I saw lots of churn working on microservices that were pre-production. When it’s like this, things are more tightly coupled than the microservice concept would

I saw lots of churn working on microservices that were pre-production. When it’s like this, things are more tightly coupled than the microservice concept would have you believe and that causes additional work. Instead of writing a new function at a higher version, you had to go change existing ones - pretty much the same workflow as a monolith but now in separate code bases. And there wasn’t a need for any of these microservices to go to production before the front end product, so we couldn’t start incrementing the versioning for the API endpoints to avoid changing existing functions. A monolith almost doesn’t need API versioning for itself (usually libraries do that), but it’s effectively a version 1.0 contract if translated to microservices.