Constructing Truth Using Algorithms


Session Title:

  • Science and Abstraction

Presentation Title:

  • Constructing Truth Using Algorithms



  • There are two kinds of truths; those of reasoning and those of fact. Truths of reasoning are necessary, and their opposite is impossible. And those of fact are contingent, and their opposite is possible.
    -Gottfried Wilhelm Von Leibniz, Monadology (1714)

    The belief that complex scientific data translates into effective communication only through collaboration with visual artists is not generally shared amongst scientists. Rather, science is frequently regard-ed as the method, amongst all others which attempt to ‘discover truth’, which has primacy and there-fore a pedantic approach to the visualisation of data is considered sufficient. However, scientific method would be seen to have a similar authority to practice if scientific ‘truths’ were demonstrated to be contingent. Then any claims of omnicompetence with respect to providing explanations for all phenomena. would be diminished and cooperation between practitioners of the two disciplines would need fewer rationalisations.

    Ironically, the recognition that algorithms are an integral part of the tools currently employed by science can contextualise scientific method as only one of many valid cultural endeavours which attempt to describe reality. It would be useful to analyse the various stages of ‘production’ (from data gathering through to visualisation of data) to determine in depth the algorithm’s role in codifying the data, simulations, analysis and theories, in order to demonstrate that the resultant ‘truths’ are contingent. This paper will rely on a simple presentation of visualised astronomical data, along with their models. to hint at their contingent character. Firstly, visualisations exist in 2-D image space.

    Therefore they are easy to view as cultural products and even lend themselves to being absorbed (perhaps inappropriately) into the domain of visual art images. Secondly, visualisation algorithms generate scientific content which is neither random nor arbitrary and yet which is not predicted by the unprocessed data or initial equations. In other words, the resultant object can not be determined by reasons alone. Therefore pursuing this aspect of the images demonstrates that scientific observations and interpretations reveal ‘contingent’ (not ‘absolute’) truths.