The team doesn't have to be great at pointing. That is the whole point of this model. A small story is 1 point for the story and 1 point for testing. Medium is 2 points for coding and 1-3 for testing. Anything else should probably be cut up into smaller stories. The idea is to have more stories... this way the work can be divided up more easily and smaller units of work are easier and more fun to accomplish.

I do agree that the PM has to be excellent at writing stories and the team has to be good at pushing back on the PM to say that a story is too big and to break it up appropriately into smaller stories.

Two week sprints are awful. Nobody should ever be in that. The problem is that the team works for 2 weeks and then after those two weeks, everyone stands up and shows what they did. If you don't hit the mark, then it is another two weeks. Now a single feature can be held up for 4+ weeks! Or even worse, you cut out things like testing to make the 2 week sprint.

The pivotal model of making smaller features and factoring in testing into the pointing ensures that nobody is working alone for 2 weeks without any oversight.

This is why velocity is a 3 week average of points and can be overridden in the tool to show people who are out on vacation/sick. The tool dynamically updates to show what stories make it into an iteration as well.

The whole idea is not to force people into a box. It is to just give PM's a way to plan the future. Obviously, things can change so it isn't some forceful thing at all.

Don't knock it until you've tried it. I had an opinion like you and then I went and worked in this system and it totally changed my mind.

I tend to agree. I have found the number of stories to have som utility as a metric, though.