I've meaning to explicitly point out one of Ullman's most annoying habits and this is a good opportunity to do it.
As usual in that video piece he willfully cherry-picks papers that are variously discredited, useless or superseded, but he does something else.
Even if the results of his favourite papers were entirely correct the effects that they find are utterly trivial.
He makes the schoolboy error of muddling statistical significance with biological/clinical significance, but he compounds this with a neat rhetorical trick: a careful elision, where he concentrates on how
important the disease is but swiftly by-passes how small is the effect he can claim for homeopathy even at the most generous possible interpretation of the data. He is most guilty of this when describing that useless Frass paper from
Chest. He is very keen to tell us how important a disease is COPD, which is true and no reasonable sceptical rational doctor could disagree, but he allows his assertion of the disease's importance to imply an importance for the effect allegedly found for homeopathy, which is so disproportionate as to be close to deceptive.
The effect of this is to claim that any positive result, however minor, legitimises a naively binary interpretation: works vs doesn't work. So, a trivial positive test on any single parameter of a poor trial is declared to show that homeopathy "WORKS!". Once he has cherry-picked a result sitting on the right side of that binary alternative it can now be promoted publicly without any connection to the underlying truth.
It's quite cunning and perfectly tuned to misleading the credulous and ignorant.