At the heart of this issue lies a fundamental tension: do we want highly personalized questions, or questions with broad applicability? Each has its merits.
The primary purpose of the Stack Exchange platform is to collect useful information that will be valuable to others. But let's face it: most programming problems are quite mundane, highly specific, and unlikely to ever benefit anyone else unless it is a commonly encountered problem. The nature of the programming profession guarantees that many of these questions are going to be unique to the asker's particular situation (what we call highly localized questions).
Unfortunately, a significant portion of the user community (including myself) was using the concept of "Too Localized" to close questions that probably shouldn't have been closed on that basis. Too Localized was taken to mean "This question is unlikely to be of use to anyone else," when it should have been interpreted as "this question involves a vanishingly small geographic area, period of time, or programming situation." "Where's my Typo" questions certainly fall into this category, but troubleshooting questions do not.
Questions that are Too Localized under the "traditional" view can often be closed for other reasons such as "unclear what you are asking," or "must demonstrate a minimal understanding of the problem being solved."