Sunday, June 3, 2012

pull; update


Les Arènes de Lutèce (the Roman arena of Paris) is not much of an archaeological site, but it's a lovely French park, surprisingly peaceful despite its location in the bustling 5e arrondissement.   A group of eight or ten men and women, mostly of a certain age, is silently practicing Tai Chi behind me;  opposite us, French school children are clambering over every visible surface and cheerfully pushing, shouting and generally attempting to terrorize each other.  This is not Worcester, Massachusetts.

When I last sat here to soak in the sun more than 20 years ago, the scene was visually and aurally identical, but today I have in my laptop a computer that weighs less than a kilo, connected to the internet because public parks in Paris give you two hours of free wifi.  The seven busy researchers in the St. Isidore research lab at Holy Cross all use mercurial for version control of their work, so I've run hg pull; hg update, and have seen every change they've committed in the day or so since I last had time to look.

Juxtaposing geographic distance with the immediacy of electronic contact may seem like a pretty tired cliché in 2012, but working step-by-step through the progress of a team thousands of kilometers away makes me realize how little we've thought about a fundamental question:  how do we make our research reproducible?   Version control systems like mercurial or git are one important part of the technological puzzle, but they don't by themselves tell us how to organize our material or working practice so that others can easily replicate our work as fully automatically as possible.

I'm introducing a new tag "RR" for the theme of "reproducible research" since I think that is arguably the biggest overarching challenge of architectural design in digital scholarship today.



No comments: