This session included a wide range of topics from agent based modeling
to social media and crowd sourcing to prediction markets. 

Agent based modeling:
Michael Wellman will provide a short summary of ongoing research.

Anderson described a potentially promising approach to utilize 
social networking platforms to obtain both 'opinion' and personally
identifiable information (PII). 
A prototype is the U.K.-based “Voice Your View” www.voiceyourview.com project
which has successfully collected public opinion data via telephone and web-based 
tools.  Data collection has been extended to include a web-based system with a 
Facebook-style API.
It has been used to obtain opinion of household expectations of future inflation.
Statistical analysis in which the PII are treated as covariates promises more
accurate inferences than analysis of opinions alone.  These methods also
provide a data collection protocol that is very similar to panel based
temporal approaches.

Research questions:

- Can signal from social media provide similar quality information
  compared to a large panel and repeated data collection?  For example,
  can it accurately determine correlations?

- Can we measure or quantify when domain experts are preferred to the
  'wisdom of the crowds'?

- What are the factors that impact diffusion in social media?  Is the 
  source and timing of the information important? Are there identifiable
  patterns of diffusion?

- Are there some specific cases where we can demonstrate some fundamental 
  defect or shortcoming in modeling from social media datasets?

- Can social media datasets mimic census data?

- Are there opportunities to determine the impact or synergies across
  different crowd sourced platforms?  For example, are there potential 
  synergies between the signal and diffusion of information from Twitter,
  and the features that impact the outcome of a prediction market?