Spacetime and code-time: Astronomical algorithms evolving through dimensions of temporality and materiality as scientists change technological practices

• In Event: Making Algorithms: Inscriptions, Benchmarks and Computers

Abstract

The role of the algorithm in knowledge creation in computationally-intensive science domains has been front and center for decades. Current trends towards making science more open can involve scientists sharing their algorithms in online code repositories. In being made publically available, the inner workings are revealed, and the ‘black box’ paradigm falls away, leaving the algorithm in a new, more exposed state while functioning as an epistemic object intersecting with a technical object (Rheinberger). The algorithm in its new shared state and may be copied (forked), altered and examined by an increasing number of people. Our case study concerns an astronomy research group with a heavy reliance on algorithms to refine and process data gleaned from telescopes. The research groups’ algorithms lie on many axes of temporality: technological advances in telescopes create data that forces the algorithms to be altered, and the actual data itself is of a temporal quality of moving objects in space throughout specific time periods (Jackson, 2011). We study the relationship the researchers have with their algorithms, most likely created years ago by a long-gone team member, with only an oral trace of its origins. The researchers use entity-based language to describe their algorithms, for example, they behave in various ways, such as break, make errors, produce inexplicable results, and yield new observations. As Rheinberger observed, the algorithms function at the margin of obsolescence and disrepair, and new forms of sharing are emerging from altered processes (1997, Jackson, 2014).

Author

• Bernadette Randles, UCLA