Negotiation & Challenge Problem meet Complexity & Dynamics
*  Purview: interaction between dynamics & complexity teams and
   challenge problem teams
  -  Can CP teams exploit work by D&C teams at run-time?

What Can Dynamics & Complexity Contribute?
*  A priori analysis/experiments may give run-time guides
  - e.g., experimental determination of order parameters may allow
    experimental results to be extrapolated to broader systems
*  Phase transition phenomena can be detected at run-time
  -  not just an off-line analysis technique
*  Run-time metrics can approximately/heuristically determine
   negotiation "position"
  -  w.r.t. phase boundary
  -  w.r.t. shortest completion time
*  Example indicators:
  -  information content exchanged between agents
  -  length of negotiation chains/size of interaction graphs
  -  duration of negotiation sessions
  -  brittleness of solutions
  -  heuristic identification of back-bone variables
*  Detection can be local
  - but may require extant protocols to be augmented with
    transition-related information

How Can D&C Information be Exploited?
*  Negotiation parameters may be tied to phase transition information
  - agents may adjust resource usage to suit easy, difficult or
    intractable problems
  - agents may abandon low-priority tasks if high-priority tasks are
    determined to be in jeopardy
  - backbone variables may be considered to be non-negotiable
  - agents may be willing to compromise earlier if they know they are
    near the phase boundary
  - hypothesis: more realistic expectations of what is achievable may
    lead to better outcomes
*  Most protocols already have suitable controls
  - incorporating D&C elements is anticipated, by CP teams, to be
    straightforward

Finding Complexity
*  Statistical techniques of D&C tend to be important in large systems
  -  100s of sensors, dozens of targets
  -  Will challenge problem provide suitable context?
   -  simulator?
*  Other dimensions of complexity may be relevant
  -  e.g., size of state space arising from complexity of sensors
  - Which characteristics of challenge problem sensors are
    representative of typical, deployed systems?
*  Performance profiles other than phase transitions play a role
  - predictions of costs of negotiation can help determine course of
    negotiation

Action Items
*  December 31: Stephen Fitzpatrick (fitzpatrick@kestrel.edu) collect
   1-page descriptions of analyses from D&C teams and distribute to CP
   teams
  -  what protocol characteristics D&C teams need to know
*  January 7, Stephen Fitzpatrick collect 1-page descriptions of
   protocols from CP teams and distribute to D&C teams
  - detailed/concrete enough to allow D&C teams to determine course of
    further investigation
  - unrealistic to expect all negotiation protocols used by CP teams
    to be described as variants of a single protocol
   -  so D&C teams will need to focus