Perhaps its just that, as software developers, we eschew any form of design that would lead to a maintainable monolithic system (The No Big Upfront Design Movement may have caused us to throw the baby out with the bath water). Maybe we just don't yet have the tools and theory to yet to put together a complex system made up of many individual agents in any easy-to-do way (e.g. Microkernels vs Monolithic kernels)

Look, we live in a era where the fastest time to market is always going to be the way to go. Microservices are nice but they slow development down a great deal.

What we need is an easier, less subjective way to build software. I think DataFlow programming will become more popular since it is easy, scales well, and applicable to more domains than many would think.

A monolithic dataflow application has many of the advantages of micro-services and monoliths alike.

I also think the industry should probably start to shy away from OOP (especially since industry totally dumped OOD). If you go on github and find a random C program, then do the same for a random C++ program - I would bet you can wrap you head around the C far before you can even begin to understand the C++. How people can revolt against microservices and yet not question the same phenomena with respect to basic SP vs OOP is again baffling to me.

I think microservice adoption is a heavy-handed approach to modularization. I very much like Jackson Strucutred Programming, Dataflow Programming etc. Dataflow is actually applicable to many more domains than some think and are about as understandable and scales about as well, if not better than microservices.