This paper has developed a perspective on how to model serendipity in a computational context. We advanced 13 criteria which were developed based on review of the prior literature on serendipitous discovery. We piloted these criteria as an evaluation framework by examining 14 patterns of serendipity that had been previously identified by van Andel. We found our criteria to be well represented, but not uniformly present, and the exceptions are interesting; for instance, we observed that A good story is liable to change. We then advanced a unified approach to modelling serendipity grounded in Deleuze’s philosophy of difference, with a debt to the dynamical interpretation of this work due to DeLanda, drawing as well on the technical strategies employed by the interdisciplinary design pattern community. This approach was developed further into a proposed experimental platform for doing collaborative research in computational creativity. We showed how four of van Andel’s patterns could be relevant in this setting, and introduced a new pattern template oriented toward facilitating the encounter of computational systems. Finally, we surveyed related work, and summarised how computational serendipity can contribute to the field of computational creativity. We suggest that more attention should be focused on the role of creativity in problem-setting, and on creative computer systems as a key stakeholder group in computational creativity.