Self-organising systems are notoriously difficult to engineer,
  particularly due to the interactions between complex specifications
  and the simultaneous need for efficiency and for resilience to
  faults and changes in execution conditions.
  %
  We address this problem with an engineering methodology that
  separates these three aspects, allowing each to be engineered
  independently.
  %
  Beginning with field calculus, we identify the largest known
  sub-language of self-stabilising programs, guaranteed to eventually
  attain correct behaviour despite any perturbation in state or
  topology.
  %
  Construction of complex systems is then facilitated by identifying
  ``building block'' operators expressed in this language, into which
  many complex specifications can be readily factored, thereby
  attaining resilience but possibly with improvable efficiency.
  %
  Efficient implementation may then be achieved by substituting
  high-performance coordination mechanisms that are asymptotically
  equivalent to particular applications of building block operators.
  %
  We illustrate this workflow by construction and simulation of example
  applications for evacuation alerts and for live estimation of crowd
  feedback at mass events.