Sunday
11:30-12:30
Description: People are often dumb, so how can crowds be wise? James Surowiecki laid the groundwork in his book, "The Wisdom of Crowds." In this solo presentation, Derek Powazek will apply those ideas to the web, concentrating on how to design websites that empower people to work together to create something truly awesome.
Four Concepts
1. Focus on small simple tasks
Good examples: Hot or Not or Threadless
Bad example: Assignment Zero (started by Wired(), told users to go write magazine articles; then refocused in smaller tasks, such as who should we interviews, pick someone to interview, etc.
2. Large Diverse Groups
Important because they help avoid group think. Online communities often do this because they tend to raise their barriers of entry over time. Example: Chevy Tahoe campaign. Powacek claiming this as failure in group think ... but the ir sales went up!
3. Design for Selfishness
Large groups of people do not participate unless they are getting something out of it. You must take into account people's selfish needs. What are they getting out of this? Why should they bother? Think about hoe people tag photos. People are just trying to get to their photos, not build the semantic web. If properly harnessed our selfishness can become smart for business.
4. Result Aggregation
For example favrd.com. Takes flickr tags and aggregates them daily. Surfaces stuff the crown likes without asking them.
The Heisenberg Problem
Say we do these four things and it produces a list of "best" things. Once we surface to their world we are showing a leaderboard pf a game ... and then we are developing a reason to participate in the "game."
Flickr has interestingness and assigns a number to it. They created a game. But screwed up ranking becuase it created an incentive for bad behavior ion the site. People then try to get to the top, game th e system, etc. This is the tension between showing the wisdom of crowds and creating bad incentives. Now they just show a random mix of interestingness. They took away the ranked list and it becomes a better display. The interface can change how it works.
In Threadless, you can vote for a week and then they display the results -- but not while voting is going on. That way you avoid the group think.
Popularity does not have to rule. It's an easy mistake when you talk about voting and rating to surface that data. But the most popular thing is not always the best thing. Most votes does not always win.
On Amazon, you see a best and worst review side by side (an d the one voted most helpful). That is a good Wisdom of Crowds approach. .They also show a histogram of reviews and show a small visual element so you can get quick insight from the crowd.
Implicit vs Explicit Feedback
Explicit feedback: voting and rating. How do you ask your audience for feedback? It depends. Different interfaces work for different projects. Never ask people to do more thinking than they have to. If you can not think of a reason why you should not use a thumbs up or thumbs down, then just use the thumbs.
Implicit feedback: Monitoring pageviews, monitoring searches, velocity (how much is something changing), interestingness (algorithms for monitoring stuff around the data). You sometimes get better data when you don;t ask you just observe.
Design matters.
How you ask questions changes the answers you get. The interface for how you collect feedback changes your site, sometimes in subtle ways (such as color). Example, kvetch.com. Powacek changed the color of the site and the comments changed. There was recently a study in Science Magazine that showed that people had different responses to different colors of borders. Color affected response.
Putting it All Together
Brooklyn's Museum Click! exhibition. Had an online gallery where people can vote on photos. First, they asked raters to rate themselves regarding how serious they were about art. Users can toggle through the results via the self rating. And you can see which photos appeared across all groups. They did an actual gallery display sized by their votes.
GetSatisfaction. Offers number of ways to aggregate feedback as well as mood. Numerous little Wisdom of Crowds techniques. Overall creates better experience.
Seeing Things
Grandma Powacek stopped making cucumber salad because it makes her hands hurt (it was the cucumber's fault not that she had arthritis). "Mother stories." Our brains are good at taking diverse feedback and turning it into a story that we tell ourselves. We fill in gaps to make the story make sense. (Listen to Radiolab podcast)In the online space all of these gaps are filled in by data (because we don't have body language, sound, smell, etc.) Often what we fill it in with things based on our insecurities.
Talking about study about in control and out of control groups. When you;re feeling out of control you make up stories to connect the dots which have nothing to do with reality. Similar in the online space because we are deprived of social data and our brains our filling in the rest. It says more about us than what is happening (it's why people are insane on the Internet!).
WOC systems work because they provide context, less mystery, brain has to work less hard to figure out what is going on. Simply getting people to get in touch they feel in control of changes the experience.
Challenge is how can we bring these WOC settings to the community settings that we create? How can we give people in-control experience that make them a little less crazy online.
0 comments:
Post a Comment