Incentive Mechanisms for Encouraging Participation in Online Communities

1
Incentive Mechanisms for Encouraging
Participation in Online Communities 

• Julita Vassileva
• MADMUC Lab Computer Science DepartmentUniversity
  of Saskatchewan, Canada

2
(No Transcript)
3
Outline

• Motivation why is participation important?
• Comtella 2002-2007 a sharing community
• Approaches for motivating participation
• Social incentives awareness
• Social incentives status
• Extrinsic incentives power
• Intrinsic social incentives reciprocal
  relationships
• Intrinsic incentives aesthetic pleasure
• Current projects
• Conclusions

4
Online communities

• Outline
• Motivation
• Comtella
• Approaches
• Social comparison
• Rewards  
• Motivating altruistic users
• Current projects
• Conclusions

• Large interest based communities
• Usenet discussion groups
• Blogs LifeJournal, MySpace, Blogger, etc.
• Game communities e.g. World of Warcraft, Second
  Life, EverQuest
• Sharing communities filesharing (BitTorrent),
  digital photos (Flickr), bookmarks (CiteULike)
• Social networking Orkut, LinkedIn, OpenBC
• Custom-made communities for particular purpose,
  e.g. knowledge management
• Expertise finding in enterprises, or peer-help
  systems in education, e.g. I-Help
• Sharing resources, e.g. lecture notes, papers
  within a research lab/group, e.g. Comtella
• Network effects
• more users - more diverse and interesting
  materials - more users.
• less users - nothing is happening - those who
  come by chance leave
• Feedback loop!
• Hypothesis after reaching a critical mass of
  participation, the community will become
  self-sustained

5
Reaching critical mass

• By chance
• YouTube, LifeJournal, Flicker, Wikipedia,
• By purchase
• YouTube - by Google 1.6 billion US in Oct06,
• Flicker by Yahoo)
• By design
• build incentives in the software, e.g Slashdot

6
The cold start problem

• Incentives
• people try to maximize utility
• they choose to do what is rewarded
• Rewards can be different
• Extrinsic rewards money, marks
• Social rewards status, power, relations 
• Intrinsic rewards contributing to a shared
  cause, aesthetic pleasure

7
Comtella

• Outline
• Motivation
• Comtella
• Approaches
• Social comparison
• Rewards  
• Motivating altruistic users
• Current projects
• Conclusions

• History
• 2002-03 P2P (Gnutella) client for sharing
  research papers (as files) in the MADMUC lab
  (Vassileva _at_ CoopIS02, Bretzke Vassileva _at_
  UM03)
• 2004 Centralized P2P to support 4th year
  students in a class on Ethics and IT to share
  URLs related to each topic/ week) (Cheng
  Vassileva _at_ ITS2004, _at_HICSS05)
• 2005 Reimplemented as a centralized web-based
  online community (ChengVassileva, UMUAI last
  issue)

8
Comtella

• History (current)
• 2006 Comtella-D system - a discussion forum for
  the Ethics and IT class link sharing (Webster
  Vassileva, AH2006)
• 2006-07 Comtella Framework multi-community,
  muti-node
• (Muhammad Vassileva, UbiDeUM2007)

9
Comtella 2002/2003 -- supporting sharing of
papers in a lab
10
Incentive Approach

• Socialawareness

Image from depts.washington.edu/.../painting/4rev
eldt.htm
11
Theories in social psychology

• Social Conformity Asch
• People want to fit in their peer group e.g. have
  similar ideas, do similar things
• Social Comparison Leon Festinger
• People tend to compare with their peer group
• Knowing that their peers may align to them, they
  behave more responsibly, care about positive
  social image and status
• Real versus Online Communities MovieLens
  experiment

12
Community visualization
Helen Bretzke
Chris Cox
Lingling Sun
13
Lessons learned

• Too complex topics hierarchy hard to classify
  papers when sharing
• Major problems of visualization
• Shows only users that are currently online
  (emphasizes loneliness)
• Size depends on who is active at the moment
• Random graphical location, but users tried to
  interpret the position
• Hard to distinguish between sizes of stars

14
Comtella supporting a class

• Supports sharing class-related web-links (popular
  magazine articles etc.)
• Links classified according to the curriculum ?
  weekly topics
• Deployed for 3 months in 2004 in a 4th year class
  on Ethics and IT, 35 users

15
(No Transcript)
16
Incentive approach

• Explicit status

17
Social psychology again

• Theory of Discrete Emotions Fear
• When people are afraid of loosing something, they
  are very sensitive to messages related to how to
  avoid the danger

18
Incentive mechanism in Comtella

• Outline
• Motivation
• Comtella
• Approaches
• Social comparison
• Rewards  
• Motivating altruistic users
• Current projects
• Conclusions

