RESUMEN

The well-known black-box model of software development outsourcing is typically effective. The approach assumes that the vendor can successfully solve a client organization's business problem without either organization having to deeply understand the other’s domain. This article reports on a field study of 209 global projects that explored the black-box approach’s limits as well as potential solutions to those limits. The study’s key finding is that the black-box approach usually works well in routine projects but fails in projects involving novelty. Novel projects require careful deviations from the black-box model depending on novelty type. Conceptually novel projects require the vendor to deeply understand the client’s business whereas projects involving novel development processes require deeper technical expertise on the client’s part. The article also presents insights into the effectiveness of capability maturity, intense architecture design effort, and development coordination tools. A knowledge congruence framework is offered to apply the findings to software development practice.

If the -adic cohomology of a projective smooth variety, defined over a local field with finite residue field , is supported in codimension , then every model over the ring of integers of has a -rational point. For a -adic field, this is proved in (Esnault, 2007, Theorem 1.1). If the model is regular, one has a congruence modulo for the number of -rational points (Esnault, 2006, Theorem 1.1). The congruence is violated if one drops the regularity assumption.