EuroSTAR software testing conference

On Tuesday 5 November Rob Evans was at the EuroSTAR software testing conference in Gothenburg. The EuroSTAR conference bills itself as “largest and most prestigious event on the software testing calendar” and Rob was there to talk about work done by Harry Collins and himself on the nature of expertise and, in particular, the importance of tacit knowledge.

The morning session was devoted to a tutorial organised with James Bach that explored the implications of the distinction between mimeomorphic and polimorphic actions first proposed by Harry Collins and Martin Kusch in their book The Shape of Actions. The idea here is that polimorphic actions depend on context-specific forms of collective tacit knowledge and hence defy automation. In contrast, mimeomorphic actions are ones that do lend themselves to mechanisation. This, in turn, maps on to different ways of approaching software testing and, in particular, the importance of context-specific testing that is sensitive to the ways in which users might interact with the software.

This tutorial session was also important as it contained the first mass-testing of the Masquerade app. We used the app to explore the expertise associated with software testing by asking trios of players to identify the least experienced tester. Although the numbers were small (roughly 50 games in total) we did find that the more experienced testers were better at identifying the less experienced player. This fits with the idea of ‘downward discrimination’, in that we would expect experienced testers to be better at spotting mistakes as they are judging the performance of those less expert than them. In contrast, less experienced testers were effectively judging ‘upwards’, putting them in the same position as the lay public choosing between competing scientists.

In the afternoon, Rob gave a keynote lecture to the conference. Harry was not able to be there but the topic again drew heavily on his work, in this case the idea of machines as ‘social prostheses’.  Here the argument is that machines, including computers, can ‘fit into’ the social body in much the same way that a medical prosthesis such as an artificial limb fits into the human body: the prosthesis has to reproduce the function of the thing it replaces but may do so in a different way and well almost always require some adaptation by the host body. In the case of computers, the usefulness of a machine thus depends both on how reliably the internal mechanism works and on how much change its human users are required to undertake in order to prepare and process the data it needs. As with the morning tutorial session, the distinction between mimeomorphic and polymorphic actions is crucial.

The conference was a hugely enjoyable day, with lots of interesting conversations. It was also extremely well organised, so a big thank you to Michael Bolton for the invitation. It would have been great to have stayed for longer but fieldwork commitments in Trondheim meant that a flying visit was all that could be arranged.


No comments.

Leave a Reply

Your email address will not be published.