[diracs_stache]

Not sure what the exact context the blog refers to (they are a scientist turned software engineer?, their field is data/software intensive and see this as an improvement area?). Our team does engineering test and evaluation that include aspects of R&D and struggle with this. Scientists have academic approaches (must have sufficient sampling/statistical significance, sometimes delaying findings/reports checking additional aspects when a sanity check will suffice) but it does give rigor. On the software side, we definitely have spaghetti code, tools that rely on some file sitting in someone's H-drive, and plug-ins built by someone's old collaborator with little documentation. This is juxtaposed by the PM types that must be agile (fine when tasks are understood and defined enough to go nicely into a sprint, not always the case). Better communication on both sides would probably alleviate some of this but that is the great challenge in any group.