Yep, totally agree. A lot of founders don't do this as they're too attached to the idea or too worried about criticism. It's a tough learning process to get over the negative feedback.
This seems quite hyped advice in SV recently, but validation can be difficult for certain classes of businesses. For example, I recently spoke to a US startup founder who had failed in what is roughly our segment, burning $300k. His advice: validate, validate, validate! In our area, however, it is essentially impossible to do so without similar to launch expenses, so we are just running with a regular development trajectory (in a far cheaper and higher demand market with other positive attributes) and the broad assumption that people want to eat food - http://8-food.com/
Everything can be iterated. The degree to which this iteration can be accomplished is really highly dependent on the complexity of your product, but, in terms of iteration complexity (and highly simplified, dropping lots of intermediate points, etc):
At the very least iterate on things like the user stories to make sure your customers can see your product fitting in to their day to day life. Without any form of iteration, you're leaving the most important thing -- product market fit -- somewhat to chance.
You're right, but as with anything the management-level questions are (1) risk versus reward; (2) (finite) resource invested versus value derived (versus other options for those resources).
In our case, we can iterate easily based upon real market information post-launch. It's literally illegal to do market research otherwise, so we can't really get a foot in the door hyper-analyzing potential segments ahead of schedule.
