I think that the inconsistency is that subject to very broad and generic conditions (most notably that certain quantities which are naturally positive are indeed positive) that all vacua in the string landscape are necessarily anti-desitter, which is inconsistent with the observation that the actual universe is desitter.
So, to get the observed universe, you have to mess with what is inside it (basically the dark sector) to turn the natural topology of the string landscape into something that resembles the real universe.
Previous efforts to look at the landscape had determined that there was lots of swampland on an ad hoc basis, but failed to identify as the new conjectures do, overarching generic conditions that cause vacua to end up in swampland.
The optimists think that may be a small, finite number of exceptions to this general rule, narrowing string theory down to a handful a vacua that can be compared to reality one by one, rather than an intractably large landscape, but most commentators appear to conclude that all string vacua are in swampland thereby conceivably disproving all of the M-theory as a representation of physical reality, which would be a gigantic bummer to the entire string theory community.
But, I welcome the comments of anyone who understood it differently.