Preference-based Location Sharing: Are More Privacy Options Really Better?
0 likes748 views
1. The document examines how adding and removing location sharing options affects user preferences. It studies four options: nothing, city, city block, and exact location. 2. The researchers hypothesize that removing the "city" option will either cause users to choose more private options proportionally, or to shift more towards the more revealing "block" option, depending on how close "city" is perceived to the other options. 3. A user study found that when "city" was removed, the share of users choosing "block" increased significantly, suggesting "city" is perceived closer to "block" than "nothing." Adding an "exact" option caused proportional increases across options, suggesting equal perceived distances.
![INFORMATION AND COMPUTER SCIENCES
Plotting the privacy calculus
For each recipient, rate
privacy and bene鍖t (-3 to 3):
I do not want [recipient] to
know where I am (Privacy)
My location could be useful
for [recipient] (Bene鍖t)
- For apps/coupons: I could
bene鍖t from [recipient]
Plot the average Privacy and
Bene鍖t of each option on a
plane
E
B
N
privacy -->
bene鍖ts-->
C](https://image.slidesharecdn.com/chipresentation-130506073022-phpapp01/85/Preference-based-Location-Sharing-Are-More-Privacy-Options-Really-Better-13-320.jpg)
![INFORMATION AND COMPUTER SCIENCES
Plotting the privacy calculus
For each recipient, rate
privacy and bene鍖t (-3 to 3):
I do not want [recipient] to
know where I am (Privacy)
My location could be useful
for [recipient] (Bene鍖t)
- For apps/coupons: I could
bene鍖t from [recipient]
Plot the average Privacy and
Bene鍖t of each option on a
plane
E
B
N
privacy -->
bene鍖ts-->
C](https://image.slidesharecdn.com/chipresentation-130506073022-phpapp01/85/Preference-based-Location-Sharing-Are-More-Privacy-Options-Really-Better-14-320.jpg)
![INFORMATION AND COMPUTER SCIENCES
Plotting the privacy calculus
For each recipient, rate
privacy and bene鍖t (-3 to 3):
I do not want [recipient] to
know where I am (Privacy)
My location could be useful
for [recipient] (Bene鍖t)
- For apps/coupons: I could
bene鍖t from [recipient]
Plot the average Privacy and
Bene鍖t of each option on a
plane
E
B
N
privacy -->
bene鍖ts-->
C](https://image.slidesharecdn.com/chipresentation-130506073022-phpapp01/85/Preference-based-Location-Sharing-Are-More-Privacy-Options-Really-Better-15-320.jpg)
![INFORMATION AND COMPUTER SCIENCES
Plotting the privacy calculus
For each recipient, rate
privacy and bene鍖t (-3 to 3):
I do not want [recipient] to
know where I am (Privacy)
My location could be useful
for [recipient] (Bene鍖t)
- For apps/coupons: I could
bene鍖t from [recipient]
Plot the average Privacy and
Bene鍖t of each option on a
plane
E
B
N
privacy -->
bene鍖ts-->
C
High privacy, low bene鍖t](https://image.slidesharecdn.com/chipresentation-130506073022-phpapp01/85/Preference-based-Location-Sharing-Are-More-Privacy-Options-Really-Better-16-320.jpg)
![INFORMATION AND COMPUTER SCIENCES
Plotting the privacy calculus
For each recipient, rate
privacy and bene鍖t (-3 to 3):
I do not want [recipient] to
know where I am (Privacy)
My location could be useful
for [recipient] (Bene鍖t)
- For apps/coupons: I could
bene鍖t from [recipient]
Plot the average Privacy and
Bene鍖t of each option on a
plane
E
B
N
privacy -->
bene鍖ts-->
C
Low privacy, high bene鍖t](https://image.slidesharecdn.com/chipresentation-130506073022-phpapp01/85/Preference-based-Location-Sharing-Are-More-Privacy-Options-Really-Better-17-320.jpg)
More Related Content
More from Bart Knijnenburg (10)
Recently uploaded (20)
Preference-based Location Sharing: Are More Privacy Options Really Better?
