Characterizing the Use of Browser- Based Blocking Extensions To Prevent Online Tracking

Transcription

1 Characterizing the Use of Browser- Based Blocking Extensions To Prevent Online Arunesh with Jessica Vitak, Arvind Narayanan, and Marshini Chetty

2 Online The New York Times The Telegraph

3 Online 124 third party sites! The New York Times The Telegraph

4 Blocking Extensions!4

5 Blocking Extensions AdBlock Plus AdBlock Ad blockers!5

6 Blocking Extensions AdBlock Plus AdBlock Ad blockers Ghostery Priv. Badger Disconnect Tracker blockers!6

7 Blocking Extensions AdBlock Plus AdBlock Ad blockers Ghostery Priv. Badger Disconnect Tracker blockers ublock ublock Origin Content blockers!7

8 Blocking Extensions AdBlock Plus AdBlock Blocking Mechanism Ad blockers EasyList Ghostery Priv. Badger Disconnect Tracker blockers Internal lists Heuristics ublock ublock Origin EasyList Content blockers EasyPrivacy Misc. Lists!8

9 Blocking Extensions Are Effective! ~ 66% reduction in third-party requests!9 Block Me If You Can: A Large-Scale Study of Tracker-Blocking Tools (Merzdovnik et al)

10 Missing Piece: Real-World Usage of Blocking Extensions Mental Models: Does blocking extension use relate with a greater understanding of online tracking? Motivations: What are the reasons behind users adoption of blocking extensions? Experiences: What kind of user experiences result when blocking extensions break websites?!10

11 Missing Piece: Real-World Usage of Blocking Extensions Mental Models: Does blocking extension use relate Goal: How can we with a greater understanding of online tracking? improve defenses against Intentions: What are the reasons behind users adoption of blocking extensions? online tracking? Experiences: What kind of user experiences result when blocking extensions break websites?!11

12 Method: Surveys + Measurement Participants: Amazon Mechanical Turk (N = 1000) Asked: Measured: Extensions used (Length of use, Learned about from, Reason behind adoption) Blocking trackers? Blocking ads? Experiences with website breakage (Websites experienced, Subsequent action)!12 Blocking third-party cookies? Set Do Not Track?

13 How prevalent are blocking extensions?!13

14 Reported Blocking Extension Usage Only Ad blockers 42.2% Tracker blockers 8.4% Ad blockers 51.2% Content blockers 20.5% No extensions 33.6%

15 Reported Blocking Extension Usage Only Ad blockers 42.2% Tracker blockers 8.4% Ad blockers 51.2% Content blockers 20.5% No extensions 33.6%

16 Reported Blocking Extension Usage Only Ad blockers 42.2% Tracker blockers 8.4% Ad blockers 51.2% Content blockers 20.5% No extensions 33.6%

17 Reported Blocking Extension Usage Only Ad blockers 42.2% Tracker blockers 8.4% Ad blockers 51.2% Content blockers 20.5% No extensions 33.6%

18 Reported Blocking Extension Usage Only Ad blockers 42.2% Tracker blockers 8.4% Ad blockers 51.2% Content blockers 20.5% No extensions 33.6%

19 Reported Blocking Extension Usage Only Ad blockers 42.2% Tracker blockers 8.4% Ad blockers 51.2% Content blockers 20.5% No extensions 33.6%

20 Reported Blocking Extension Usage Only Ad blockers 42.2% Tracker blockers 8.4% Ad blockers 51.2% Content blockers 20.5% No extensions 33.6%

21 Reported Blocking Extension Usage Only Ad blockers 42.2% Tracker blockers 8.4% Measured: 9.95% Ad blockers 51.2% Content blockers 20.5% No extensions 33.6%

22 Does blocking extension use relate with a greater understanding of online tracking?!22

23 Mental Models of Online Entities that Track Information Tracked Outcomes Mechanisms Comfort with!23

24 Mental Models of Online Entities that Track Information Tracked Outcomes Takeaway: Blocking extension usage only weakly relates with an advanced understanding of online tracking Mechanisms Comfort with!24

25 Mental Models of Online Entities that Track Advertisers Information Tracked Outcomes Websites Visited Government Agencies p = 0.02 Internet Service Provider Mechanisms Third-party Companies p = 0.01 Comfort with 0% 25% 50% 75% 100% Users Non-users!25

26 Mental Models of Online Entities that Track Advertisers Information Tracked Outcomes Websites Visited Government Agencies p = 0.02 Internet Service Provider Mechanisms Third-party Companies p = 0.01 Comfort with 0% 25% 50% 75% 100% Users Non-users!26

27 Mental Models of Online Entities that Track Advertisers Information Tracked Outcomes Websites Visited Government Agencies p = 0.02 Internet Service Provider Mechanisms Third-party Companies p = 0.01 Comfort with 0% 25% 50% 75% 100% Users Non-users!27

28 Mental Models of Online Entities that Track Information Tracked Visible Outcomes (e.g., targeted ads, personalization) Outcomes Invisible Outcomes (e.g., increased revenue, Mechanisms Comfort with price discrimination) p < % 25% 50% 75% 100% Users Non-users!28

29 Mental Models of Online Entities that Track Information Tracked Visible Outcomes (e.g., targeted ads, personalization) Outcomes Invisible Outcomes (e.g., increased revenue, Mechanisms Comfort with price discrimination) p < % 25% 50% 75% 100% Users Non-users!29

