If the problem is focusing on short-term writing which becomes worthless quickly, then simply hiding or showing dates shouldn't much affect how long readers stay on the page: most short-term stuff shows its colors very quickly. (How many sentences does it take to figure out you're not interested in a rant about John Kerry from 2004?)
I think McKenzie's argument is that using a date can turn long content into short content, which many people do on accident, and while he doesn't quantify it (which would be the value of A/B testing) I think he has enough evidence to establish the direction of the effect. Not using a date is obviously not sufficient to turn short content into long content, but I do think it may be helpful at getting one into the right state of mind, as it focuses the attention on sorting things by content rather than time. (Imagine trying to find all of Robin Hanson's writ...
Another month has passed and here is a new rationality quotes thread. The usual rules are: