The issue with regards to understanding the North Korean thought process is that we have almost no human intelligence (HUMINT) in the country. Though satellites can tell you what the North Koreans are doing (where their equipment is, where their prison camps are, and how much material they possess), it cannot tell us why they are doing it.

Interestingly, despite the many flare-ups and provocations on the Korean peninsula, the status quo has proven incredibly resilient over the course of the past 60 years. There has not been a full-scale war, despite the many incidents that would normally ignite one in other situations. Making this case, in 2003, Victor Cha and David Kang wrote in Foreign Policy:

“Since the end of the Korean War in 1953, there have been more than 1,400 incidents across the DMZ, resulting in the deaths of 899 North Koreans, 394 South Koreans, and 90 U.S. soldiers. Tensions have been so high that in 1976 the United States mobilized bombers and an aircraft carrier battle group to trim one tree in the DMZ.”

“Tensions,” however, do not necessarily mean that the situation is as unstable as people might assume. So why hasn’t the peninsula broken out into open war, especially with such loss of life over the years? The reality is that neither side truly wants it. The costs outweigh the benefits for all involved — and if the North Koreans legitimately recognize this they will not undertake actions that they believe could ignite open war. In conventional terms, the North Korean military, despite its size, has been inferior in quality to the South Korean (ROK) armed forces for years, and any protracted war would likely end in the elimination of the North Korean regime.

For South Korea and the United States, retaliatory options against North Korea for its many provocations, short of outright invasion (which China would actively protest), are extremely limited.

This effectively gives the North Koreans free range to provoke nearly as much as they’d like. As any US/ROK retaliation short of full invasion will likely catalyze an escalating chain of events that ultimately leads to the need for an invasion, North Korea can wave its guns around and fire wildly — scaring off adversaries without fear of effective reprisal. The appearance of being unstable and crazy essentially acts to serve as a deterrent on its own — rendering the U.S. and ROK unwilling to retaliate for fear of setting off a full scale war.

A key question for the U.S. and ROK to work out is what a response will be in case the North Koreans do something that not only appears crazy, but is legitimately crazy — such as firing off a few artillery shells at Seoul. We do not actually know to what extent the Kim regime is willing to provoke and take actions beyond what it sees as necessary to save face and legitimize the regime internally.

If North Korea takes actions that it predicts will likely push the US/ROK response over the edge (and ultimately cause the end of the Kim regime), that would indeed be crazy.

The key for the U.S. is to work on establishing some sort of system of intelligence. Getting information out of North Korea is incredibly difficult for a few basic reasons, and much of the problem amounts to geography. North Korea has two land borders, one of which is heavily fortified, and the other is China. This is not like the Cold War game with the Soviet Union, where incredibly long borders neighboring many countries allowed for relatively easy access. Not to mention, the U.S. actively coupled public diplomacy efforts into its overall strategy in order to assure an on-the-ground presence, however innocuous, within Soviet territory.

It is incredibly important that we work harder on establishing HUMINT resources inside North Korea. It is incredibly important that we develop plans to address flagrantly crazy actions taken by the North Korean regime. And it is incredibly important that we do none of this alone.

Wallin is a senior policy analyst at the American Security Project.