Ran Cheng

• Rewarding participatory acts with points and
  status
• The user earns points by
• sharing new links, rating links, etc.
• Points accumulate and result in
• higher status for the user
• Memberships

Gold
10
60
Silver
Bronze
30
19
Weidong Han
20
(No Transcript)
21
Evaluation

• Subjects 35 students taking Ethics and IT (CMPT
  490) in T2 (2003/2004)
• Centralized - servants reside on a server
• redundancy is no longer needed
• participation focus is on sharing new URLs of
  articles
• Duration January 11 April 5, 2004
• 1 topic gt 1 week, except topic 6 gt 3 weeks

Without status and visualization
With status and vis
1 2 3 4 5 6
7 8 9 10
topics
22
Results 1 group contributions
  Original Contribution Original Contribution Comments Comments Ratings Ratings
  number number number
Overall 821 100 888 100 578 100
before 7 331 40.32 176 19.82 73 12.63
week 7 131 15.96 162 18.24 112 19.38
after 7 490 59.68 712 80.18 505 87.37
23
Results 2 new contributions Vs. visualization
usage
Correlation 0.66
24
Results 3 Memberships

• 58 of the users indicated that they have tried
  to upgrade their memberships.
• Most of the users checked weekly the evaluation
  of their contribution.
• The membership card was clicked 10.8 times per
  user in the last four weeks.
• 48 of the users clicked the cards every week to
  check their participation and contribution
  levels.

25
Summary - quantitative eval.

• Hierarchical memberships were very motivating
• Some users contributed many links of low quality
  (15) to maximize their nodes
• Several users commented negatively about the
  decline of quality in the end of the term
• People used more often the default view (original
  contribution).
• Very few users made the effort to make an extra
  selection to see other views
• People who used the visualization more often
  contributed more original papers

26
Lessons learned

• Multi-views are not useful
• Stars need to be more attractive
• Need to motivate social comparison in the quality
  of the contributions ? find a way to visualize
  user reputation
• User Status is very effective in increasing
  participation in sharing new papers, but
• stimulated low quality papers excessive number
  of contributions
• has to stimulate contributions early in the week

visualization
status
27
Adaptive incentive mechanism requirements

• To ensure sustainability, the incentive mechanism
  needs to
• Reward contribution of new resources, but
• Discourage excessive contribution
• Encourage timely contributions
• Encourage high quality contributions
• Ensure a way to measure the quality of
  contributions ? reward ratings

28
Incentives approach

• Introducing an Extrinsic Incentive Currency,
  which gives power to the user
• Adaptive Rewards
• Community model reflects the current needs of
  the community
• More papers or more ratings?
• Depending on the time in the week
• Individual model reflects the users recent
  tendency of contributions
• Good or bad papers and ratings?
• What number of contributions (too few / too
  many)?

29
Extrinsic incentive for rating

• Separate rewards for rating - C-points
• Currency, earned with each act of rating
• Can be invested to sponsor own links (like
  Googles sponsored links)
• Decays over time

30
Comella 2004/2005 adaptive rewards mechanism

• Points are earned by
• Contributing new links (depending on their
  quality)
• Rating other links (depending on the ratings
  quality)
• Status - weighted sum of the points earned with
  each activity.
• weights are dynamic
• current values are shown to the user,

31
(No Transcript)
32
Points for rating
33
Social visualization
Size (3) number of original
contributions Colour (4)
membership (status) Brightness (4) reputation
(quality of
contributions) State (2) offline or online
Generated using OpenGL with parameters -
size, colour, temperature/brightness ? More
realistic and attractive design
34
Visualization Final Design
35
Case study

• Comtella used in the Ethics and IT class at the
  UofS
• Jan-April 2005
• 32 students
• Two groups of 16 throughout the term
• Test - Comtella 1 with adaptive rewards
  mechanism and c-points, visualization
• Control - Comtella 2 no adaptive rewards
  mechanism, no c-points, no visualizaton
• Groups formed to have equal gender and Canadian /
  foreign representation
• Test and Control groups are separate communities
• no interaction of shared links, ratings etc.
• however, students were in the same classroom for
  lectures project teams across both groups
• We compared the numbers of contributions in each
  group (links, ratings)
• Post-study online questionnaire

36
Questions and answers (1)

• Did the users in the test group (Comtella 1) give
  more ratings?
• Yes nearly twice as much as Comtella 2 1065 vs.
  613 ratings (significant)
• Did the summative ratings in Comtella 1 reflect
  better the quality of the contributed links?
• Yes in Comtella 1, 56 (9 users) felt that the
  final summative ratings that their links received
  reflect fairly their quality, while in Comtella
  2, only 25 (4 users) thought so.
• Did the users in Comtella 1 tend to share links
  earlier in the week?
• Yes users in Comtella 1 shared 71.3 of their
  contributions in the first 3 days after
  introducing the topic users in Comtella 2 shared
  60.6 of their contributions in the first 3 days.
  The difference was significant for all topics
  and ranged between 7-14.

