You might be thinking of this [1] paper? The core idea being if your problem space lends itself to monotonicity (see the paper for a more precise definition than I can give), then you can build a globally consistent database (around a CRDT) where the end-user doesn't need to concern themselves with inconsistent states while consistency is reached.
I wasn't clear - by "the paper" I meant the original CRDT paper. I read it, thought I understood it on some level, but had not drawn the dots between the theory and real world problems.
Regardless, thanks very much for linking the paper! Right up my alley.
[1] https://arxiv.org/pdf/1901.01930.pdf