No different than any other part of the stack...everything that should be simple has become pointlessly complex. How many people here are evaluating Kubernetes for trivial services that could be easily addressed with twenty-year-old scaling models?
Do you have a particular example of something that should be simple but has become pointlessly complex? In my experience when people complain about complexity they are often complaining about a change from the thing they know to something they don't. On objective examination it can be the case that the old way wasn't simple either, because the complexity is inherent. Typically we move it around, put abstractions around it, but don't really change it fundamentally.
Re: when people complain about complexity they are often complaining about a change from the thing they know to something they don't.
If something is truly better, then it should be easy to articulate why in a concrete manner, such as exploring total lines of code that need to be changed per a change request scenario that's realistic for your particular org (and not a cheesy textbook example).
It's also good to see it actually working in similar organizations. Often people copy from dissimilar organizations and make a mess. Microservices are like this. They don't fit many shops or orgs; but because Netflix finds them useful, everybody has to copy.
Often "new things" are passing fads. It's good to vet things for actual productivity improvement instead of just using buzzwords such as "separation of concerns". Otherwise, the org will have separation-of-wallet.
One area of complexity I constantly rail against is dependency injection framworks or IoC... In most cases you can use what's built in, or don't need it.
Layers of crap in "Enterprise" .net and java projects. If you actually have lots of complexity in the app, with lots of test coverage, or multiple target libraries, sure... but most don't have that.
Aside from that, it really is resistance to change.
