Sometimes a tweet is more complete when a location is attached - I can just comment on something, rather than having to specify why the comment is relevant. Less typing, less fuss, cleaner words.
Often taking a picture makes it more relevant/meaningful when location data is attached.
When posting a Runkeeper update, it is often much better with the whole map visible to people (like when I cycled from SF to Palo Alto. Without a map, it's just a 56km cycling trip, with the map it's an adventure)
What I don't do, however, is make it possible to discern where exactly I live. Doing cluster analysis on all my location-aware updates everywhere will only get you to about a ~500m radius of my building. At least 300 people live in this radius.
Even if you pinpoint my building, you don't know my name. That leaves you with about 10 apartments to check. Without eyes-on-the-ground surveillance you aren't getting anything.
And if you're going to the trouble of eyes-on-the-ground surveillance none of my location-aware updates matter. You can see where I am.