Post date: Mar 27, 2014 8:14:21 PM
Aaron started by talking about how large an effort (number of students and stakeholders) the work was. Something to keep in mind with respect to such a big data collection effort. He then introduced some of the issues that are important to addressing accessibility to transit. Transit accessibility affects social isolation, and the ability to work of people with disabilities. Aaron talked about the importance of universal design to increasing accessibility: For example a low floor bus is an improvement over having a lift (which may possibly break).
He then discussed social computing for tracking and improving city services. He talked about the impact of wall of shame vs more collaborative approaches (where multiple stakeholders are involved and feedback is supported).
He then introduced the complexities of transit. The difficulty of getting data, the routes that drivers take, the changes in drivers and on-time goals for different seasons, the difficulty of separating buses that travel on the same route. These sorts of issues affect data quality, one of the themes of this course. AVL systems can help with some of this by tracking buses in real time and predicting when they will arrive at each stop.
Tiramisu, a social computing application that supports bus tracking, was their solution. It shows 'real-time' data that people have just entered, 'historic' data for things that have sufficient historical data to be estimated on that basis, and 'scheduled' data from the transit authority when the other types of data are not available. It also allows users to report problems.
In terms of application acceptance, he discussed the importance of minimizing power use and data upload, the fact that data is collected after the value is given to users, and the importance of having the data right away (for others trying to get on this bus right now). In addition the data is only useful for that route (and time of day) and even historically is only useful for at most 3 months.
Initially, a closed trial with 28 users was conducted; then an open trial now spreading to 3 cities. They tried to share problem reports with the transit agency. However union rules meant that the transit agency could not act on the data because of existing policies -- the data must be collected by the transit agency in Pittsburgh. Learning about policies is critical to successful data collection.
The huge amount of interesting publications associated with Tiramisu suggests many opportunities for future reading and thought. Thanks Aaron for a fascinating lecture.