TREDS - the turtle research database Pacific conservation programmes run themselves

TREDS, the Turtle Research and Monitoring Database System, is where researchers across the Pacific record tagging, nesting, strandings and beach surveys, and follow an individual turtle across years and across islands. I developed and extended it under two contracts with the Western Pacific Regional Fishery Management Council; the Secretariat of the Pacific Regional Environment Programme and the Marine Research Foundation were the technical monitors, not the contracting parties.

The work that decides this kind of system is not the schema, it is where the data is entered. A turtle is tagged on a beach at night, on an island with no reliable connection to anything. So every programme runs its own copy, locally, and the design question is how a national conservation office with no IT department installs, runs and backs up a research database on its own. I restructured what was there into separate business-logic and data-access layers over the Access data, and packaged it so that installing it is not a project.

TREDS has no web front end and I did not build one. The ASP.NET work on this project was a separate site, which distributed the release and managed who was entitled to download it. I mention it because the two are easy to collapse into one claim, and they are different systems with different users.

The number I can stand behind is from my own instrumentation on that download site: 170 downloads in the thirteen months after the 15 February 2009 release. It measures distribution, not use. A download is not an installation and an installation is not a record entered, and I had no telemetry from inside the application, by design - it runs offline in places where sending anything home would have been the wrong default. For a system whose whole point is to work with nothing behind it, that is the trade I would make again.

What I inherited was half-finished, which is the ordinary starting condition for scientific software: written by whoever was available, against requirements that arrived while it was being written. The restructuring mattered more than any feature I added, because it is what let the next person change one layer without breaking the other two.