Agile team – empirical vs scientific methods

As we all know that human behavior does not follow strict rules. At most some statistical model can be used to model approximate human behavior. Complexity increases exponentially as more people enter into and environment has many dynamic known and unknown dimensions. On the basis of statistical model if someone attempts to predict behavior of a team member, naturally lot of probability come into play. Instead of such a scientific method, empirical analysis and model fits better with human behavior due to cost and time factors.

For any successful team (scum or not scrum), I prefer to use analogy of navigating a yacht which requires continuous adjustment to reach a destination even in calm water.

In any sufficiently complex software engagement diverse skills are required. This requires team to be consisting of generalists and specialists. This specialization breaks scrum on surface which assumes that team members have broad skill set. This may not be true when team gets formed. Team members may not have T shape skill on day one but certainly can develop (provided team sticks together long enough). If work required high level of specialization, Spotify like model (guild) may be helpful.