Well, in my opinion as well, the active page in the navigation should not be represented by an anchor, but by another HTML element that degrades to something logical. If the user agent does not support CSS, the navigation remains simply a set of links, giving no clue to where the reader is in the page architecture (unless breadcrums or a rootline are provided).

Usually, I would use strong instead of a to represent the current link in a navigation list. That way in a text-based browser, the user easilly distingueshes which list element he is currently visiting.

The usability aspects of this technique could possibly outweight the mentioned “frustration and confusion”. It might even be pointed out, that the user can distinguish between visual styles of different types of navigation items (currently viewed vs. the rest). That means that he/she can also benefit from knowing not to click on the currently visited list item. Check the example here.

Of course, HTML is way too flexible for us all to agree on what exactly is semantically correct and what not.. I believe the technique presented is perfectly legitimate (and quite ingenious), but I’m adding my opinion to the bunch as well, just to add some extra supporting arguments :-)