« Mobile is redefining everything we know about information | Main | Next generation of LBS growth will come from geoweb feeds »
February 06, 2009
Latitude reality check
OK. So Google have released Latitude , which is a kind of real-time mobile location sharing via Google Maps). There has been huge coverage, some seeing it as a new threat to privacy ("Google puts spy in your pocket" London's Metro Newspaper, 5th February 2009), but by others as new evidence of the productive linkage of location and social networking (see Andrew Grill's London Calling, for example).
I joined up through my Google account and followed myself as Google tracked me in real time across London. A work colleague texted me with incredulity as he watched me move across the map.
So. It works (kind of, see below), it's fun and it fits with the strategy to enhance Google Maps, but is it an advance for those already using tracker services like Loopt or Brightkite? In short, no. What Google have done is to enhance their platform with an additional feature as John Spinney has pointed out. Google's brand will ensure that it gets distribution and use, and this will change the landscape for LBS in various subtle ways.
Let's look closer at what Latitude does and how it performs. Latitude uses the existing Google Maps for Mobile 'My Location' feature in the Google Gears API which uses Google's growing database of cellID locations to define your location using trilateration. This kind of positioning is accurate to 100m in certain city centre locations, and to 500-1000m in the suburbs, and has been around for almost a decade. As I followed myself using Latitude on my train journey home, at first in central London I was snapped accurately to the closely spaced stations, then as we moved into the suburbs my location jumped from place to place on both sides of the railway track, many of which I have never been to in my life.
As my phone has GPS, I can switch on GPS and see 'My Location' updated to the nearest few metres... and I saw Latitude map my GPS track as I walked home from the railway station. 'My Location' also uses WiFi positioning when the device can detect WiFi hotspots whose locations have been mapped by Google or its partners. Hence, when I switched off GPS and switched on my phone's WiFi radio, Latitude quickly zoomed in on my house: I assume my home hotspot has been mapped.
So, although Latitude has now brought positioning to a mass market, the underlying accuracy fundamentals have not been altered... and you still don't know what kind of positioning you are getting or how it is performing. So, I look at Andrew Grill's Latitude location as I write this and I see his marker sitting on a road in West London. Is this a cellID estimate that is bouncing around depending on the room in the house he is in (and different cell base stations are serving his mobile device), or is it a wifi position or a sample from a GPS track? Over time by looking at the position you can start to guess which method is being used, but without more information from Google it it is hard to know how the positioning is performing. This problem of positioning quality for LBS is something that Placr is working on in the EU funded MetaPos project (see my recent presentation at GRACE).
This means that although Latitude 'works', the lack of positional quality information means that LBS developers and advertisers cannot yet bank on the positional streams to offer services. It also means that (by accident) Latitude doesn't do a very good job of invading your privacy, even assuming you have got consent to track someone...
Jonathan
Posted by raper at February 6, 2009 10:52 AM