Draft Blog Post

Observing a PANDA in its natural habitat

We've been developing the PANDA project using agile practices. Two keys to agile development are stakeholder involvement and frequent releases.

Stakeholder involvement is crucial: with close contact with active users, we can get frequent feedback on whether we're building the right features and building them correctly. Earlier models of software development were modeled on large-scale engineering projects, like building skyscrapers and jetliners. For that kind of project, the cost of failure is too high to risk, but for most software applications, bugs are inconveniences, not crises. This is especially the case if the development process also includes frequent updates which promptly provide bug fixes and add important missing features. Because of this, we've been eager to reach the point where the application was ready to offer to reporters in working newsrooms.

Coincident with our recent beta release, we went to the NICAR conference in St. Louis, where we helped almost two dozen PANDA servers enter the world in less than an hour. We took this as a good sign that we'd succeeded at making PANDA easy to install. These early users flushed out a couple of bugs--thank you for making PANDA better!

While we were at NICAR, one awesome new feature was eagerly discussed. Besides fixing bugs, a recent development cycle focused on delivering typed searches. Users wanted to be able to search for specific values and ranges, especially in number and date columns.

Today we are releasing an updated PANDA which fixes those bugs and delivers typed search. We've made a new installer, as well an easy-to-run upgrade process for existing PANDA installations.

While we're looking forward to learning more from people who've installed PANDA, we've learned quite a bit from the Chicago Tribune's roll out of the application. We had a series of meetings with editors and other staff to show them how to use the application. In the course of demonstrating and discussing PANDA, our insights and our community's reactions have lead us to update the prospective features list.

For example, as we explain to staff how we want them to use PANDA, we are stressing the importance of replacing the default dataset title, which is the original filename, with something more clear. We are also urging uploaders to diligently use the description field to explain the source of the data, so that other PANDA users will be able to judge how they use that data in reporting and writing. In the course of these discussions, we realize that if this is so important, PANDA should make it easier for users to do the "right" thing. We'll evaluate the upload process and adjust the workflow to promote best practices.

The PANDA board will meet in-person soon to for an intensive session of prioritizing features. If you're using PANDA already, we would love to hear more from you to help us focus development for the remaining months of our grant. You can use the comments here, or feel free to join us on the user mailing list.