Epidemiology during a pandemic is about one part science to three parts PR. Sending a consistent message to the public and getting them to change their behaviour at all is more important than getting the details right.

In the case of masks, the WHO decided early on that masks for the public were of limited use in slowing coronavirus spread: they'd stop some infections but could be detrimental if misused and could threaten the supply to medical professionals. This was a reasonable conclusion at the time. To communicate that to the public, the party line was: masks no good. Many national organisations took their cue from the WHO.

Now the risk/reward tradeoff has shifted: better research shows that masks, even ersatz ones, are good and the supply lines for professionals have improved. So the message has to change: masks good.

The actual scientists know that it's not black and white, but pretending it is is a useful tool for health organisations.

At some point I expect we'll see a flip-flop on the party line that "there is no evidence that covid-19 infection confers immunity against future infection". It's nonsense - there are hundreds of years of circumstantial evidence - but it's just about defensible in that there haven't been any sufficiently powerful blinded trials to directly demonstrate it specifically for Covid-19. Once it's deemed to be in the public good, they'll accept some study showing exposure does indeed confer immunity.