Because sometimes those choices have a large effect on the amount of work/mental tax in the future - for example, if you know that a new feature will have to in

Because sometimes those choices have a large effect on the amount of work/mental tax in the future - for example, if you know that a new feature will have to interop with yours in the near future and the cost is low to implement the right extensibility point now, it would be absolutely stupid not to - I would call that bad engineering that costs time & effort. Obviously one doesn't want to go down the rabbit hole too early, but the other extreme is just as bad.