But consider actual masking practice, rather than the theoretical undertaking.
In practice, they are used in a way that legitimises closer contact (socialising, working, travelling, supporting) than would otherwise be appropriate.
At the point it matters, for example getting to see an elderly relative in hospital, masking isn’t effective enough that it should be counted. Yet it is.
If I was to make a bet, it would be that if masks are eventually identified as being useful, the confounding factor is exactly that they make us increase contact so much that it offsets their entire benefit.
You are seamlessly slipping from a claim of practice back to just a theory.
Those are fine things to speculate about, and lead to some interesting conversations, but these are not practical claims.
One could further speculate that the emotional needs of a deeply social species would have still overridden whatever guidance was given, and in a no-mask world the primary difference might be that some people mentally adjust for a greater degree of perceived risk (and still carry out their planned visit) than they otherwise would have.
If your bet is correct, I think it's necessary to consider more than one dimension of benefit when evaluating the overall scenario, e.g. the benefit of seeing one's loved ones one last time (and similar scenarios) needs to be considered when evaluating which policy would have been better (assuming the impact of a masking policy is roughly equal to the impact of a non-masking policy).
If we learned tomorrow through some better study with more certain data that masks reduced deaths by X percent, I don’t think you could fairly claim that during the period of time without that information, behaving as if masks did nothing was the same as behaving as if they are effective.
The cost of betting wrong is asymmetrical in the two scenarios.