MIT+CCI

MIT CCI- Tom Malone et al resources
[] MIT Media Lab forum
 * Tom Malone, Sandy Pentland & Karim Lakhani Discussion- Video** 2007

[] MIT Media Lab forum 2002- Rheingold Smart Mobs "Future of Work deals with collective intelligence" Lakhani- studies distributed innovation Malone- whole category of "Collective Intelligence" How does one take advantage of all the knowledge in the room?
 * Tom Malone, Sandy Pentland & Karim Lakhani Discussion- Video** 2007

What is it? Malone- groups of individuals doing things together that seem intelligent Collective Intelligence around a very long time. As long as people. Sometimes exhibit collective stupidity. Therefore how to recognize differences in collective intelligence v stupidity. In last few years a new kind of collective intelligence. Just the beginning of new classes of intelligent entities. To take advantage need to understand the possibilities at greater depth. CCI how can people and computers be connected so they collectively act more intelligently than any computer or person has before?? Analogous in field of AI- how can we design computers that are as or more intelligent than people? Suggests that his CCI question an important & understudied question.
 * Favorite example Google.** Entire system- millions of people all over the world creating pages and Google technology assembling. Amazing example of Collective Intelligence- people + technology- never existed before.
 * Wikipedia-** amazing not the technology- it's the organizational design- invented an organizational structure that let's thousands of people all over the world create a high quality product with little central control & mostly volunteers.
 * InnoCentive-**


 * SANDY PENTLAND**

How can we avoid collective idiocy? Try and move from ad hoc way we organize ourselves to something more based on data for decisionmaking Key thing in modern organization managing by looking at emails, etc But still people seek F2F- stuff that really matters in orgs doesn't find it into memos. Invisible. Can't be managed. But now can manage F2F in real time. His approach using a name badge but pays attention to who you're talking to but not your voice. Recording signals e.g. degree of animation. German Bank- ad campaign development. Shows both F2F & electronic flows can predict with good accuracy e.g. somebody getting overloaded. Can also measure quality of group interaction- can identify bottlenecks to change network of information flow - results in better collective intelligence. Personal intelligence tool that collectively produces better results. Alert people to group dynamics in real time e.g. alert to groupthink Can begin to use technology to make more formal intelligence. Prediction markets- a real problem gossip and rumour- hot tips, rumour bias the market. Identifying gossip demands having group history.
 * Sensible Organizations through Sensor Networks**
 * But can tell what's happening by looking at pattern of communications.**

Malone-- Sandy's work an organizational microscope

Stumbled into collective intelligence- Worked for GE radiology systems-- as 1996-97- competitor leveraged open source to leapfrog what GE could do. Model of self-organizing Masters in Technology & Policy-- people using open source & Linux. Couldn't understand. Used to spending big bucks on Sun server. Shifted research to puzzle of open source model of organizing being successful. Told by colleagues that open source a fad. Saw something different. 1st thing always interested in why people are working for free. In open source a tremendous heterogeneity motivations. Communities don't care about motivations. Provides hints about where take hold. Need participation architecture to attract a diverse groups. IP- who owns. Open Source- Property issues for collective intelligence will come front and center Governance- Wikipedia a flat architecture but each page a war. Open Source world provided inspiration. InnoCentive- take science products and broadcast for bounties. People successful at solving problems- not in field of expertise- bridging knowledge domains. Solutions used in back pocket. Apply knowledge in different settings. Open Source- asymmetry costs and benefits- some people can write beautiful code overnight. -- Average time spent on InnoCentive 2 weeks. Favorite example Threadless. Online T-shirt competition-- produce best scoring designs. Trajectory- 1.5 shirts - 133,300 designs- 41k designers- change notion of how a company should work. Organizational change. Do much of that work in the community.
 * Andrew Lakhani--**

e.g. prediction markets but want to generalize. Possible for simple algorithms to do a better job at predicting than experts. Where human sees algorithm leaving out a factor then humans will participate. Division of labor emerge over time.
 * Malone- CCI projects.**
 * Collective Prediction-** better ways to have groups of people predict what will happen

Is that true for groups of humans and computers. Is it true that a group of people & computers will do well on other tasks. Is there in fact "collective intelligence". A single "g" factor? If not, subsets, what causes them? Then can rewire groups. What can we do to increase their collective intelligence?
 * Measuring Collective Intelligence-**
 * Psychometrics-** Intelligence- that thing that correlates with a wide range of performance on a wide range of tasks

What are the limits to Collective Intelligence? Is there a sceptical side? Privacy issue- Pentland-- cell phone company knows a lot about you. Most orgs have name tags with RFIDS- trade off of privacy & advantage- how little of that info needs to be used to reap benefits of collective intelligence. Vision- only share unidentifiable Sensor based society we've already become. No panacea. If can detect common errors and feedback can do a better job of what we already know but address problem states.

