On the other hand, programmers are very good at wasting weeks creating incredible software architectures to try the latest library they read about or try a design pattern, or in general over engineer something so that it’s “better designed”, more generic, more “flexible”, and so on.

In other words, if a project manager sets an expectation that a task is a “task that takes two weeks”, the developer will somehow manage to use about two weeks to make it, that is find the “best way” to do it given that timeframe, where “best” is probably evaluated under some kind of metric which has nothing to do with product or user value.

Sometimes the question “explain me why it should take more than two days, when i know it takes two days to do just that in the context of a minimal MVP” goes a long way in making programmers focusing on delivering the maximum value for the product.

So I think there is actually some value in having a technical person continuously challenging developers in shortening their path to implement a feature.