously. This can include information on location, missile type, bearing, apogee, and range. Sometimes information is lacking, and occasionally conflicting, or wrong.

Within minutes, the world should know if there is a failure. If ground-based radar doesn’t detect it, then the test failed before the missile even made it over the horizon. If the first stage of the missile does not land in the designated area of the ocean, it could mean that the stages didn’t separate correctly or the first stage malfunctioned. If what’s known as a “splashdown” does occur, it will give us information about the bearing of the missile. During North Korea’s space launch last year, the first stage of the rocket exploded into approximately 270 pieces before splashing down, indicating that North Korea may have intentionally destroyed the stage to prevent examination.

What will North Korea launch in its first ever ICBM test? I’m hoping a dummy warhead that will test the ability to deliver a device through the heat, pressure, and vibration of re-entry through the atmosphere. Since North Korea does not have its own space-based sensors, it will not be easy for the country to track the dummy warhead’s progress far afield. It also will not be easy for the North Koreans to recover it if they launch it south into an ocean. There is a small chance that the North Koreans could be reckless enough to try an atmospheric nuclear test from an ICBM, though it is very doubtful, especially on their first few ICBM tests. At least, I really hope they don’t. So, Kim Jong Un, if you happen to be reading The Atlantic today, please don’t restart the Korean War.