Surowiecki doesn't say that. Not magic. Sometimes works well. Too complex Help put a more firm scientific foundation under what's CI good for and not. I**n order for it to work well need to collect right people & Computers and need to connect them in the right ways.** Pentland- build tools to improve e.g. No HBR or Sloan course on "community management"- not yet mechanisms. Technological issues- e.g. open source pharma- how would that take place? Carla Baldwin- information shadow of products. Do simulations. Scale of limitations are legal as well. How can rents/profits accumulating? e.g. Mozilla Foundation an effort.
 * Malone-** widespread magical thinking about Wisdom of Crowds.
 * Lakhani-** prediction markets in a research org. Data shows markets accurate but issue that managers don't want to use? Managers geared to be information hubs. **Organizational limits to collective intelligence**

e.g. Wikipedia. CCI- Climate Change project- using technology to collectively solve e.g. multi user simulations, collective decisionmaking to make items bubble up, + argumentation For each issue, a series of positions, and a series of arguments for & against. By structuring discussions make more effective. Lakhani- Sunlight Foundation- enable citizens to observe & give feedback CCMixter- YouTube points to potential.
 * Q: Just talking about consumer applications??**

Pentland- detect depression using patterns of socialization- Social discord- in UK can detect using cell phone using. Lagatum Center- intended to be a collective intelligence center- bring people from developing countries. come, build a network to support change, propagate best solutions Pentland- tool gives insight into Charisma- wants to immunize people against that! e.g. Boards of Director Good things can happen without little accountability & centralized control. Free markets an example e.g. cotton gets distributed Failed example, MIT, Wharton, Pearson, SharedInsights In end a team of professional writers hired to write the book. As experiment in Wikipedia book style writing failed. **Where failed not motivated enough to spend time contributing**. But many didn't have writing ability or insight. Of 4k who registered plenty who had the ability if motivated enough. 2nd step how to manage the interdependencies of different parts of the book.** Integration, flow. Not worked out enough Interdependence- how closely tied are our works important Debates about authority at heart of Wikipedia. Should we look at credentials. Confounded.
 * Lakhani-** will learn from practice to develop the theory
 * Pentland- f**or last few centuries bought into notion of individuals but much more creatures of our environment than we recognize. Focus on connections to understand what we think and how we act to increase sensitivity.
 * Malone-** chaos theory builds with emergent behavior- also applies to interesting examples of collective intelligence. CI focuses on more on the cognitive
 * Malone-** "centralized mindset" - put somebody in control.
 * Lakhani- centralize is main intuition. Instinct to decentralize will take time.**
 * Malone-** Future of Work makes the case for growth in decentralization
 * Malone-** Motivation a factor. e.g. Trust, or like contributing (Wikipedia), money
 * We are Smarter than Me-** registered 4k people. Only a few dozen contributed but most not very useful.
 * Failed at collecting the right people.
 * Lakhani-** Softwre- Village idiot is the compiler that demonstrates performance.
 * Malone-** most of what is know is in the heads of my friends. Nobel winner Simon quote
 * Pentland-** decisionmaking - thinking & not- complicated purchasing behavior- as more complicated intuition works better
 * Malone-** role of rewards in performance-
 * Pentland-** notion of a reflective aid- to see how come across to others. Anonymize data to compare to others to understand what they do
 * Lakhani-** mixed feeling about ratings. Can skew behaviors.
 * Much of the novelty coming from infrequent contributors. Regular participants polish.**
 * Pentland-** novel things don't get acceptance. e.g. NSF won't fund
 * Malone-** best selling book of all time, the Bible, written by a group

Participant- accountability in open source is very useful. Comment on challenge of participation in wikis