Shb Workshop Final


Published on

slides from Security and Human Behavior workshop at MIT

Published in: Technology, Education
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Shb Workshop Final

  1. 1. Privacy, self-expression and social network sites <ul><li>Adam Joinson </li></ul><ul><li>Centre for Information Management </li></ul><ul><li>University of Bath </li></ul>
  2. 2. Acknowledgements <ul><li>Collaborators: </li></ul><ul><ul><li>Mina Vasalou </li></ul></ul><ul><ul><li>Dave Houghton </li></ul></ul><ul><li>Funders: </li></ul><ul><ul><li>ESRC </li></ul></ul><ul><ul><li>EPSRC </li></ul></ul><ul><ul><li>TSB </li></ul></ul>
  3. 5. Expressive privacy <ul><li>“ protects a realm for expressing ones self-identity or personhood through speech or activity. It protects the ability to decide to continue or to modify ones behavior when the activity in question helps define oneself as a person, shielded from interference, pressure and coercion from government or from other individuals” (DeCew, 1997, p.77). </li></ul>
  4. 6. The business case for expressive privacy <ul><li>SNS face the danger of becoming a phone book - much thumbed, little value </li></ul><ul><li>Value created by continued connections (limited growth? possible issues), by interaction (photos, apps etc). </li></ul><ul><li>This increases attention, time on site - and info. for direct marketers. </li></ul><ul><li>But, is this likely to be reduced if privacy is violated? </li></ul>Source: Fogg, B.J & Iizawa, D. (2008). Online persuasion in Facebook and Mixi: A cross-cultural comparison. Persuasive 2008, LNCS 5033, pp. 35-46.
  5. 7. Study 1: FB, Trust, Privacy <ul><li>Do people create value in the same way across countries? </li></ul><ul><li>Links between trust, privacy and value creation </li></ul><ul><ul><li>...and trust in who (site, peers?) </li></ul></ul><ul><li>N=426, 5 countries </li></ul>
  6. 13. Trust and uses 1... Trust in Facebook Trust in FB users Social connection .14** -0.06 Photographs .24** 0.02 Social investigation .15** -0.08 Content .23** -0.05 Status updates .21** 0.03 Shared identities .25** -0.03
  7. 14. Trust and uses 2... Trust in Facebook Trust in FB users Freq. site visit .16** 0.02 Hours per week .13** -0.07 Freq. status changing .24** 0.01 Length of time on site -.10* -0.05 Number of friends 0.07 -.13** Privacy settings -.17** -0.007
  8. 15. <ul><ul><li>In conditions of low trust, users engage in privacy management through (lack of) engagement with the site </li></ul></ul><ul><ul><li>permissive privacy approach associated with limited use...protective approach associated with wider use (and more friends) </li></ul></ul>
  9. 16. Study 2: Twitter and privacy management <ul><li>Aim: identify linguistic markers for intimacy, see if related to audience size... </li></ul><ul><ul><li>People are less intimate, expressive to a larger audience </li></ul></ul><ul><ul><li>Twitter norm is open...infinite audience regardless of followers... </li></ul></ul><ul><li>Part 1: Secret tweet vs. normal tweets </li></ul><ul><li>Part 2: Audience size and markers </li></ul>
  10. 17. Twitter or Secret Tweet? <ul><li>“ my husband is having a sexting affair. I am considering posting the pics she sent him to FB and tagging her in them.”… </li></ul><ul><li>“ got a running nose.. again. *sniff. sniff.*.” </li></ul><ul><li>“ I found $200 at work. I kept the money but I am afraid to spend it.”…. </li></ul><ul><li>” can think of no better way to spend this sunny afternoon than playing in a charity golf day at West Cornwall. </li></ul><ul><li>“ I think I'm better than everyone because I have a really big beard. My facial hair game is impeccable” </li></ul><ul><li>” wishing I hadn't gone to bed at 12:45 a.m., only to get up at 6 a.m.!” </li></ul><ul><li>“ Been sleeping with my former professor for 3 years now, and he's married with 2 kids, but the sex is the best I've had and still is” </li></ul>
  11. 18. Linguistic markers for intimate secrets <ul><li>Identified using LWIC analysis of secret vs. normal tweets </li></ul><ul><li>Discriminant analysis - identified 16 markers (91% successful identification of secret vs. normal) </li></ul>
  12. 19. Markers (all p <.05) Marker Normal Tweet Secret Tweet Word Count 17.30 21.1 Personal Pronoun 6.95 20.78 You 0.92 0.51 She/He 0.53 3 Articles 10.44 4.13 Past Tense 1.96 4.34 Swear Words 0.59 0.06 Family Words 0.33 2.69 Human Words 0.73 1.43 Sexual Words 0.45 1.78 Fillers 0.7 0.21 Question marks 1 0.03 All Punctuation 23.56 16.95 Exclamation marks 4.03 0.16
  13. 20. Results <ul><li>Randomly selected twitter users (n=154) </li></ul><ul><li>Audience size split at Dunbar’s number (150 +/- 50) </li></ul><ul><li>Last five tweets taken </li></ul><ul><li>No differences in markers </li></ul><ul><li>No differences in all linguistic measures </li></ul><ul><li>Evidence that privacy management via communication = banality </li></ul>
  14. 21. Summary <ul><li>Value creation in social network sites related to privacy and trust... </li></ul><ul><li>....permissive approach to privacy (either user selected, or via site design), low trust, related to limited use (and uses), and banality </li></ul><ul><li> maximize value (social, operational), you should provide an environment for expressive privacy. </li></ul>