A recent article on Techcrunch brings up the issue and you can read more about it here: http://techcrunch.com/2010/04/17/open-database-places/
With the proliferation of location and how many apps and products are being formed around location. A big part of that is points of interest data. Applications like Foursquare, Gowalla, … all require information for places around you. Now you would think this would be a lot easier to get and you can use Google to search places by category or by some query term to return places around you. You could also license data from places like Localeze or InfoUSA. The problem is see with the issue as it stands today comes down to 2 things:
1) Categorization
2) Uniquely identify a place
Categorization
The problem here is the ability to get a good sense of what category a place belongs to whether you search with a query or filter by category. In addition we would need the ability assign a place to multiple categories
Uniquely identify a place
This is a bigger issue and in the state of the world today is hard to deal with because each provider either does not have a proper way to uniquely identify a place or you cannot identify a place across services. For example: if I added McDonald’s in Cupertino, CA how can I be sure I get the right one when another user does a similar search. Do I hash the name, address or name, and lat/lon?? What if you switch data providers how do you move your venues over? To me this is the biggest problem that needs to be addressed and would solved with a standardization on an open database of places.