Knowledge from Crowds – Be1er with Ins6tu6ons + Algorithms h1p://goo.gl/q1DNL Shaun Abrahamson @shaunabe
Jan 13, 2015
Knowledge from Crowds – Be1er with Ins6tu6ons + Algorithms h1p://goo.gl/q1DNL Shaun Abrahamson @shaunabe
Tap the crowd for learning
Where did Stanley come from?
Crowd data recycled into knowledge
Collec6ve contribu6ons into holis6c understanding
Compe6ng for be1er predic6ons
Our signals into rankings
New type of ins6tu6on to deliver mobile services
Labo
r
Capita
l
Assets
Influ
ence
Data
The crowd as the gateway to critical resources
Sharing, Commenting, Reviewing Open Innovation, Co-Creation, Micro Tasks
Labo
r*
Capita
l
Assets
Influ
ence
Data
“Big Data”
* very often “knowledge” work
Labo
r / Kn
owledge Work
Capita
l
Assets
Influ
ence
Data
Institutions vs Algorithms
4 PERSPECTIVES!!
OUTCOMES!PEOPLE!TOOLS!
ORGANIZATION!
OUTCOMES!
Ronald Coase “Given that produc6on could be carried on without any organiza6on that is, firms at all, why and under what condi6ons should we expect firms to emerge?” About 75 years ago
Why do we organize work in a certain way?
New Ins6tu6ons + tools
! Sales!
! Marketing! Operations! Production!R + D!
New data + algorithms
! Sales!
! Marketing! Operations! Production!R + D!
CASE:!What went wrong at Wikipedia?!
Source: Mary Meeker presentation at All things D. ! The end of all other encyclopedia business models
Why aren’t editors staying?
Active E
ditors R
eten
tion
afte
r 1 y
ear (
%)
PEOPLE!!
FINDING AND MOTIVATING THE MOST IMPORTANT KNOWLEDGE RESOURCES!
What skills might you tap into?
175 million people on !
Why will they par6cipate?
Money!
Experience!Attention!
Good!
Stuff!
What data by-‐products might you rely on?
Why did you start contribu6ng to Wikipedia?
What kind of work environment do you want ?
ORGANIZATION!!
WHAT INSTITUTITIONS ARE CRITICAL TO BENEFIT
FROM CROWD LABOR + INFLUENCE!
Elinor Ostrom “It's a problem, it's just not necessarily a tragedy ... The problem is that people can overuse [a shared resource], it can be destroyed, and it is a big challenge to figure out how to avoid that.” About 2 years ago
Organizing to resolve the “Tragedy of Commons”
Who owns what? Brand vs IP vs Confiden6ality
Cheap access to dispute resolu6on
Collec6ve choice processes
Sanc6on bad behavior -‐ Don’t feed the trolls
Nes6ng to scale
How Wikipedia community sees itself
The community's role, as some kind of nebulous science-‐fic6on super-‐en6ty, is to: + Organize and edit individual pages + Structure naviga6on between pages + Resolve conflict between individual members + Re-‐engineer itself -‐-‐ crea6ng rules and pa1erns of
behavior There are other roles J Source -‐ h1p://meta.wikimedia.org/wiki/The_Wikipedia_Community
TOOLS!!!
COLLECTING DATA AND CREATING NEW UNDERSTANDING!
Making it easier to contribute
Understanding individual contribu6ons
My Klout! My Giving (Crowdtwist)!
My Creative Impact(Jovoto)!
Understanding collec6ve health and performance
Rocket Science vs People Science?
Making sense of all that data
Wikimedia founda6on’s focus People + Tools
CASE: EDX.ORG!!
NEW EDUCATIONAL INSTITUTIONS !
+ !DATA TO GET SMARTER
ABOUT EDUCATION!
Content + Community = Learning
Represents about 40 years worth of classes at MIT
155,000 registered!23,000 tried the first problem set!9,000 passed the midterm!7,157 passed the course!
Ricardo + Arthur doing “online learning”
Community not just content
“One of the best things about 6.002x was the community built by the students themselves. The atmosphere was great: people shared their enthusiasm and knowledge, and lended a hand to those like me who didn’t have the basics for the course.” - Arthur Amaral, 18 years old, Brazil Source: http://blog.edx.org/
Ins6tu6ons -‐ Nes6ng + Collec6ve Choice
Anant Agarwal “We can watch how many a1empts students made before they got an exercise right, and if they got it wrong, what they used to try to find a solu6on. Did they go to the textbook, go back and watch the video, go to the forum and post a ques6on?” About 1 month ago
Data to learn how to teach
CASE: GIFFGAFF [TELEFONICA]!
!NEW INSTITUTIONS!
TO BENEFIT FROM THE KNOWLEDGE OF !
CUSTOMERS!
Social produc6on for a complex service
Homepage hints at how this works
What tasks can be performed?
Who is doing what on GiffGaff
Value Created!
Income/Expenses!
! Sales!
! Marketing! Operations! Production!R + D!
From sales + support to new app development
Encouraging Par6cipa6on + Rewarding Behavior
Self policing mechanisms
Gaming the system -‐ posts from users who you suspect are abusing the payback system by using mul6ple accounts to give themselves solu6ons or kudos. Tou0ng for SIMS/Kudos -‐ posts which are ac6vely asking for kudos or solu6ons, it is fine to have this in your signature but not to ask in a post/topic. Incorrect Accepted Solu0ons -‐ if you spot an accepted solu6on which is incorrect or if a user has accepted one of their own responses as a solu6on unjus6fiably. Incorrect Tags -‐ If you see that a post has been tagged with an irrelevant or inappropriate tag. Inappropriate Content -‐ Posts which are disrespecmul to other users, profanity, adver6sing, naming and shaming and generally causing discord or disharmony on the forum.
How is this growing? Also NPS = 73 (Apple = 79)
Labo
r / Kn
owledge Work
Capita
l
Assets
Influ
ence
Data
Institutions vs Algorithms
Anant Agarwal “We can watch how many a1empts students made before they got an exercise right, and if they got it wrong, what they used to try to find a solu6on. Did they go to the textbook, go back and watch the video, go to the forum and post a ques6on?” About 1 month ago
Data + Algorithms for Knowledge Management
Elinor Ostrom “It's a problem, it's just not necessarily a tragedy ... The problem is that people can overuse [a shared resource], it can be destroyed, and it is a big challenge to figure out how to avoid that.” About 2 years ago
Community + Institutions for Knowledge Management
Knowledge from Crowds – Be1er with Ins6tu6ons + Algorithms h1p://goo.gl/q1DNL Shaun Abrahamson @shaunabe