HPII: High-Performance InterOrganizational Interaction
for Disaster Response
An evolutionary game
Fred Phillips
October 19, 2012
Organization of the talk
• Background on disasters and statement
of today’s problem
• Relevant prior work
• Proposal for a game to be used as
– Experiment
– Data-gathering instrument
– Training tool for disaster aid agencies
– (Possibly) theoretical advance.
Disasters
and the
Inter-agency Problem
Distribution of Natural-Hazard Deaths in the US
KA Borden & SL Cutter, Spatial patterns of natural hazards mortality in the United States. Int’l
Jour of Health Geographics 2008, 7:64, http://www.ij-healthgeographics.com/content/7/1/64
Public Disasters: Examples
•
•
•
•
•
•
Exxon Valdez spill
Deutsch Bank building
Mortgage crisis
Hurricanes Katrina, Sandy
Fukushima
BP blowout
“Public Disasters”
• Caused by nature, by individuals, or by public
or private institutions - or a combination
• High-impact events
• Low-probability events individually - though
collectively, public disasters are quite probable
• Multiple organizations, often from different
sectors, are
– To blame for the event, and/or
– Involved or accountable for remediation
The Disaster Cycle
Today we
look at
this segment
of the
cycle.
Most of these failures of disaster
response/remediation involved
failures of inter-institutional
relationships
•
•
•
•
Jurisdictional disputes
Quest for glory & budget
Unrestrained externalization of corporate costs;
Lack of accountability e.g., in the selling and
securitization of mortgages;
• Difficulty of introducing change when multiple
stakeholders guard the gate; and
• Lobbying and industry control of supposed
watchdog institutions
We all know the plot…
After a disaster each aid
organization wants to…
•
•
•
•
•
•
Display its expertise
Be “in charge”
Avoid legal liability
Avoid embarrassment
Get good press
And try to shift responsibility to, or even
sabotage the efforts of, other agencies,
in order to get these things.
The Players in Three Crises
Exxon Valdez
Deutsch Bank NY
Mortgage crisis
•Exxon Valdez captain,
crew
•Exxon Corp.
•State of Alaska
•US Dept of Interior
•US Environmental
Protection Agency
•Deutsche Bank
Executives
•Insurors
•New Yorkers
•Governments
•EPA
•New York courts
•Planners
•Construction
companies
•Legislators
•Community Groups
•Home buyers
•Mortgage originators
•Mortgage buyers
•Mortgage insurors
•Financial
intermediaries
•Investors
•US Federal Reserve
Bank
•Alyeska Corporation
How can we optimize interagency cooperation in postdisaster situations, balancing
these conflicting forces?
Some Prior Work
Classifying the tools
Within the organization
Inter-organizational
• High-Performance • Alliance management
Normal Organization (HPO) • Accounting rules
times
theory
• Negotiation
• Game theory
• Some HPO theory
Crisis
times • Crisis management • Game theory?
• “Swift Trust in Hastily
Formed Networks”*
•???
* R. Zolin, US Naval Postgraduate School (undated)
Phillips (2011)
• Mapped these ideas to the disaster situation,
in a multiple-perspectives systems schema:
–
–
–
–
•
Moral hazard
Adverse selection
Integrity/Breach of Trust
Moral authority
- Externalities
- Responsibility
- Accountability
- Transparency
Phillips, Fred, “Inter-institutional Relationships and Emergency Management.” International Journal of Society
Systems Science, Vol. 3, No.1/2 2011 pp. 40 - 57.0
Delton et al, in Proceedings of
the NAS 2011, showed:
The incidence of altruistic (cooperative)
behavior depends on
– The actor’s assessment of the chances of
ever meeting the other party again.
– The actor’s assessment of the probable
frequency of meeting the other party again.
•
Delton, Andrew W., Krasnow, Max M., Cosmides, Leda, and Tooby, John, “Evolution of direct reciprocity under
uncertainty can explain human generosity in one-shot encounters.” Proceedings of the National Academy of Science,
July 2011. http://www.pnas.org/content/early/2011/07/20/1102131108.abstract 1
Novak (Science 2006) uses game theory to
study evolution of cooperative behavior in
prisoners-dilemma type games.
• Bit-string ‘organisms’ are coded for cooperative or selfish
strategies.
– A ‘strategy’ is a vector of probabilities.
• Game is played.
• Strings producing high payoffs may ‘reproduce’
– ‘mating’ with other successful strings (offspring have sub-strings from
each parent)
– Occasional random ‘mutation’
• Many ‘generations’ of plays show
– Emergence of cooperative behavior
– Five basic cooperative super-strategies
•
•
•
Nowak, Martin, “Why We Help.” Scientific American, July 2012, 34-39.
Nowak, Martin, “Five Rules for the Evolution of Cooperation.” Science 314:1560-1563, December 8, 2006.
Nowak, Martin, Super-cooperators: Altruism, Evolution, and Why We Need Each Other to Succeed. Free Press, 2012.
Novak: Five basic cooperative
super-strategies
1. Direct reciprocity
– “evolution of forgiveness”
2. Spatial selection
– Accords with Delton et al.
3. Indirect reciprocity
– Decision to help is based on recipient’s reputation for
helpfulness.
4. Kin selection
5. Tribal selection
– Differs from spatial or kin selection b/c altruism is not
directed at an individual.
