Game Theoretic and Economic Models in IT

Game Theory and Economic Models in IT

Recent progress in theory for diverse game contexts

repeated/finite, sustaining efficiency and cooperation, auctions, bargaining and negotiation, incomplete/imperfect information.

Becoming standard language for decision making: economics, sociology. Recent Nobel prize for Nash.

Networks:

QoS: Service and resource pricing games for better QoS provisioning, congestion control. Current research activity.

Denial of Service: Uncertainty in opponents, actions. Game approach to design monitoring and enforcement, balance costs and risks.

Logistics - recent successes

Physician Resident Matching program (preference matching)

Cal power exchange (Dutch auctions)

Explore application to DOD scenarios

Collaborative reasoning, negotiation.

Embed game theoretic tools in agents/decision aids; group decisions.

From
Heisenberg vs Bohr
to
Picard vs Q:

New Challenges for Game Theory and National Security

US Defense Policy: 3 Goals

Strengthen the bond of trust between the American people and those who wear our nation’s uniform. We’ll give them the tools they need and the respect they deserve.

We will work to defend our people and our allies against growing threats: the threats of missiles; information warfare; the threats of biological, chemical and nuclear weapons. We will confront the new threats of a new century.

We will begin creating the military of the future, one that takes full advantage of revolutionary new technologies. We will promote the peace by redefining the way wars will be fought.

President Bush at SecDef Rumsfeld Swearing-In Ceremony, 012601

National Intelligence Council

Global Trends 2015

Biological Warfare Threat

More players with offensive BW program

More lethal agents

More proliferation of technologies

MI2 Game: The leaders of x believe they and only they have the antidote to a genetically altered agent. Your job is to prove them wrong (even if, and especially if, they are right).

R. Preston “The Bioweaponeers”, New Yorker

The Past is Prologue

1962 Cuban (Missile) Crisis = More than 13 Days

Virtual War: Nuclear Operations During the Cuban Crisis , P. Kozemchak , OSD/NA

A. Wohlstetter, R. Wohlstetter On Controlling the Risks in Cuba, RAND and IISS Paper

ARPA / DNA Long-Range R&D Program (Lukasik, Haas)

1973 US 6^th Fleet vs Soviet Eskadra

Center for Naval Analyses

Schlesigner, Kissinger, Hearings

1989 Gorbachev Coup

Odom

1991 Iraq Biological Weapons

Powell, Jumper

Desert Fox

Leide, Zinni

Confronting WMD
Prevention Options

Assume: The Democracies declare

(1) Any use of WMD against the citizens of the US or its allies is an act of war.

(2) The use of WMD is the last act of the user.

Your Mission:

Create an X-FORCE to prevent use.

Non-lethal and lethal options

Information Requirements

Measures of risk and performance vs ROEs\

pkozemchak@darpa.smil.mil

pkozemchak@darpa.ic.gov

“All things are ready if our minds be so.”

Current DOD readiness reporting systems (including JMRRs) fail to measure “decision superiority” of our commanders, value of information to warfighters

Joint Vision 2020 and GPRA

Is a 12-hour restoration time for mission-critical networks in “national security emergencies” the best we can do?

What is DOD’s Social Welfare Function?
The Strategic Review Meets The Impossibility Paradox

Research Topics

Omniscience (o) and Omnipotence (O) in Games

Quantum Strategies (Meyer)

Brams Theory of Moves on Hamlet, Superior Beings, Biblical Games

Copenhagen M. Frayen

Hapgood T. Stoppard

Warrior Codes in J. Keegan The Book of War: 25 Centuries of Great War Writing

Bayesian Games and Mechanism Design

The Market for Innovation

Akerloff and Heal on supergames and product quality

Surreal Games (Conway, Knuth)

If surreal numbers are the answer, what’s the question? “Fair fights” in networks?

Nash equilibria in large dynamic games (A. Green, Courant)

“Who dares wins.”

Can Game Theory Build Better Gamers? And Vice Versa?: An Experiment

“On the Net nobody knows you’re… an algorithm.”

Got game? Get game theory!

Create an experimental gaming network.

