TitleFrom one to many: generalisation and evidence in failure analysis.
Publication TypeJournal Article
Year of PublicationSubmitted
AuthorsDe Bal, I

In this paper, I use cases and reasoning from failure analysis (a part of engineering
science which deals with artefact failure and its causes) to draw attention to a relatively
unstudied problem of knowledge generalisation: when we are focusing on creating new
things; designing new artefacts and technologies. Using three cases from failure
analysis practice, I present a two-fold mechanism-based procedure to determine when
generalisations to non-existing artefacts are warranted. This procedure builds on (1)
Cartwright's notion of capacities (2) literature on mechanisms and (3) Steel's
comparative process tracing, developed for the biomedical sciences. I will show that,
while they provide guidance, these literatures and concepts are not enough to grasp
how we use information from current artefacts and failures to create new things - we
will need a lot more specific information and adequate ways to present it. The account
developed in this paper is relevant for both philosophers and failure analysts. For
philosophers, it can provide input for a theory of evidence. For failure analysts, it allows
them to present stronger arguments for their recommendations by making the required
evidence explicit. My account can furthermore provide inspiration for similar inferences
in other innovation contexts such as pharmacology.

Citation KeyDeBal2017evid
Download PDF (Author PDF)
PDF author (public):