Yes, 'maybe not' is very relevant to this discussion, but few people seem to agree with my understanding of what he says about the right solution:
Optionality doesn't fit in the type system / schema, because it's context dependent. For some functions, one subset of the data is needed, for others a different subset. Trying to mash it into the type system / schema is just fundamentally misguided.
Yes, he's rather explicit in saying Maybe is a poor tool.
I'll have to watch the talk a second time to be sure, but I'm not sure he proposes any solution at the level of type systems. Not using Maybe or using Union is not what he is advocating.
For him (and me too) types are the wrong thing to put data in because, among other things, it forces you back into PLOP. His point is to remove entirely the need to fill slots with nothing.
Obviously the talk is more about specs than types. While tactfully avoiding the debate around types, he's still starting the talk with types to help those that are only there to decomplect their thinking.
Optionality doesn't fit in the type system / schema, because it's context dependent. For some functions, one subset of the data is needed, for others a different subset. Trying to mash it into the type system / schema is just fundamentally misguided.