For each game of interest:

What’s the best strategy?

Fight it. Prove it. Improve it.

No published solutions for multiplayer online games

Use prize authority to sponsor algorithm tournaments

Can you beat the OPFOR in Half-Life?


Recent progress in theory for diverse game contexts
	repeated/finite, sustaining efficiency and cooperation, auctions, bargaining and negotiation, incomplete/imperfect information.
	Becoming standard language for decision making: economics, sociology. Recent Nobel prize for Nash.
Networks:
	QoS: Service and resource pricing games for better QoS provisioning, congestion control. Current research activity.
	Denial of Service: Uncertainty in opponents, actions. Game approach to design monitoring and enforcement, balance costs and risks.
Logistics - recent successes
		Physician Resident Matching program (preference matching)
		Cal power exchange (Dutch auctions)
		Explore application to DOD scenarios
Collaborative reasoning, negotiation.
	Embed game theoretic tools in agents/decision aids; group decisions.


	Strengthen the bond of trust between the American people and those who wear our nation’s uniform. We’ll give them the tools they need and the respect they deserve.
	We will work to defend our people and our allies against growing threats: the threats of missiles; information warfare; the threats of biological, chemical and nuclear weapons. We will confront the new threats of a new century.
	We will begin creating the military of the future, one that takes full advantage of revolutionary new technologies. We will promote the peace by redefining the way wars will be fought.
			President Bush at SecDef Rumsfeld Swearing-In Ceremony, 012601


Global Trends 2015
Biological Warfare Threat
	More players with offensive BW program
	More lethal agents
	More proliferation of technologies
		MI2 Game: The leaders of x believe they and only they have the antidote to a genetically altered agent. Your job is to prove them wrong (even if, and especially if, they are right).
			R. Preston “The Bioweaponeers”, New Yorker


1962 Cuban (Missile) Crisis = More than 13 Days
	Virtual War: Nuclear Operations During the Cuban Crisis , P. Kozemchak , OSD/NA
		A. Wohlstetter, R. Wohlstetter On Controlling the Risks in Cuba, RAND and IISS Paper
		ARPA / DNA Long-Range R&D Program (Lukasik, Haas)
1973 US 6^th Fleet vs Soviet Eskadra
	Center for Naval Analyses
		Schlesigner, Kissinger, Hearings
1989 Gorbachev Coup
	Odom
1991 Iraq Biological Weapons
	Powell, Jumper
		Desert Fox
			Leide, Zinni


Assume: The Democracies declare
	(1) Any use of WMD against the citizens of the US or its allies is an act of war.
	(2) The use of WMD is the last act of the user.
Your Mission:
	Create an X-FORCE to prevent use.
		Non-lethal and lethal options
		Information Requirements
		Measures of risk and performance vs ROEs\
			pkozemchak@darpa.smil.mil
			pkozemchak@darpa.ic.gov


	Current DOD readiness reporting systems (including JMRRs) fail to measure “decision superiority” of our commanders, value of information to warfighters
		Joint Vision 2020 and GPRA
		Is a 12-hour restoration time for mission-critical networks in “national security emergencies” the best we can do?


Omniscience (o) and Omnipotence (O) in Games
	Quantum Strategies (Meyer)
	Brams Theory of Moves on Hamlet, Superior Beings, Biblical Games
		Copenhagen M. Frayen
		Hapgood T. Stoppard
		Warrior Codes in J. Keegan The Book of War: 25 Centuries of Great War Writing
Bayesian Games and Mechanism Design
	The Market for Innovation
		Akerloff and Heal on supergames and product quality
Surreal Games (Conway, Knuth)
	If surreal numbers are the answer, what’s the question? “Fair fights” in networks?
		Nash equilibria in large dynamic games (A. Green, Courant)


“On the Net nobody knows you’re… an algorithm.”
Got game? Get game theory!
	Create an experimental gaming network.
		For each game of interest:
			What’s the best strategy?
			Fight it. Prove it. Improve it.
			No published solutions for multiplayer online games
			Use prize authority to sponsor algorithm tournaments
				Can you beat the OPFOR in Half-Life?