Challenges in the Health Care System How to

Challenges in the Health Care System How to evolve an overly-constrained, mature, complex, interdependent system? Institutional change is too slow 1. 2. – – Host-pathogen systems are dynamic Major changes in who is providing health care – More women, alternative medicine, “Physician's will be put of business by nurse practitioners” – Data-poor ⇒ data-rich ⇒ data-overload 3. – – The patient’s genomic “pre-existing condition” problem c. f. “Blink” by M. Mc. Dowell Increasingly broad patient cultural-ethnic background – Diversity 2007 Who is the most genetically diverse ethnic group?

“Normal” Technology Development Phases How to organize a movement, that changes/coordinates 100 s of organizations and impacts 700, 000 physicians? How do you then build processes that support new “utility”? How do new structures then become “transparent” and the building blocks of new options and structures?

Extreme stages can prevent development X X Collective reinforcement of hype can lead to a interruption of the developmental cycle

Diversity: A Weapon of Mass Construction Norman L Johnson Referentia Systems Inc Norman@santafe. edu http: //Collective. Science. com Jen Watkins Los Alamos National Lab jhw@lanl. gov http: //public. lanl. gov/jhw

Mauboussin: Deep Blue, Wisdom of the Crowds & Demise of Experts 1. 2. 3. Utility of experts is being eroded What system benefits from expert performance – Computers: rule-based or probabilistic, limited options (low heterogeneity / complexity, but lots of data) – Experts: rule-based, many options (moderate complexity and data) – Collective intelligence: probabilistic, many options (high complexity, lots of data) Conditions for Collective Intelligence – 4. Diversity, an aggregation mechanism, incentives Examples of Collective intelligence – Discovery: Needle in the haystack 1. State prediction: Jelly beans in jar 2. Future prediction: Academy Awards Diversity 2007

Mauboussin: What system benefits from expert performance Domain description: Rule-based, limited options Rule-based, many options Probabilistic, limited options Probabilistic, many options Expert performance: Worse than computers Better than computers Equal or worse than computers Collectives outperform experts Expert agreement: High Moderate-Low (70 -90%) (50 -60%) (30 -40%) (<20%) Examples: Credit scoring Simple medical diagnosis Chess Go College admissions Poker Stock investing Econ. forecast Source: Michael J. Mauboussin, ”Are You an Expert? ” Mauboussin on Strategy, October 28, 2005. Diversity 2007

Mauboussin: Questions asked 1. What about a “crowd of experts”? 2. What about incentives on web-based systems? 3. How do you use Diversity in investment? 4. What about behavior influencing individuals or crowds? Diversity 2007

Test Question on the most important topic of our times How did we get here? 1. Evolution 2. Intelligent design 3. Creationism Diversity 2007

How to make sense of it all? Consider Leadership “Of all the hazy and confounding areas in social psychology, leadership theory undoubtedly contends for nomination. And, ironically, probably more has been written and less is known about leadership than about any other topic in the behavioral sciences” - Bennis 1959

Theories of Leadership Power-based: leaders sustained by power Innate traits theories - Leaders are born not made v v Trait theory (Stogdill 1974) Great Man theory (Carlyle 1843) Structure-based: leaders fill structural roles Role theory v v v Social theory and structure (Merton 1957) Supervisory behavior (Pfeffer & Salancik, 1975) (Adair 1965) Performance-based - leaders as performers Ex: Situational theory (Tennenbaum 1958, Maier 1963, Yuki 1989) Required leader style is situation dependent Ex: Contingency theory (Fiedler 1954) - Context based Leader’s effectiveness is based on ‘situational contingency’ Performance-based - Collective/plurality/shared-based: leaders as enablers Ex: Shared, relational, collective, situational theories Ex: Distributed leadership (Gronn 2002) - Plurality based Leadership happens everyday in formal and informal interactions and is spread over leaders, followers, and context Ex: Adaptive leadership (Linsky 2002) - CAS-based “Leadership is an interactive event in which knowledge, action preferences, and behaviors change, thereby provoking the organization to become more adaptable. ” Performance-based Leadership + Emergence Leadership 2007

Landscape Model for Leadership Previous presentation useful to understand the literature, but less helpful to develop new approaches and resources for leadership. Where Leadership Arises How Leadership arises Some Individuals Emergent: Unpredictable, opportunistic, from interactions Structurally determined: predictable Leadership 2007 Classic Leaders from power or traits (structure) Most of the Collective

Landscape Model for Leadership Examples Where arises? How arises? Some Individuals Most of the Collective Classic Leadership theories for individuals Aggregation methods from collective input democracies, markets, etc. => “lower” collective intelligence Emergent Structurally determined Leadership 2007