- 1. Preference-based Location Sharing Are More Privacy Options Really Better? Bart P. Knijnenburg Department of Informatics, UC Irvine Alfred Kobsa Department of Informatics, UC Irvine Hongxia Jin Samsung R&D Research Center
- 2. INFORMATION AND COMPUTER SCIENCES Outline Should sharing pro鍖les be simple or complex? What happens when you add/remove sharing options? Users choose based of their perception of the options A method for designing better sharing options
- 3. INFORMATION AND COMPUTER SCIENCES Pro鍖le-based location sharing Figure 5 Locaccino privacy settings page We address in detail the topic of privacy policies, presenting the necessary formalism in section 3.1.2. Who Can Locate Me This page is intended to provide privacy awareness to the user, in the sense that the
- 4. INFORMATION AND COMPUTER SCIENCES Why so complex?
- 5. INFORMATION AND COMPUTER SCIENCES Locaccino researchers: -Users will otherwise err on the safe side Decision scientists: -Depends on perception of the options
- 6. (part 2/3) Preferences for sharing your location with friends Use the demo phone on the right to set your sharing preferences for different people at different times. Note that there are several pages of preferences, and on some pages you may have to scroll down. On the different pages you will find settings for sharing with: Your friends Your colleagues Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Steven, Frank During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Continue
- 7. (part 2/3) aring your ds t your sharing preferences for references, and on some pages tings for sharing with: Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Steven, Frank During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Continue (part 2/3) aring your ds t your sharing preferences for references, and on some pages tings for sharing with: Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Steven, Frank During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Continue (part 2/3) aring your ds t your sharing preferences for references, and on some pages tings for sharing with: Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Steven, Frank During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Continue
- 8. Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ages Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs:
- 9. Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ages Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ages Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ? ? 1. removing the City option
- 10. Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ages Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ? ?2. introducing the Exact option
- 11. INFORMATION AND COMPUTER SCIENCES Without Exact (E) With Exact (+E) Without City (C) With City (+C) Main manipulation g preferences for and on some pages ring with: Best friends John, Dave, Ethan During work hrs: nothing city city block exact location r es Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location es Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Steven Frank During work hrs: Outside work hrs: John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Steven Frank During work hrs: nothing Outside work hrs: nothing and on some pages ring with: Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Frank During work hrs: ring with: John, Dave, Ethan During work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Frank During work hrs: nothing g preferences for and on some pages ring with: Best friends John, Dave, Ethan During work hrs: nothing city city block exact location g preferences for and on some pages ring with: Best friends John, Dave, Ethan During work hrs: nothing city city block exact location g preferences for and on some pages ring with: Best friends John, Dave, Ethan During work hrs: nothing city city block exact location g preferences for and on some pages ring with: Best friends John, Dave, Ethan During work hrs: nothing city city block exact location
- 12. INFORMATION AND COMPUTER SCIENCES Plotting the privacy calculus We believe that people choose based on their perception of the options How do people perceive these options? Privacy calculus: Trade-off between privacy and bene鍖t E B N privacy --> bene鍖ts--> C
- 13. INFORMATION AND COMPUTER SCIENCES Plotting the privacy calculus For each recipient, rate privacy and bene鍖t (-3 to 3): I do not want [recipient] to know where I am (Privacy) My location could be useful for [recipient] (Bene鍖t) - For apps/coupons: I could bene鍖t from [recipient] Plot the average Privacy and Bene鍖t of each option on a plane E B N privacy --> bene鍖ts--> C