37
Questions and answers (2)

• Did the users in Comtella 1 participate more
  actively in general?
• Yes they read more papers (3419 vs. 2416) and
  logged in the system more frequently (1714 vs.
  982).
• Is there a significant difference in the total
  number of contributed links between the test and
  the control group?
• No 613 in Comtella 1 versus 587 in Comtella 2
• There was no excessive paper contribution in
  either case.

Cheng R., Vassileva J. (2006) Design and
evaluation of an adaptive incentive mechanism
for sustained educational online communities,
User Modeling and User Adapted Interaction, 16
(2/3), 321-348.
38
Lessons learned

• Incorporating an incentive mechanism can
  stimulate a desired behaviour in an online
  community
• the c-points stimulated ratings
• can be useful for collaborative filtering systems
• An adaptive rewards mechanism can orchestrate a
  desired pattern of collective behaviour
• the time-adaptation of the rewards stimulated
  users to make contributions earlier
• It is important to make the user aware of the
  rewards for different actions at any given time

Cheng R., Vassileva J. (2006) Design and
evaluation of an adaptive incentive mechanism
for sustained educational online communities,
User Modeling and User Adapted Interaction, 16
(2/3), 321-348.
39
Pulling in the lurkers in Comtella-D 2005/06

• Outline
• Motivation
• Comtella
• Approaches
• Social comparison
• Rewards  
• Motivating altruistic users
• Current projects
• Conclusions

• Comtella-D combines the functionality of
  Comtella (sharing links) and a discussion forum
• Applied in the Ethics and IT class 2005/2006, 20
  students

Andrew Webster
40
Online community composition
Core Membership (5-15)
Peripheral Membership (85-95)
41
Incentive approach

• Intrinsic incentive
• Immediate gratification - when users perform a
  desirable action (e.g. rating a post) reward them
  with an aesthetically pleasing effect
• Intrinsic social incentive
• Visualizing the asymmetry of interpersonal
  relations
• Expectation that users will try to correct the
  asymmetry

42
Approach

• Emphasize what is valued in the community
• Rating is important reward it (just with
  esthetically pleasing effect)
• Highly rated content is valued emphasize it
  visually
• Connect peripheral members to core - make
  relations visible
• Who reads my postings?
• Reciprocity?

43
Community energy
_at_work Energy
Stored Energy
The quick red fox jumped over the lazy red
dog. By Andrew
The quick red fox jumped over the lazy brown
dog. By Andrew
The quick red fox jumped over the lazy brown
dog. By Andrew
The quick red fox jumped over the lazy brown
dog. By Andrew
All generalizations are false, including this
one. By Mark Twain
All generalizations are false, including this
one. By Mark Twain
44

• Community visualization content
• Topics and individual postings that are rated
  higher
• appear hot, those rated lower appear cold
• colours ease navigation in the content
• aesthetically pleasing, intuitive

45
Approach

• Connect peripheral members to core
• Emphasize what is valued within the community
• Make relations visible

46
We want to connect the dots
47
Modeling relations member Visibility
Blog entry
Discussion post
Shared photos
Mr. Manhattan
cosmotron
Mr. Manhattan sees cosmotron
1.0
0.4
cosmotron doesnt see Mr. Manhattan
1.0
48
Relations visualization
(0,1)
(1,1)
You watch them
Unknowns
0.4
1.0
How much they see you
You both watch each other
They watch you
From Mr. Manhattans perspective
How much you see them
(0,0)
(1,0)
49
Approach

• Connect peripheral members to core
• Emphasize what is valued in the community
• Make relations visible

50
Relation visualization (RelaViz)
51
Study Comtella Discussions

• Online discussion forum for 2 courses
• CS 408 (N19 simulated core)
• Phil 236 (N32 peripheral members)

PHIL
Test interface
CMPT
Control interface (typical discussion forum)
52
Listing forums control interface
53
Listing forums test interface
54
Post header
Control Interface
Test Interface
55
Results
Contribution Counts Contribution Counts Contribution Counts Contribution Counts Average Access / Views Average Access / Views Average Access / Views
Group Threads Posts Comments Ratings Logins Reads Relavis
CS test 72 326 17 55 66.3 233.6 4
CS ctrl 60 299 5 11 48.6 180.2 n/a
Phil test 6 10 0 6 15.9 28.1 1.1
Phil ctrl 1 6 1 4 7.9 19.2 n/a
Significant, plt0.02
56
More results

• Counted the number of interactions between
  members of the groups core (test), core
  (control), periphery (test), periphery (control).
  