Landscape Model for Leadership Distribution? Emergence? Some Individuals High Leadership emerges without precedence opportunistic None Classic Leadership Most of the Collective “Lower” collective intelligence Secondary considerations Embodied and disembodied emergent solutions (upper row) The emergent solutions can either be embodied in individuals or captured in the interactions between the individuals (disembodied). Maturation: creating structure by “capturing” the emergent property • Emergent leadership can go “classic” with rules and regulations • Ex: early development of e. Bay: e. Bay structure followed emergent social processes Leadership 2007

Landscape Model for Leadership Distribution? Involves some Individuals Involves all of Collective High Leadership emerges without precedence ? None Classic Leadership “Lower” collective intelligence Emergence? Leadership 2007

Examples within Leadership Landscape Unpredictable Degree of Emergence: How performance arises Localized Emergent: Leadership outside of structure as in a hero or savior Localized Deterministic: Classical top-down leaders supported by structure Deterministic Localized Distributed Emergent: Emergent functions in societies as in the fall of the Berlin wall, future symbiotic intelligent systems, … Distributed Deterministic: Democracies, commodity/currency exchanges, prediction markets, recommender systems, … Distributed Degree of Distribution Where performance is located Leadership 2007

Expert Performance in Finance Why can’t financial experts outperform consistently the S&P 500 “collective” (including good + bad performers)? • Professional money managers fail to beat the S&P 500 at an average rate of 70% per year. • 90% trail the S&P over a 10 -year period. • Only a few beat every year for 10 years – Soros, Miller, …. “These are the people who have more knowledge and more training than the vast majority of investors. And yet, neither the superior knowledge nor the superior experience helps them in the long run. ” Bill Mann, TMFOtter Diversity 2007

Expert Performance in Finance Why can’t financial experts outperform consistently the S&P 500 “collective” (including good + bad performers)? Where Experts Have Value Experts ? Simple Value Collectives Complex Domain Diversity 2007 Michael Mauboussin - Legg Mason Capital Management

Ants Solving “HARD” problems The ant colony (and individuals) finds the shortest path Food Does a “classic leader” find the path? How is this possible? Nest

A Model for Collectives Solving Hard Problems How can groups > solve hard problems, > without coordination, > without cooperation, > without selection? The Maze has many solutions > non-optimal and optimal. Start Individuals > Solve a maze > Independently > Same capability End When In “Learning” But because a individuals the maze, global solve the maze individuals perspective is again, create they a missing, they eliminate diversity of cannot shorten “extra” loops experience. their path. This is where diversity helps.

Normalized number of steps Averaged Performance 1. 3 Using novice information, with two different collections 1. 2 1. 1 . Average Individual 1. 0 0. 9 Using established information 0. 8 0 5 Shortest path 10 15 Individuals in Collective Decision 20

How collectives find the Shortest path Paths of three ants Collective path Unlike in natural selection, no one individual is the fittest!

Noise and Robustness Noise: Replace “valid” information with “false” information An “expert” individual • Individuals are very sensitive to noise A collective • Collectives are insensitive 10 steps become 21 steps 10 steps become 9 steps Lack of experience Contingency from diversity

Conclusions on Emergent Problem Solving Collectives reliably solve a problem “perfectly” that experts cannot reliably solve The emergent solution is not initially embodied in any individual (no one ant finds the shortest path). Diverse collectives not only perform better, but they are also more robust to misinformation. The accuracy of emergent solution correlates with diversity Diversity is defined as uniqueness of information/skills contributed Diversity of performance is also required Competition, optimization or stress all reduce diversity, performance and/or robustness. Collective “solves” a problem that individuals are unaware of => emergent problem definition and solution Performance from synergistic diversity has a sweet spot Collective performance is bounded by individual performance and complexity of the problem. Diversity 2007

! Simple Complex Domain Diversity 2007 Value of Collectives Value of Experts Expert Performance in Finance

Landscape Model for Leadership Distribution? Emergence? Involves some Individuals Involves all of Collective High Localized leadership emerges without precedence or structure Emergent collective “Leadership” from synergistic diversity None Classic Leadership “Lower” collective intelligence Leadership 2007

Leadership Landscape Considerations Distribution Emergence Involves some Individuals Involves all of Collective High Leadership emerges without precedence or structure Emergent collective “Leadership” from synergistic diversity None Classic Leadership “Lower” collective intelligence Secondary considerations Embodied and disembodied emergent solutions (upper row) Maturation: creating structure by “capturing” the emergent property Problem solving capability and robustness increases as you move up & right • Higher diversity (and complexity) is required from lower left to upper right The efficiency-quality tradeoff problem - particularly for fast change • Lower left can be the most efficient (less communication and coordination required) and has a speed advantage in dealing with fast change • But Upper right has the best and most robust solutions Mixed “classic” leadership models are becoming essential • “Localized” Leader as facilitator to develop emergent collective intelligence Leadership 2007

