Wait, no the hell! There have been a finite number of jobs advertised, right? You must be able to average *those*.

It'd be stunningly misleading to only average those for which you know the correct value, and if you set the "value" of those still open to "how long they've been open so far", that would be similarly very misleading (since that set will tend, for some distributions, to be skewed towards those that will remain open for a very long time indeed).