I haven’t seen any systematic, data-based investigation of how well cloth masks work against the Delta variant, and it is too early to expect it. Nonetheless I tried to do some simple mental modeling of my own.
We do know that the viral load from Delta is much higher. That could make masks less effective, because perhaps they cannot stop the spewer from spreading the virus so easily. (Oddly, you don’t see many people admitting such an effect might be possible, as this seems to be a politically incorrect idea to present.)
Yet there is a countervailing factor. The Delta variant spreads far more rapidly than “classic Covid.” So a given “small effect” of a mask is more important. It used to be that a mask (sometimes) stopped the spread to one person, who in turn might spread to 1.3 others. Now, if your mask stops the spread to that person, it might be stopping the subsequent spread to seven other people (we don’t know the exact number, but yes I have heard “seven” bandied around as a possibility).
To be clear, in this scenario masks are still less effective than before in preventing Covid spread. But the rate of return on wearing a mask, relative to no mask, can be higher. It is simply that the whole curve has shifted downward in a disadvantageous way. But if the mask has any effectiveness at all, that effectiveness is now magnified greatly.
Under some plausible numbers the protective potency of wearing a mask might be about five times higher (seven divided by recent non-Delta R, or something like that). Unless masks are five times (or more) less effective in stopping spread, masking could become more important rather than less important.
But the net effect could go either way.
That said, seeing other people masked should make you feel less secure than it used to. The new potential upside “rate of return effect” from masking is choking off the greater second order effects of the more rapid spread. The “are you going to get Covid from this particular masked person near you?” calculus seems to be decidedly worse than before.