Transcript Wisdom of the crowd, wisdom of the mould

Emergence
Julita Vassileva
Social Computing Class 2009
• Based on the book:
• http://www.amazon.com/Emergence-Connected-BrainsCities-Software/dp/0684868768 , Price: $10.20
The Science of Complex Systems
• Used to analyse existing systems in nature and
society
• Now we are creating complex (self-organizing,
emerging) systems ourselves in our software
applications
– Emergent systems to recommend new books,
recognize voices, find mates
– Artificial emergence: systems designed to exploit
the laws of emergence like nuclear reactors
exploit the laws of atomic physics
Smart mould
• In 2000 Toshiyuki Nakagaki announced that he
had trained slime mould to find the shortest
path through a maze
www.abc.net.au/science/news/stories/s189608.htm
Morphogenesis
• One of Alan Turing’s last papers (1952) – on
biological development in mathematical terms
•
During the final years of his life Turing was working on what would now be called
Artificial Life or A-Life. He used the Ferranti Mark I computer belonging to the
Manchester University Computing Machine Laboratory to simulate a chemical
mechanism by which the genes of a zygote may determine the anatomical
structure of the resulting animal or plant.
• Single cells following individually simple rules
can lead to very complex structures and
behaviours
• Bottom-up, not top-down
• Self-organization
The Interdisciplinary Science of Selforganization: the laws of emergence
•
•
•
•
•
•
Adam Smith – economics
Friedrich Engels – dialectical materialism
Charles Darwin – evolution
Alan Turing – cell biology and computation
Jane Jacobs – city neighborhoods (sociology)
Marvin Minsky – human intelligence
Self-Org. Theories in Philosophy
•
•
•
•
•
•
•
•
Tectology (A. Bogdanov, 1922) ,
General Systems Theory (L. von Bertalanffy, 1937)
Self-Organization (W. Ross Ashby, 1947)
Cybernetics (N. Wiener, 1948) – steering a complex system
towards achieving a goal
Cell automata, self-reproducing systems (J. von Neumann, 1950)
Chaos theory – mathematical basis (Lyapunov, Poincare,
1890ies)
Complex Systems
Emergence
Complex Adaptive Systems with
Emergent Behaviours
• Examples: ant colonies, markets, city neighbourhoods,
brain neurons
• Masses of simple, relatively stupid units following a
few simple rules, rather than a single, intelligent,
“executive” create collectively efficient structures
which are better than any conscious design
• If you put infinite number of monkeys in front of keyboards
for infinite number of years, what is the chance that
Shakespeare’s “Hamlet” will emege?
• Need to adapt to external conditions (environmental
feedback) – provides a gradient, a notion of moredesirable (viable) vs. less desirable(viable)
configuration
Ants
• One queen and many
workers taking flexible
roles
• Communicating through
pheromone trails for food
finding
• Gestures – to express
emotions
Manchester in 1842
• Friedrich Engels: “I have never elsewhere seen a concealment of
such fine sensibility of everything that might offend the eyes and nerves of
the middle class. And yet it is precisely Manchester that has been built
less according to a plan and less within the limitations of official
regulations – and indeed more through accident – than any other town”.
Good neighbourhoods, 1961
• Jane Jacobs: “Under the seeming disorder of the old city, wherever
the old city is working successfully, is a marvelous order for maintaining
the safety of the streets and the freedom of the city. It is a complex order.
Its essence is intimacy of sidewalk use, bringing with it a constant
succession of eyes. The order is all composed of movement and change,
and although it is life, not art, we may fancifully call it the art form of the
city and liken it to the dance… an intricate ballet in which the individual
dancers and ensembles all have distinctive parts which miraculously
reinforce each other and compose an orderly whole”.
Pandemonium intelligence:
the demons in your mind
• Selfridge, 1957: Pandemonium: A paradigm for
learning (pattern recognition)
– Distributed, bottom-up intelligence, based on layers of
demons, each with a vote
• Holland, 1960ies – Genetic Algorithms: like
Pandemonium, but with evolution:
– Reproduction, cross-over, mutation, natural selection
(fitness function), all on paper
• Jefferson and Taylor, 1980ies – Tracker
– Simulating the track-following behaviour of ants
– 3 factors: reproduction, mutation, competition
–  Darwinian evolution
A-Life: Back to the mould
• Mitch Resnick, 1984: Slime-mould simulation
– StarLogo simulation with two key variables:
-- #of cells,
-- temporal length of the pheromone trail left.
Principles of design of systems
that learn bottom-up
– More is different, there is a critical mass when the
behaviour of the whole changes – state transition
– Ignorance is useful, use simple blocks, self-interested
and with only local awareness
– Encourage random encounters, to allow cancellation
of errors and discovery
– Look for patterns in the signs, to allow information to
flow (the pattern creates a gradient, env. feedback)
– Pay attention to your neighbours – local information
can lead to global wisdom
Emergence mechanisms
•
•
•
•
Neighbour interaction
Pattern match
Feedback
Indirect control
Neighbour Interaction
• Ant-colonies: ants interacting through
pheromones and observing their neighbours
• Cells in the embrio – all cells reading different
sections of the same DNA, taking cues from
their neighbours about what section to ”read”
• SimCity – a meshwork of cells that interact
with their neighbours following simple rules
• Sidewalks as providers of interactions that
change people’s behavior
Pattern match
• Patterns emerge as optimal points
(equilibriums) in interactions and their driving
forces
– Example: Trading streets in old cities, e.g. Florence
– What is equilibrium? A balance of forces , e.g. a
point of mutual convenience for everyone
involved.
• Feedback loops and state transitions
• Complex systems “learn” patterns
How does our immune system
learn?
• During our lifetime, vocabularies of antibodies
are built from exposure to different threats
• Antibodies learn to recognize a threat
• Antibodies learn to neutralize a threat
• Antibodies remember the strategy over the lifetime
– The recognition unfolds purely on a cellular level
– The immune system need not to be conscious to
be capable of this type of learning.
Florence XI –XXI Century
• Do cities learn?
– Silk weavers and goldsmiths settled on Por Santa Maria
street north of Ponte Vecchio in the XI-th century, they are
still there now.
– The pattern is “learned” by the city, but not consciously; it
persists due to individual conscious decisions, but at a
lower level (self-interest, max. utility)
Listening to Feedback
• Positive feedback and Negative feedback
– Positive feedback loops boost growth (self-enforced)
– Negative feedback steers towards a target (e.g. a
thermostat)
– Homeostasis – a balance of forces, equilibrium point
• Examples
– Cities – Paul Krugman’s “self-organizing economy” with a
very simple math model of city growth driven by
centripetal and centrifugal forces explains polycentric
“plum-pudding”pattern of modern metropoilis
– Online communities, like Slashdot or Comtella (feedback?)
– Systems with user ratings: Amazon and eBay (feedback?)
The Art of Control
• Emergent system “is a little chaos machine, unexpected
things happen and you only control if from the edges” - Eric
Zimmerman, scientist, artist, game designer
• “The rules make the game”, emergent systems are rulegoverned as well (low level rules)… if an agent stops
following the rules, anarchy or chaos results
• Group behaviour evolving in unpredictable ways in online
games,
– e.g. SimCity has no pre-defined objectives
– The more autonomous the system, the more irrelevant the
player is
– Game designers wonder how far off the edge should a player
be, to keep him/her interested in the game?
The Noosphere
• Is collective intelligence emerging on the web?
Image from: http://noosphere.princeton.edu/