[extrapickles]

There is a list of improvements in the article under "Latest Paper".

Its mainly the need to carefully characterize and document all the test equipment and EM Drive support equipment to make sure its not caused by the test setup. This would involve testing with a dummy load so things like interactions with chamber wall can be ruled out or if there are, the effect can be mitigated or accounted for.

Basically they proved their test setup produces thrust, but haven't proven the thrust was a direct result of the EM Drive. They need to more rigor to prove its only the EM Drive and not a setup error (eg: something like the FTL neutrinos turning out to be a poorly tightened connector).