30 Mental Models of Online Entities that Track Information Tracked Visible Outcomes (e.g., targeted ads, personalization) Outcomes Invisible Outcomes (e.g., increased revenue, Mechanisms Comfort with price discrimination) p < % 25% 50% 75% 100% Users Non-users!30

31 Mental Models of Online Entities that Track Information Tracked Uncomfortable (suspicion of companies, data mining concerns) p = Outcomes Comfortable (nothing to hide, Mechanisms nothing is free) p = Comfort with 0% 25% 50% 75% 100% Users Non-users!31

32 Mental Models of Online Entities that Track Information Tracked Uncomfortable (suspicion of companies, data mining concerns) p = Outcomes Comfortable (nothing to hide, Mechanisms nothing is free) p = Comfort with 0% 25% 50% 75% 100% Users Non-users!32

33 Mental Models of Online Entities that Track Information Tracked Uncomfortable (suspicion of companies, data mining concerns) p = Outcomes Comfortable (nothing to hide, Mechanisms nothing is free) p = Comfort with 0% 25% 50% 75% 100% Users Non-users!33

34 What are the reasons behind users adoption of blocking extensions?!34

35 Reasons Behind Adoption Takeaway: Each extension type has a Ad blockers primary reason behind adoption: Tracker blockers User Experience: block pop-ups, reduce clutter on websites Privacy: block targeted ads and data mining Content blockers Security: block harmful & malicious content!35

36 Reasons Behind Adoption Ad blockers 100% 75% AdBlock Plus Tracker blockers 50% AdBlock Content 25% blockers 0% User Exp. Privacy Security!36

37 Reasons Behind Adoption Ad blockers 100% 75% AdBlock Plus Tracker blockers 50% AdBlock Content 25% blockers 0% User Exp. Privacy Security!37

38 Reasons Behind Adoption Ad blockers 100% 75% AdBlock Plus Tracker blockers 50% AdBlock Content 25% blockers 0% User Exp. Privacy Security!38

39 Reasons Behind Adoption 100% Ghostery Ad blockers Tracker blockers 75% 50% Priv. Badger Content 25% Disconnect blockers 0% User Exp. Privacy Security!39

40 Reasons Behind Adoption 100% Ghostery Ad blockers Tracker blockers 75% 50% Priv. Badger Content 25% Disconnect blockers 0% User Exp. Privacy Security!40

41 Reasons Behind Adoption 100% Ghostery Ad blockers Tracker blockers 75% 50% Priv. Badger Content 25% Disconnect blockers 0% User Exp. Privacy Security!41

42 Reasons Behind Adoption Ad blockers 100% 75% ublock Tracker blockers 50% ublock Origin Content 25% blockers 0% User Exp. Privacy Security!42

43 Reasons Behind Adoption Ad blockers 100% 75% ublock Tracker blockers 50% ublock Origin Content 25% blockers 0% User Exp. Privacy Security!43

44 Reasons Behind Adoption Ad blockers 100% 75% ublock Tracker blockers 50% ublock Origin Content 25% blockers 0% User Exp. Privacy Security!44

45 What kind of user experiences result when blocking extensions break websites?!45

46 Experiences with Broken Websites Takeaway: Browser extensions are effective in detecting trackers, and only rarely break websites Users subsequent reactions vary based on perceived value of content and trust in website!46

47 Experiences with Broken Websites Frequency: 40% of extension users experienced broken websites 94.6% experienced broken websites either rarely or sometimes in any given week!47

48 Experiences with Broken Websites Reported Experiences: Webpages failed to load completely and the content failed to appear 28.7% Embedded videos failed to play 24.3% Webpages appeared distorted, and the elements looked out of place 13.0% Pop-ups that drove functionality failed to appear 8.1% Images failed to load completely 7.5%!48

49 Summary Prevalence: ~1/3 of participants used no extensions; Ad blockers most prevalent but do not explicitly block trackers Motivations: Depends on extension type; primarily for user experience reasons Mental Models: Non-users and users of extensions had similar understandings of online tracking Experiences: Users rarely experience broken websites; When they do, make decisions based on trust and value of content!49

50 Moving Forward What can we do to better protect users from online tracking? Suggestion: Offload tracking protection to the browser!50

51 !51

52 !52

53 Moving Forward How can we better help users report breakage they see on websites? Suggestion: Improving the reporting interfaces of browser extensions (e.g., Directly by pointing to elements)!53

54 Characterizing the Use of Browser- Based Blocking Extensions To Prevent Online Arunesh with Jessica Vitak, Arvind Narayanan, and Marshini Chetty

55 Bonus Slides!55

56 Blocking Extension Usage Ad blockers Tracker blockers Content blockers Ad + Tracker blockers Ad + Content blockers Tracker + Content blockers Ad + Tracker + Content blockers No extensions 0% 11.25% 22.5% 33.75% 45% Blocking Trackers Not Blocking Trackers!56

57 Mental Models of Online Entities that Track Information Tracked User Information (e.g., demographics, location) Behavioral Activities (e.g., websites visited, products clicked) Outcomes Device Information (e.g., IP Addr, browser) Mechanisms Comfort with 0% 25% 50% 75% 100% Users Non-users!57

58 Mental Models of Online Entities that Track User Activities Information (e.g., clicks, history) Tracked Outcomes Cookies Mechanisms Comfort with 0% 25% 50% 75% 100% Users Non-users!58