What’s up with this “EM Drive”?

Punchline: If someone offers you a high-risk, high-payoff investment opportunity, here are some clues that it might actually be a zero-payoff opportunity.

Likely that only other experimental physicists will appreciate the details, because failure investigations can help us do better in the future.

The AIAA’s Journal of Propulsion and Power has kindly made the “Measurement of Impulsive Thrust from a Closed Radio-Frequency Cavity in Vacuum” article available for free. You can read about the experiments at Johnson Space Center that the authors claim demonstrate an effect that defies the laws of physics and changes the game in interplanetary space travel.

Having had a couple of hours to read the article through, I think I’m starting to understand what they did. We’ll forgive them using supoptimal apparatus (it was probably what they already had) and having an unorthodox way of engaging with the broader scientific community. And I would have liked them to have converted their time and distance measurements into units of acceleration and force, what they describe actually doing (like turning the cone around, measuring temperatures, calibrating the electrostatic thingamabob) shows that they acted in good faith to address several of the more obvious systematic errors. So I think it’s more likely to be an error rather than a deliberate hoax. Or, who knows, our understanding of the basic principles of physics could be completely wrong.

Ha ha. I’m not holding my breath.

The first big red flag is that the duration of time that the RF was on is different in every plot. There’s no discussion of how they decided when to turn it on or off. The operator may have been unconsciously conveying his desired outcomes through the power switch. I write, “may”. There may be a reasonable explanation.

The second big red flag is that the tests are conducted inside a metal can — and not a very large one, at that — and there’s no discussion of how it might interact with the RF they’re spewing out. They went to elaborate lengths to “tune” something about the RF. I don’t understand that part well. Maybe the cavity they’re exciting isn’t the one they think it is because of the chamber wall. A frequency-dependent phase shift is often sign of a time delay, and light has a well-known velocity, so this might be something that could be calculated, so this hypothesis of mine could be easily falsified by looking closer at those details (which weren’t published). They seem to be aware of the possibility of interaction with the environment, so they turn the copper cone around, but they never move it to the other side of the support bar, which is an obvious axis of symmetry.

The third big red flag is that they show raw data from a few events, rather than averages of multiple runs. If it’s truly repeatable, we would expect them to repeat it, at least a few times, and tell us about it. Now, it could be that they only barely got it to work once, and a follow-up paper will confirm the observation. But how do we know that they didn’t cherry-pick the plots that looked good, and all we are seeing are the random noise outliers? That has happened before — cold fusion — but it doesn’t necessarily mean that the authors are wrong — as in gravitational waves.