Perturbations of interdependent (science) teams reveal perfect and dysfunctional teams

Lawless B. Perturbations of interdependent (science) teams reveal perfect and dysfunctional teams. Oral presentation at 2017 SciTS Conference. Clearwater Beach, FL. Jun 14, 2017. Team Formation And Cohesion. Online at: http://www.scienceofteamscience.org/2017-agenda.

Cooke & Hilton (2015) reported interdependence statistically associated with the best science teams, but not theoretically, with an unknown effect on team size. With our goal a metrics of team performance derived from a theory of interdependence, we have established that a team’s performance cannot be determined by its membership (Lawless, 2017); that scientifically evaluating teams with self-reports or interviews cannot perfect or fix dysfunctional teams, or predict team performance (e.g., winners of sport competitions, political contests, or jury decisions); but that competitions (perturbations) among teams improves science and social welfare. Team size is a fundamental barrier to a complete theory of teams: How to aggregate the contributions of team members? Cummings (2015) found that the worst performing scientific teams were the most interdisciplinary, but by removing interdisciplinarity, that the best scientific teams were highly interdependent, implying that forced interdisciplinarity adds redundancy to teams, impeding teams by reducing interdependence. From a traditional social science perspective, Centola & Macy (2007, p. 716) speculated that redundancy improves a team’s efficiency. However, Lawless (2017) found support for Cummings by comparing oil firms as teams operating under a dictatorship, where they were highly redundant, versus oil firms as teams operating in a democracy, where they were highly interdependent; e.g., compare Sinopec’s 124.6 employees/M BBL of oil to Exxon’s production with 15.5 to see that redundancy creates inefficiency and serves as a source of corruption. With Fourier pairs from Cohen (1995), our theory of interdependence for two factors or competing teams is: [A,B]=iC → σAσB ≥ 1/2 (1) From Equation (1), the exact knowledge of the standard deviation for factor A (σA) precludes simultaneously the exact knowledge of factor B, leading to several matches of theory and observation; e.g., from Arrow (1951/1963), aggregating preferences of three or more individuals is impossible without a vote (viz., the majority rule in a democracy) or a unilateral decision (e.g., a dictatorship). Thus, a team of interdependent members does not aggregate summarily. If aggregation occurs with degrees of freedom, then, without adjustment, ∑nindividuals = dof. But if the perfect team acts as a single unit, then ∑dofteam = 1. Assuming the best available individuals fill the roles of a team and log (dofteam) equals to its entropy gives log(dof(perfect team)) ≤ log(dof(dysfunctional team)) (2) With this model from theory, team fit is crucial; interdisciplinarity is functional only if it improves team fitness; as a team’s dof increase, due to redundancy, role conflict, lack of communication, etc., team performance deteriorates, allowing us to revise Eq. (1) to the standard deviation of entropy produced by team structure (least entropy production, or LEP) times that for performance (maximum entropy production, or MEP): σLEP σMEP ≥ 1/2 (3) From Eq. (3), as σLEP→ 0, in the limit σMEP → ∞; thus, the best performing (science) teams expend the least effort on team structure, generating MEP for a team’s mission. In contrast, when a team becomes dysfunctional, illuminated by a perturbation, say a team divorce, Eq. (3) is reversed: as σLEP → ∞, in the limit σMEP → 0; i.e., a dysfunctional team expends entropy to tear its structure apart. Concluding, with further research, ceteris paribus, we expect to find that larger team structures generate more entropy than smaller ones (i.e., more arrangements are possible), requiring more energy (revenue); that the perfect team operates emotionally in a state similar to a ground state, the dysfunctional team at a perturbed, excited state; and that the perfect team’s generation of information to itself and outsiders is subadditive (Von Neumann information) while a dysfunctional team’s information generation is additive (Shannon information), the two forming a metric for a team’s performance, whether the teammates are humans, machines or robots, a key step for the science teams of tomorrow.