What Communication Supports Multifunctional Public Goods in Organizations? Using Agent-Based Modeling to Explore Differential Uses of Enterprise Social Media


Jeremy Foote
Purdue University


Jeff Treem
University of Texas at Austin


Bart van den Hooff
Vrije Universiteit Amsterdam

Introduction

Enterprise Social Media software holds great promise

However, ESM launches often fail

  • Often blamed on
    • Leadership
    • Technology

There are two main types of information good

  • Connective
    • Often (nearly) synchronous
    • Often one-to-one
  • Communal (Information storage and retrieval)
    • Asynchronous
    • Non-relational
  • These are public goods (Fulk et al., 1996) and there are motivations to free ride

Enterprise social media is a mix of both types of information good

For many public goods, participation increases when there is a “critical mass”

  • As an ESM system is more used, it becomes more valuable
  • As an ESM system becomes more valuable, more people want to use it

What is the experience of members of an organization as they decide whether to use a new ESM system?

  1. Interview study
  2. Agent-based models

Interview-based Case Study

Site and Method

  • Branch of a financial company
  • Implemented an ESM 2 months earlier
    • Similar to Facebook / Twitter
  • 39 interviews
  • Open coding → axial coding → selective coding

Findings

Two stages of participation

  • Initial use
    • Determining whether the system would have value
    • Exploring, testing, and experimenting
  • Continued use
    • Integrating the ESM into daily work practices
  • Perceived critical mass determined whether people moved from intial use to continued use

Motivations for using the ESM influenced how critical mass was perceived

  • Connectivity
    • Those seeking to connect perceived whether others responded to a post or accepted friend requests
  • Communality
    • Those seeking information wanted the ESM to have content that was useful for completing tasks

Agent-based model

ABMs can complement individual-based approaches

  • Agent-based models use computational agents who perceive their environment and act according to rules
  • We can simulate how behavior would scale up
  • We can create “alternate worlds” where initial conditions differ

ABM Design

  • Agents who probabilistically
    • Log on to the site
    • Attempt to connect
    • Look for information
    • Add information

Agent Behavior

  • When attempts to connect or look for info are successful, agents increase their likelihood of being active and connecting / looking for info
  • Finding info also increases probability to contribute info
  • Info decays randomly at a constant rate

Findings

The multi-function nature of ESM can make them more likely to succeed

  • A connection-only system requires a big initial critical mass; otherwise, they quickly become a ghost town
  • Users come for information, making them available for connection

Prioritizing information or connection depends on the nature of the community

  • When information decays quickly, information searches are disappointing and people leave the system
  • If this also reduces people’s willingness to be active (as in our model), it can even kill the connective good

Lots of limitations

  • Doesn’t include structure / networks
  • Doesn’t include expertise / different types of knowledge

Suggestions for organizations

  • Better to build contributors than information
  • Focus initial efforts on creating core group of engaged contributors
  • Notifications may help to make it more probable that others will be “active” and findable

Thanks!


Jeremy Foote
Purdue University


Jeff Treem
University of Texas at Austin


Bart van den Hooff
Vrije Universiteit Amsterdam