Revisit Traits of Good Leadership Do some of these also apply to Distributed Leadership? Good Leadership traits v. Performance: Accurately and reliably solves problems => Collectives outperform leaders v. Approach: Able to communicate and persuade others without resort to negative or coercive tactics v. Resources: Able to understand a wide range of areas => Collectives have greater resources v. Integrity: Owning up to mistakes, rather than putting energy into covering up => Collectives are more robust v. Personality: Calm, confident and predictable, particularly when under stress => ? Collectives have Leadership 2007 detrimental herd (heard) effects

Themes for Complex Adaptive Systems Processes in “one” system – – – Role of diversity Optimizing performance or robustness Multi-level viewpoint: system(internal, external) Interplay of structure (rules) and options Situated intelligence Co-Development of multiple systems How systems develop? Dynamics and effects of change Strategies for responding to change Diversity 2007

Conclusions Challenges: v Creating hype, utility and transparency – while in a quickly changing system Consider a Leadership landscape that covers all resources v From power-based, structure-based resources to performance-based, distributed and emergent theories v New leadership resources reflect greater needs by society Match “Leadership” resources to systems characteristics v Data-poor systems of high complexity require experts v Data-rich systems of low complexity require computer resources v Data-rich systems of high complexity require human-computer solutions Complex systems require diversity for performance and robustness Leadership 2007

Test again How did we get here? 1. Evolution 2. Intelligent design 3. Creationism Diversity 2007

Rat Studies of Maximum Carrying Capacity Cooperative social structure Control - no imposed social structure NIMH psychologist John B. Calhoun, 1971 Both systems loaded to 2 1/2 times the optimal capacity. Social order system can carry 8 times the optimal capacity.

References Johnson, N. L. (1998). "Collective Problem Solving: Functionality Beyond the Individual. " from http: //collectivescience. com/Documents 1. html Johnson, N. L. (2002). "The Development of Collective Structure and Its Response to Environmental Change. " S. E. E. D. Journal 2(3). Lichtenstein, Uhl-Bien, Marion, Seers, Orton and Schreiber. “Complexity Leadership Theory: An interactive perspective on leading in complex adaptive systems” Emergence: Complexity and Organization Volume 8, Number 4, 2006. <Entire issue is of interest> Sawyer, R. K. (2006). Social Emergence: Society as Complex Systems. Symbiotic Intelligence Project http: //www. collectivescience. org/symintel. html Watkins, J. H. (2007). “Prediction Markets as an Aggregation Mechanism of Collective Intelligence. ” from http: //public. lanl. gov/jhw. Watkins and Rodriguez (2007). “A Survey of Web-based Collective Decision Making Systems” from http: //public. lanl. gov/jhw. Diversity 2003 Los Alamos

Norman L. Johnson Norman@santafe. edu Jennifer H. Watkins Jhw@lanl. gov Diversity 2003 Los Alamos

The Problem with Collective Effects Ants foraging for food chose one path out of two equidistant paths. Cooperation leads to exclusive behavior in stable environments Food Non-linear or Chaotic behavior: Positive reinforcements can amplify random weak signals >> global chaos. Nest (Deneubourg et al. 1990) Diversity 2005 Los Alamos

Definitions of Leadership Good Leadership traits: v. Performance: Accurately and reliably solves problems. v. Approach: Able to communicate and persuade others without resort to negative or coercive tactics. v. Resources: Able to understand a wide range of areas, rather than having a narrow (and narrowminded) area of expertise. v. Integrity: Owning up to mistakes, rather than putting energy into covering up. v. Personality: Calm, confident and predictable, particularly when under stress. Leadership 2007

Definitions of Leadership “Leadership is like the abominable snowman whose footprints are everywhere but who is nowhere to be seen. “ “Leadership is like pornography (or complexity), you know it when you see it, but you can’t define it. ” Leadership 2007

How to Predict and Manage Change Broadest understanding – – Governing processes Specific system (inference, network, etc. ) Environment and context Dynamics: Specific triggers, non-linearities, etc. Methods - Resources – Theories & Models – – Diversity 2007 Dynamical systems vs. Complex systems Simulations, experiments, … Data sources (poor or rich or overwhelmed)? Science: Integration of theory, data & simulations Predictive methods: risk assessment, uncertainty management

Self-Organizing Adaptive Systems Agent Interaction Emergent properties rules “Solutions” arise from the dynamics from a diversity of potential solutions. Decentralized, robust, adaptable, fault-tolerant, scalable, . . . Fundamental concepts Chaotic behavior or non-linear response Performance AND robustness Emergent properties Leadership 2007

Network View of System of systems Environment (culture, economy, demography, technology, nature) Individual • Sensory • Memory • Motivation • … Social-organizational -information network • Diversity • Connections • Strengths • Asymmetry • Change Dynamics on the network performance - stability - resilience - transients • Change of states • Creation/destruction of structure & options • Dynamics under stable conditions • Dynamics in response to change Individual types Personal • Peers • Motivation • Bosses • Sensory • Clients, … • … Groups Personal • Media • Motivation • Organization • Sensory • … Regulations Personal • Feds • Motivation • Agencies • Sensory • …