Challenges Moving Forward
There are several challenges Google, or similar solution providers, will face before making this method of using user's mobile phones a source of quality real time traffic data on a large scale.
Mobile Phone Adoption
The larger the data set Google can draw from the more accurate results will be. The use of smartphones with GPS capability is certainly rapidly expanding, as is the use of its Maps application on smartphones. However, the large scale adoption of both cases is likely to be more distant than anyone reading this article anywhere near its publish date is likely to accept.
Distinquishing Motorists from Pedestrians
I suspect it will be hard to accurately distinguish people in cars on the road from those that are walking along side it, or perhaps using alternate forms of transportation such as a bike (or who knows, even the subway). I suspect Google's move to open this up to only major secondary roadways is based in part on that. Making sure they pick roads that have a degree of traffic on it sufficient to out-weigh any pedestrians on it is crucial to accurate results.
Processing Power and Algorithms
Probably Google's least worry, it certainly does take time to rally the appropriate hardware and engineers behind a solution to process so much real time data (another reason for keeping it to major roadways).