Finally, the moment of truth is at hand. After months of preparation and calculations, after checking and rechecking miles of wiring, thousands of screws and hundreds of computer algorithms, it’s launch time.

In the command center, all eyes gaze at the giant monitor. The rocket’s red glare fills the screen and the boosters maneuver the kill vehicle toward its target: a simulation of an incoming ballistic missile.

Will the kill vehicle hit its target, or will it miss?

If the test is designed properly, it can be a huge success even if the incoming missile gets through.

In any missile defense test, people naturally want to know whether the good missile shot down the bad one. But that shouldn’t be the only — or even the primary — criterion for determining whether the test was a success or failure.

A test that ends with the target destroyed is not necessarily good. What if the test was designed to make interception of the target missile all but certain? What have we learned? Rigged or “dumbed-down” tests can teach us nothing.

In missile defense, the stakes are enormously high. Hundreds of thousands, if not millions, of lives can be lost in a single attack. The tests must be rigorous and challenging so that we learn as much as we can. Tougher tests increase the likelihood of missing the targets, but they increase the rate of progress in advancing missile defense technology.

Consider our Ground-based Midcourse Defense system, which features 30 interceptors deployed in Alaska and California and is the only system protecting the homeland from long-range ballistic missiles. Of 15 intercept tests to date, only eight ended up with targets destroyed; indeed, the last successful intercept was more than four years ago.

That is not to say the tests were failures.

In a 2010 test, the interceptor missed its mark. The media reported it as a failure, yet the test proved that a new kill vehicle could discriminate between an incoming warhead and the surrounding decoys and debris with similar radar footprints — an extremely difficult task. The test was a success because it validated the concepts and technologies underlying the experimental kill vehicle.

“Failure” is a normal — in fact, necessary — part of experimentation and progress. That holds true for any field of scientific inquiry, including weapons development. The Polaris submarine-launched ballistic missile, developed in the 1950s, “failed” 12 out of 17 times — all within a span of about a year. Despite this track record, the missile was deployed, and its many technical advances were incorporated later in the Trident SLBM.

If one looked only at the test “failures,” one could easily dismiss both systems as unproven and unreliable. In reality, the tests had been pushing technology to the extreme. As a result, the developers learned more, faster. The tests produced a complex technology so advanced that the Trident SLBM today stands as a bedrock of our strategic deterrent. It is expected to remain in service until 2040.

Defense systems can work, even if they fail to hit their targets every time during testing. Our sea-based Aegis Missile Defense system failed to destroy the target in six out of 30 tests, yet it is regarded today as the most effective component of U.S. missile defense.

Tests are, and must continue to be, designed to push the system to the point of failure. The goal is to learn as much as we can about the performance of each of the many interceptor components in different phases of the test.

Aggressive testing allows us to improve our missile defenses as quickly as possible, advances that are needed now more than ever. So let’s have some more test “failures,” because we want our defenses prepared for success if North Korea or some other bad actor decides to attack a U.S. or allied city.

-Michaela Dodge is a defense policy analyst in the Heritage Foundation’s Allison Center for Foreign Policy Studies.