- 14. INFORMATION AND COMPUTER SCIENCES Plotting the privacy calculus For each recipient, rate privacy and bene鍖t (-3 to 3): I do not want [recipient] to know where I am (Privacy) My location could be useful for [recipient] (Bene鍖t) - For apps/coupons: I could bene鍖t from [recipient] Plot the average Privacy and Bene鍖t of each option on a plane E B N privacy --> bene鍖ts--> C
- 15. INFORMATION AND COMPUTER SCIENCES Plotting the privacy calculus For each recipient, rate privacy and bene鍖t (-3 to 3): I do not want [recipient] to know where I am (Privacy) My location could be useful for [recipient] (Bene鍖t) - For apps/coupons: I could bene鍖t from [recipient] Plot the average Privacy and Bene鍖t of each option on a plane E B N privacy --> bene鍖ts--> C
- 16. INFORMATION AND COMPUTER SCIENCES Plotting the privacy calculus For each recipient, rate privacy and bene鍖t (-3 to 3): I do not want [recipient] to know where I am (Privacy) My location could be useful for [recipient] (Bene鍖t) - For apps/coupons: I could bene鍖t from [recipient] Plot the average Privacy and Bene鍖t of each option on a plane E B N privacy --> bene鍖ts--> C High privacy, low bene鍖t
- 17. INFORMATION AND COMPUTER SCIENCES Plotting the privacy calculus For each recipient, rate privacy and bene鍖t (-3 to 3): I do not want [recipient] to know where I am (Privacy) My location could be useful for [recipient] (Bene鍖t) - For apps/coupons: I could bene鍖t from [recipient] Plot the average Privacy and Bene鍖t of each option on a plane E B N privacy --> bene鍖ts--> C Low privacy, high bene鍖t
- 18. Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ages Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ages Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ? ? 1. removing the City option
- 19. INFORMATION AND COMPUTER SCIENCES Decision theory hypotheses: Three situations: 1. City is subjectively distinct from Nothing and Block: -Luces Choice Axiom holds -The ratio N : B+E will stay the same Users will prefer both sides proportionally when City is removed E B N privacy --> bene鍖ts--> C privacy -->
- 20. INFORMATION AND COMPUTER SCIENCES Decision theory hypotheses: 2. City is subjectively closer to Nothing than to Block: -Tverskys Substitution Effect holds -C is a substitute for N -When we remove C, N will increase more than B+E Users will err on the safe side E B N privacy --> bene鍖ts--> C
- 21. INFORMATION AND COMPUTER SCIENCES Decision theory hypotheses: 3. City is subjectively closer to Block than to Nothing: -Tverskys Substitution Effect holds -C is a substitute for B -When we remove C, B will increase more than N Users will prefer the more revealing side E B N privacy --> bene鍖ts--> C
- 22. Results Without Exact (E) With exact (+E) Withoutcity(C)Withcity(+C) 48.8 51.2Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> Privacy --> 26.7 29.9 18.6 24.8 Bene鍖t--> Privacy --> Exact Block City NothingWithout Exact (E) With exact (+E) Withoutcity(C)y(+C) 48.8 51.2 Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> 26.7 29.9 18.6 24.8 Bene鍖t--> Exact Block City Nothing
- 23. Without Exact (E) With exact (+E) Withoutcity(C)Withcity(+C) 48.8 51.2Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> Privacy --> 26.7 29.9 18.6 24.8 Bene鍖t--> Privacy --> Exact Block City Nothing Results Size: percentage of people choosing this option
- 24. Results W Without Exact (E) With exact (+E) Withoutcity(C)Withcity(+C) Privacy --> Privacy --> 48.8 51.2Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> Privacy --> 26.7 29.9 18.6 24.8 Bene鍖t--> Privacy --> Position: perception in terms of privacy and bene鍖t
- 25. Results Without Exact (E) With exact (+E) Withoutcity(C)Withcity(+C) 48.8 51.2Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> Privacy --> 26.7 29.9 18.6 24.8 Bene鍖t--> Privacy --> Exact Block City Nothing left vs. right: comparison of without vs. with Exact
- 26. Results Without Exact (E) With exact (+E) Withoutcity(C)Withcity(+C) 48.8 51.2Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> Privacy --> 26.7 29.9 18.6 24.8 Bene鍖t--> Privacy --> Exact Block City Nothing bottom vs. top: comparison of with vs. without City
- 27. Without Exact (E) With exact (+E) Withoutcity(C)Withcity(+C) 48.8 51.2Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> Privacy --> 26.7 29.9 18.6 24.8 Bene鍖t--> Privacy --> Results City is closer to Block (0.45pts) than to Nothing (2.25pts) Substitution effect! Only the share of Block differs signi鍖cantly between C and +C (24.2pp) Exact Block City Nothing
