You want to apply Bayesian reasoning to this. Conservation of momentum is a very well-tested, long-established physical law. Our priors going into this experi

You want to apply Bayesian reasoning to this. Conservation of momentum is a very well-tested, long-established physical law. Our priors going into this experiment should be "it is impossible to produce thrust without throwing mass or energy out the back of your thruster!" at something like 99.99999% So now this experiment has happened! That should indeed reduce our belief in the above. But there are other explanations, like, "the experimenters screwed up." So previously our beliefs were: There is no way to get thrust without blowing mass or energy out the back end of a thruster (and if you do it as energy, it's a completely enormous ridiculous amount of energy): 99.99999% There is a free lunch: 0.00001% Now we think: There is no way to get thrust without blowing mass or energy out the back end of a thruster (and if you do it as energy, it's a completely enormous ridiculous amount of energy) (and those experimenters screwed up): 99.99998% There is a free lunch: 0.00002%