Chris Lee in Ars Technica on NASA’s EM-drive, the rocket engine that either runs on heretofore undiscovered physics, or experimental errors. Spoilers: it’s probably the latter.
Then there’s the error analysis: the authors estimate many measurement uncertainties so that each thrust measurement has an uncertainty of about ten percent. That sounds brilliant, right? Except the authors ignore the main uncertainties. In one experiment at 60 Watts of microwave power, the authors measure thrust of 128 microNewtons, while all three data points for 80 Watts of microwave power have thrusts of less than 120 microNewtons. Indeed, the thrust at 60 Watts for all data overlaps pretty much perfectly for all data taken at 80 Watts. They can only claim a slope by turning the power down to 40 Watts, where they do consistently measure less thrust.
In other words, you apparently can’t get more than 120 microNewtons of thrust out of this machine. Why? The paper doesn’t speculate on that question at all. The more important point is that the individual uncertainties in their instrumentation don’t account for the variation in the thrust that they measure, which is a very strong hint that there is an uncontrolled experimental parameter playing havoc with their measurements.
Lee does a great job of pointing out how breathtakingly exciting this new EM-drive paper would be, if it were weird in the “good lord, we have discovered an entirely new physics” way, instead of weird in the “we have conspicuously omitted all the details which would satisfy skeptics” way. Not that anyone is accusing the EM-drive folks of deliberately misleading the public, but this paper seems to raise more questions about the measurement of the results than it does about the physics at play.
Postscript: Ethan Siegel in Forbes describes exactly what it would mean for physics if this is validated.