Location-aware programs such as Foursquare, Gowalla and Dopplr allow users the ability to chart and share their past, present, and future locations. This is all good fun: letting your friends know you are at a cool bar in the Lower East Side or are about to go on a trip to San Francisco, these new social tools are beginning to redefine our bodies in space (c.f. Sousveillance).
[tl;dr summary: location aware software needs to stop treating locations as points, but rather points, areas, fields, lines and rays.]
I’m going to talk about Foursquare now to illustrate my critique, and in-turn use their nomenclature, but all services currently exhibit these fundamental problems.
First, some background.
In Foursquare, users can create Places which have a name and spatial location – usually an address. A common problem of duplicate Places (both accidental and malicious) is dealt with through the promotion of hard-core users into one of three classes of Super Users (disclosure: I’m a Super User Level 1, please give me a cookie) which can weed out the duplicates. When you check in with your mobile device, you are presented with a list of Places surrounding you. Generally your Place is in the first screen, but often it is not due to weak GPS signals or if the Place isn’t in the database.
The Problem: Accuracy versus Precision
There are two fundamental metrics we can use to evaluate the success of locating ourselves in space. One is accuracy, the other is precision. Accuracy deals with how close your measurement is to the true value; while precision indicates the repeatability of the measurement across a series of trials. The problem arises when the Place you are trying to check into doesn’t fit into the 1:1 modal of address:place. Currently Foursquare only accounts accuracy as a point in space, never a field, area, polyline or ray.
For example, let’s take LaGuardia Airport, a facility I am in and out of constantly. How do we define LaGuardia as a Place which is both accurate and precise?

We can certainly agree that LaGuardia Airport is located in the following Places:

• The Universe
• Sol System
• Earth
• North America
• The United States of America
• New York State
• New York City
• The Borough of Queens

We can all agree with the above list, which seems rather silly. However, where it gets fuzzy is when we try to increase our accuracy – the distance to the true Place LaGuardia Airport inhabits. LaGuardia isn’t just in one point in Space-Time since LaGuardia Airport comprises over 500 acres of land, parking garages, 5 terminals, concourse shops, and the gates, taxiways and runways. Again, there is no “one Place” La Guardia inhabits; rather than a point, La Guardia (and similar typologies) are areas.
This causes services such as Foursquare a great deal of issues. The result on Foursquare looks a little like this for LaGuardia and Grand Central Terminal respectively:

There are some interesting things to note here:
You can be on the other end of the LaGuardia airfield and not be able to check-in to LGA – which decreases the precision of the Place by users creating new LaGuardias which then have to be merged with the “one true LGA.” As you can see, LGA exists as both points in space, but also more accurately as an area.
The Russian Doll Problem
This brings up a related, but separate issue of User’s understanding of Space-Time. Users conception of Place is as varied as you would imagine, ranging from gamut from the general to the highly specific. To continue our LaGuardia Airport example, users have created Places for the following:

• LaGuardia Airport
• Individual terminals for LaGuardia Airport (LaGuardia Airport – Delta Terminal)
• Individual Gates (LGA C9)
• Individual concourse shops
• Individual flights (UA 690 Ord To Lga)

Because the precision of our instruments have improved many magnitudes, Users have created work-arounds creating even smaller levels of detail to accurately describe their location.
Possible Solutions
It would be remiss of me to just point out problems without putting forth any possible solutions. Suspending your judgement, here are some, “What if we…” situations:
What if we provide Places a Typology Option?
What if we stopped pretending that all places can be resolved with a single cartesian point? Let’s build in tools to our software where we can let Users begin to accurately define the space and place around them. This will undoubtably make the storage and retrieval more difficult, but the payback is greater than the pain: better recording of actual locations and how we humans actually use the city and world.
If Flickr can create whole maps from a field of data which is fairly correct, then we can build tools which will increase the overall accuracy.
What if we provide Users a Russian Doll option?
What if we allow Users to check in to Places which are both implicitly and explicitly higher in the location stack? In other words, going back to our La Guardia Example, if I check in to Gate C9 I should also be checked into the following locations:

• Central Terminal
• La Guardia Airport
• The Borough of Queens
• New York City
• New York State
• The United States of America
• North America
• Earth
• Sol System
• The Universe

Some tuning between both the implicit location – the Universe all the way to (say) The Borough of Queens – and the explicit – LaGuardia – must be experimented with so that Users won’t be overwhelmed by the data and visual noise. User testing should be also done on creating a clear definition and option so that Places retain their meaning.
What if we provide Users an Accuracy Slider option?
What if we provide Users an option of how accurate they are comfortable living at? Going back to LaGuardia, how about allowing a User to set a high and low threshold of accuracy so they would only see (say) LaGuardia Airport and The Borough of Queens? They might be OK with ambiguity, and the system should respect their careless ways – both in storage and retrieval.
Conclusion
Humans and computers understand location differently. It is the best interest of Users, and by extension location-based services, to understand and map the world correctly; and where possible create visualizations which combine an accurate understanding of the world which is usable by Humans. This is why cartography is so difficult and a balance between “correctness” of a map and how it is used is so difficult.