Novak:
• “Humans, more than any other creature,
offer assistance based on indirect
reciprocity, or reputation.”
• Why? Because we have language (and
Facebook, and credit-scoring agencies!)
to make a person’s reputation widely
known.
Currao, A New Role for Emergency Management:
Fostering Trust to Enhance Collaboration in Complex
Adaptive Emergency Response Systems. Master's
thesis, Naval Postgraduate School Monterey CA, 2009.
• In spite of the 9/11 Commission Report and a revised incident
command system, effective interagency collaboration at emergency
incidents within New York City has not been fully achieved.
• This thesis explores
– how ... collaborative efforts [depend on] inter-organizational trust, and
– whether emergency management agencies [can] assume a leadership
role in fostering and implementing trust-building programs [for]
collaborative agency partnerships.
• [I interviewed] senior management of seven public safety agencies...
• Conclusions:
– Trust enhanc[es] effective interagency partnerships [and] increased
problem solving capacities.
– The “leadership in building trust” concept is complex, ..., [requiring] a
synthesis of agency skills to meet homeland security challenges.
The Game
• Research gap/opportunity
– The mathematics of kin and group selection
are still controversial.
– Post-disaster cooperation is a special case.
• Research questions:
– How can the evolutionary game idea be
adapted to post-disaster cooperation among
agencies?
– Is Delton’s “probability assessment”
(extended spatial selection) idea most
relevant to disaster situation, or Novak’s
indirect reciprocity?
A disaster aid agency’s planning
(hypothetical)
Level of emergency preparedness
<- Operational Skills | Cooperation Skills ->
20
18
16
14
12
10
8
6
4
2
0
1
2
3
4
5
6
7
8
% of budget spent on alliance training
Low
|
Capacity to cooperate
Mid | High | Very High
Game: How much cooperation
to extend?
Agency #1
lo
mid
hi
very hi
lo
o
-
mid
+
o
hi
o
+
+
o
very hi
o
o
-
Payoffs are “Level of effective disaster relief.”
These are payoffs to the victims, not to the
players!
Players will be told:
• “In this disaster scenario, you must deal
with an agency that you will interface
with in x% of your own agency’s future
missions. Here is that agency’s history
of (non)cooperation.”
• x will be drawn from a suitable
probability distribution, on each play.
• The distribution will be conditional on
the size of the other player.
Players will decide to:
• Extend a
– Low
– Mid
– High, or
– Very high level of cooperation to the other
agency....
• Assuming their budget allocation
supports that level of cooperation.
It is highly likely that
• A small agency will interface with a
large agency in a large % of the smaller
agency’s missions.
• A larger agency, with
– A bigger budget and
– A larger number of total missions
Will interface with the particular smaller
agency in only a small % of its missions.
What evolves, in this
evolutionary game?
1. Agencies’ strategies for extending
cooperation to other agencies.
2. Agencies’ views of their own mission.
• News item: Scottish fire departments now
emphasize fire prevention skills over
dousing skills.
• This is “double-loop learning.”
Uses of the game
1. As experiment, to find out what could
happen under evolutionary scenario.
2. As an online game, played by disaster
agency managers, so the researchers
can gather data on empirical behavior
of agencies.
3. As thesis topic, for a student with
programming skills?
4. Later, as a training tool for agencies.
(Practical impact)
Kristalina Georgieva, the EU’s
humanitarian aid
commissioner
• On ECHO’s
principles: “Humanity,
neutrality, impartiality,
and independence.”
• On disaster
preparedness: “By
2016, we are
proposing that all
member states have
risk management
plans in place.”
(Possible Research Impact)
• Particularizing “evolution of cooperation”
research to the disaster situation.
• Delton’s “probability assessment,” or
Novak’s indirect reciprocity?
• New wrinkles in game theory?
– Disaster victims as passive 3rd player
– Double-loop learning
Other readings
• In addition to those cited in the slides:
•
•
•
•
•
•
•
•
Chen, Bin. (2008). “Managing Inter-organizational Partnerships: Preconditions, Processes and
Perceived Outcomes,” Best Papers Proceedings of the Sixty-Sixth Annual Meeting of the
Academy of Management.
“Crisis and Consequence.” BizEd, (January/February 2009), 24-30.
Faraj, Samer, and Yan Xiao, “Coordination in Fast-Response Organizations.” Management
Science 52-8, August 2006, 1155-1169.
Kale, Prashant, and Harbir Singh, “Managing Strategic Alliances: What do we know now, and
where do we go from here?” Academy of Management Perspectives, August (2009) 45-62.
Kemeny, John G. (1980). "Saving American Democracy: The Lessons of Three Mile. Island."
Technology Review 82 (June-July).
Muhtada, Dani. “Ethics, Economics and Environmental Complexity: The Mud Flow Disaster
in East Java.” Systems Research and Behavioral Science, vol. 25 (July, 2008), 2.
White, Leroy. “Connecting Organizations: Developing the Idea of Network Learning in InterOrganizational Settings.” Systems Research and Behavioral Science, vol. 25 (September,
2008), 6.e
Tatham, Peter; Kov當s, Gy gyi. The application of ‘swift trust’ to humanitarian logistics.
International Journal of Production Economics 126(1) July, 2010. p. 35-45. DOI:
10.1016/j.ijpe.2009.10.006.