We're using XSLT because we don't want our core code to have to output a bunch of different formats. With XSLT, we can output one format and allow XSLT to handle transformation of the XML to whatever output format is desired. Unfortunately, I've been using the tried and fuzzily true "hacker's learning methodology" of just using Google and books to apply a solution to a problem as it arises.
A couple of times this has led to me discovering that an applied solution was not the best solution -- just as novices learn with any language they stick with. So do I take the time to really learn XSLT or do I just finish up the task at hand and hope that I've learned enough to just Get Things Done? In other words, I face a recursive dilemma: I need to learn enough XSLT to determine if it's worthwhile to learn enough XSLT.