• Periphery test users interacted more often with
  the core group than periphery control users
  (plt0.01)
• Within the core group, members of the test group
  engaged in more symmetrical relations

57
Lessons learned

• Non-status and non-power based rewards can work
  well too
• Users can be motivated by
• Pleasing effects following desirable actions
• Gentle social comparison of contributions (not
  ego)
• Appropriate visualization of social relationships
  ? to stimulate reciprocation
• Modeling and visualizing relationships is
  interesting and useful

58
Claims and questions

• The users behavior can be dynamically
  orchestrated by
• Providing social awareness through visualization
• Providing explicit rewards (status, power,
  esthetic pleasure, social binding) for desired
  user activities
• Adapting the rewards according to what activities
  are currently needed most by the community.

• What should be the score?

59
Current projects

• Outline
• Motivation
• Comtella
• Approaches
• Social comparison
• Rewards  
• Motivating altruistic users
• Current projects
• Conclusions

Simulating the evolution of online communities
Modelling interpersonal relationships using
trust Trust-based formation of communities of
interest
Integrating Personal Information
Management Personal rights-management for
blogs Social annotations and tagging

• Bridging across communities of interest UMtella,
  Women in Science

Social networks for recommending content
60
Sharing /collaborating through blogs

• Why share?
• Need to be useful for self first, then to others
• Sharing own desktop? ? Personal info management
  (PIM)
• Need to be able to manage access seamlessly
• Blogs personal info space, accessible from
  everywhere
• Currently open for everyone to see (like a
  homepage)
• Managing access rights very much needed
• Who sees what? Delegating access rights to groups
• Collaborating allowing others to modify blog (a
  la Wikipedia)
• Prototype a blog system allowing users to
  manage access rights to their blogs

Indratmo
Indratmo, Vassileva (to appear) A Usability Study
of an Access Control System for Group Blogs,
Proc. International Conference on Weblogs and
Social Media ICWSM'2007, March 26-28, 2007,
Boulder, CO,
61
Bridging online communities

• Currently, online communities are islands.
• Each one has to reach critical mass to be
  sustainable
• But if there were bridges allowing users to
  seamlessly visit across,the critical mass will
  be reached with smaller number of committed users
• Also new users will find it easier to start in a
  friendlier environment
• Can we enable users to seamlessly jump across
  communities, without abandoning their old
  communities?
• Three problems
• Identity management across communities (MS
  InfoCard?)
• Translation of user data across communities
  (shared UM ontology)
• Negotiation of policies across communities
  (purpose-based /decentralized) user modelling
  to determine what user data is shared, how it is
  interpreted, what adaptations should be taken
• Exploring solutions in the Comtella system
• Mutli-community, multi-node framework users can
  create own communities
• Different user roles (owner, admin, moderator),
  status, rights and privileges
• Communities and nodes are autonomous, with own
  policies.
• http//umtella.usask.ca/um/

Tariq Muhammad
Muhammad, Vassileva (to appear) Policies and
Decentralized User Modelling in Online
Communities, Proceedings UbiDeUM workshop,
UM2007.
62
Social networks for recommending content

• Information, like diseases spreads using social
  networks
• Open model of the relationships of influence
  between users,
• allow users to add /remove people of influence
• use these relationships to recommend content
• applied to recommend RSS (lots of new items)
• Outperforms classic recommender system on static
  data
• Try it out at http//keepup.usask.ca we need
  users for evaluation!!

AndrewWebster
Webster A.S., Vassileva J. (to appear) Push-Poll
Recommender System Supporting Word of Mouth,
(to appear) in Proceedings User Modelling,
UM2007, Corfu, Greece, June 25-29, 2007.
63
Conclusions

• Outline
• Motivation
• Comtella
• Approaches
• Social comparison
• Rewards  
• Motivating altruistic users
• Current projects
• Conclusions

• Motivating participation is a virtually
  unexplored area in social computing
• On the cross-roads of
• Economics (mechanism design)
• Game theory
• Social Psychology
• HCI
• Distributed AI
• Applications in education, online communities
  and game design, web 2.0, enterprise 2.0 etc
• In this talk I presented a spectrum of approaches
• All were successful (encouraged participation)
• Choosing one is appropriate one depends on the
  community

64
More Info?http//bistrica.usask.ca/madmuc

• Google for
• MADMUC or Comtella