- 28. Without Exact (E) With exact (+E) Withoutcity(C)Withcity(+C) 48.8 51.2Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> Privacy --> 26.7 29.9 18.6 24.8 Bene鍖t--> Privacy --> Results Distance from City to Nothing and Block is more equal Luces choice axiom! Both Nothing (14.3pp) and Block (9.1pp) differ between C and +C Effect on Nothing is larger, because City is somewhat closer to Nothing Without Exact (E) With exact (+E) Withoutcity(C)y(+C) 48.8 51.2 Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> 26.7 29.9 18.6 24.8 Bene鍖t--> Exact Block City Nothing
- 29. Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ages Best friends John, Dave, Ethan During work hrs: nothing city city block exact location Outside work hrs: nothing city city block exact location Other friends Ben, Robert, Thomas, Mike, Paul, Stev Frank During work hrs: Outside work hrs: ? ?2. introducing the Exact option
- 30. INFORMATION AND COMPUTER SCIENCES Decision theory hypotheses: Two situations: 1. Exact is subjectively close to Block: -Tverskys Substitution Effect holds -E is a substitute for B Only B will decrease when E is introduced E B N privacy --> bene鍖ts--> C
- 31. INFORMATION AND COMPUTER SCIENCES Decision theory hypotheses: 2. Exact is more distant: -Simonsons Compromise Effect holds -B is no longer an extreme, but a compromise -B attracts some from C and N -Still some Substitution of B to E In sum: Sharing increases across the board B N privacy --> bene鍖ts--> C E
- 32. INFORMATION AND COMPUTER SCIENCES W Without Exact (E) With exact (+E) Withoutcity(C)Withcity(+C) Privacy --> Privacy --> 48.8 51.2Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> Privacy --> 26.7 29.9 18.6 24.8 Bene鍖t--> Privacy --> Results Exact is subjectively close to Block Substitution effect! the availability of Exact mainly affects the share of Block (21.5pp) Without Exact (E) With exact (+E) Withoutcity(C)Withcity(+C) 48.8 51.2 Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> Privacy --> 26.7 29.9 18.6 24.8 Bene鍖t--> Privacy --> Exact Block City Nothing
- 33. INFORMATION AND COMPUTER SCIENCES W Without Exact (E) With exact (+E) Withoutcity(C)Withcity(+C) Privacy --> Privacy --> 48.8 51.2Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> 40.432.6 26.9 Bene鍖t--> Privacy --> 26.7 29.9 18.6 24.8 Bene鍖t--> Privacy --> Results Exact is further away from Block Compromise effect! Block is only reduced by 8.3pp, as it regains some share from Nothing (reduced by 13.7pp) Without Exact (E) With exact (+E) Withoutcity(C) 48.8 51.2 Bene鍖t--> Privacy --> 41.0 29.7 29.3 Bene鍖t--> Privacy --> Exact Block City Nothing
- 34. INFORMATION AND COMPUTER SCIENCES Two conclusions 1. With fewer options, users do not just err on the safe side Instead, they deliberately choose the subjectively closest remaining option 2. An extreme option does not just increase sharing among those who already share a lot Instead, it increases sharing across the board
- 35. INFORMATION AND COMPUTER SCIENCES Applying these results Problem: when designing the optimal set of options, users choice depends on the available options Problem for user tests! Combinatorial explosion of the experimental conditions More efficient approach: -Ask a sample of users about the perceived Privacy and Bene鍖t of the options -Map them on a plane
- 36. INFORMATION AND COMPUTER SCIENCES Applying these results Bene鍖t--> Privacy --> 6. Introduce to increase overall sharing 2. Remove one redundant option 1. Remove this dominated option 3. Introduce an option to close this gap 4. If desired, remove the left option to increase the right option 5. If desired, replace this option with one that is perceptually close
- 37. INFORMATION AND COMPUTER SCIENCES Take-home message Do you want to design an optimal list of (location-sharing) options? Measure the subjective perceptions of the options, and apply decision theories to decide which are best! Paper draft: http://bit.ly/chi2013privacy More papers: www.usabart.nl Follow me on Twitter: @usabart
- 38. INFORMATION AND COMPUTER SCIENCES Guide for questions: 1. With fewer options, users do not just err on the safe side Instead, they deliberately choose the subjectively closest remaining option 2. An extreme option does not just increase sharing among those who already share a lot Instead, it increases sharing across the board 3. If you want to design the optimal set of options, measure the perception of the options And apply decision theories to argue which are best