Fragmentation of identity through structural holes in email contacts

1
Fragmentation of identity through structural
holes in email contacts

• danah boyd, Jeff Potter, Fernanda Viegas
• (Sociable Media, MIT Media Lab)

2
Research Questions

• How does social network structure impact
  individual construction of identity?
• How is this behavior made explicit online?
• How can this be observed within the context of
  email?

3
Construction of Individual Identity

• Interrelated ideas of identity
• Social identity public presentation of self
• Internal identity private view of self
• Fragmentation vs. Multi-Faceted Identity
• Fragmentation conflicting internal identity
• Multi-faceted coherent internal identity,
  fragmented social identity

4
Managing Faceted Selves

• Differentiated presentation changed according to
  context
• How? Fashion, language, location/context, people
• Why? Privacy, social appropriateness, reputation
  differentiation
• Who? Dependent on self-monitoring habits,
  marginalization, fear of retribution
• Fragmented social network (e.g., work, clubs,
  family, )
• Separate social circles provide for segmentation
  of presentation

5
Identity online

• Confusion of context
• Ease of moving between multiple contexts
• Data aggregated across locations
• Email address serves as context
• Allows for privacy and faceted behavior

6
Relating Network Structure

• Structural holes bridges (Burt)
• Maximize control information flow
• Simmelian ties (Krackhardt)
• In public settings, personally constraining by
  restricting appropriate behavior aggregate of
  all associations
• Control of network structure
• Minimize uncontrolled personal information flow

7
Structuring social networks via email

• Recognizing the power of multiple recipients
• Copy/paste phenomenon to appear personal or
  contextual
• Slight content alternations for context
• Making others aware of audience

8
Ego-Centric Visualization

• Visualization tool to observe social networks
  embedded in email
• Focused on structure
• Analyzed Mikes email habits
• 5 years worth of complete data
• Maintains multiple email addresses for different
  contexts
• (Dis)advantages of using one persons behaviors

9
Introducing Mike

• Social characteristics
• 24-year old, gay-identified, white male
• Born in northern CA, attended Yale (art
  computer science)
• Friends jobs in Boston, SF, Chicago, NYC
• Uses many forms of media to stay connected
• Mikes primary social communities
• Family, high school friends
• Undergraduate friends
• Gay men in/outside Boston, in NYC
• Boston, Texas, California work colleagues

10
Mikes dataset

• 80,941 messages
• 1.03 average recipients per msg
• 15,537 unique people
• 7,250 people w/ 2,618 knowledge ties (excluding
  listservs)
• 662,078 ties between all respondents (using only
  messages with
  million)
• 226 trusted ties 23 reciprocal

11
Defining Connectivity

• Knowledge ties
• If A sends a message to B, A knows B
• B does not necessarily know A
• Awareness ties
• If B receives a message from A - B is aware of
  A
• If B and C both receive a message from A - B and
  C are aware of each other
• Trusted ties
• If A sends a message to B and blind carbon copies
  (BCCs) D - A knows and trusts D
• (D has the ability to respond and reveal that A
  included people without Bs awareness)

12
Visualizations

13
Visualizations Overview

• Goal is to allow one to quickly see how Mikes
  network is connected and view structural holes
• Methodology
• Spring/Wire explanation
• View of entire world
• Close-up views of network

14
Visualizations Methodology

• Basic spring/node algorithm used to place nodes
  in optimal location
• - annealing algorithms dont work with 15,000
  nodes
• Colors are used to indicate the relationship to
  the person
• - based on which of Mikes email address the
  person uses
• - most common address used

15
Visualizations Spring/Node (1/2)

• Basic spring algorithm used to place nodes
• -Ties act as springs, pulling connected nodes
  closer together
• -Nodes act like magnets and repel each other

16
Visualizations Spring/Node (2/2)

• All nodes start out at random location, spring
  algorithm is run several hundred iterations
• This (eventually) results in connected nodes
  being nearby and non-connected being far away

17

18
Visualization Entire World (1/2)

• Color key for
• all images

19

20

21

22

23

24

25
Social Implications

• Using one persons email, we can observe the
  social networks of hundreds of people - what are
  the implications of this?

26
Thoughts moving forward

• More detailed analysis
• Use visualizations to have ethnographic
  conversation with Mike
• Extend to multiple users
• Visual comparison valuable
• Allow for interactivity
• More detailed analysis of ego-centric graphs
• Learn more from social network analysts

27
http//smg.media.mit.edu/projects/SocialFragments/