The monolithic codebases are easier to crawl for any problem that can't be conveniently isolated to a single microservice.
The same would be true in a monolith: The context to understand what's happening would be contained to a few files.
When the work starts crossing through domains and potentially requiring insight into how other pieces work, fail, scale, etc. then the microservice model blows up complexity faster than anything, even if you have the API documented.
Maybe you're right but I'm aghast at how much of engineering over the last 15 years has been breaking up working monoliths to fit better within the budget of an external provider (first it was AWS). Those prices can change.
There are good reasons to use microservices but so often they're used for the wrong reasons.
