Susie Wyshak's Hub

Notes from a SXSW 2009 panel run by Derek Powazek – a fabulously passionate and intelligent speaker, pure delight. Here are some things he said:

Does not mean if you put a bunch of people in a room they are smart. You have to work to enable the crowd to be smart.

Francis Galton was the first to observe wisdom of the crowds, in a guessing contest in the early part of the 1900s. Individually everyone was wrong but collectively they were nearly got the answer right.

Requirements of the crowd:

• Need diversity within the crowd
• Independence – Everyone needs to contribute for their own reasons, which avoids groupthink.
• Decentralization – NO one is in charge. You don’t need permission to post.
• Aggregation – All data goes into one place (which is where many websites fail).

How to successfully generate wisdom of crowds:

• Create small simple tasks for users - A blank comment form is not a simple task. Anything can grow (like a petri dish). There needs to be “an answer.” (For me Yelp comes to mind or review systems where you can pick an option to give your opinion then add more text to that.) Threadless rating is another example ( 0 – 5). You can’t tell people to “write stories.” It’s too “blank” like looking at a blank sheet.
  “Here’s a list of people we wish we could interview. Who should be on our list?”
  OK…now “who wants to do the interviews?”

  This worked, giving users small simple tasks.

• encourage participation by a large diverse group – Encourage new members to join. LOWER barriers to entry to encourage new voices. No easy answer on how to include a lot of people but not have lowest common denominator rule.
• very important to design for users’ selfishness – Large groups only participate if they’re getting somethng out of it (attention? release for saying something to wthe world?) When you design an interface that doesn’t take into account people’s selfish reasons for participation, you’re going to get little to no participation. (Is this worth my time? What do I get out of it?)Threadless’ submit page. Fame, friends, and chance for twenty five hundred dollars!You can create something wise and successful by using whatever users created based on their own selfish reasons.
• Aggregate the results -

The Heisenberg Problem – Showing a leaderboard based on ranking, after aggregating, may lead to people participating to WIN, not for whatever their other reasons were.

Interestingness on Flickr is really a game. The way to win is to be #1 in interestingness. This created an incentive for bad behavior where people tried to figure out what the algorithm was, spammed lots of groups with their photos etc. like SEO. So they changed views of interestingness and instead of showing rank they show them randomly.

Threadless doesn’t display results while voting is going on to avoid group think – so it’s an individual decision.

Polls that require you to submit before seeing the results achieve the same goal.

Popularity does not have to rule.

The most popular is often not the best. “Most votes” doesn’t have to mean something wins.

Seeing histogram (like Amazon and Yelp) to see the distribution of stars on reviews is a way to get around only seeing what’s the “best.”

Implicit vs. Explicit Feedback

Explicit feedback (voting and rating). Digg (+1), Neat or not, Drag bars from The Click Exhibition, Hot or Not, Threadless etc. How do you ask your audience a question for which you want explicit feedback?

Never use more options than you think you need. Do you need a 10 scale on Hot or Not?

Never ask people to do more thinking than they need to.

Use thumbs up or down if that will do the trick.

Implicit Feedback

1. Monitoring page views around an item
2. What people are searching for (predict news/popularity by analyzing)
3. Velocity – How much are things changing? (e.g. uptick in sxsw searches in March)
4. Interestingness – Favored is good since it’s not voting but like monitoring “email to a friend” function which can be like a wisdom of the crowds.You get better data not by asking them but by monitoring behavior.

Design Matters

How you ask questions changes the answers you get.

The interface for how you ask feedback produces what you get.

Color can even influence what people submit / contribute. (Kvetch.com example – his site). When he changed the colors he saw that people’s submissions became funny whereas with previous colors they were angry.

Color research:

• Blue – does better for creativity and imagination. In an ad, resesearchers found people like. Blue – creates a calm emotional response where people feel freer to create.
• Red groups – better in recall and attention to detail. Red – danger and mistakes. People try to avoid mistakes. Red creates fear response — they get very detail oriented and don’t want to mess up. You can’t focus on the ad or on being creative when you see red as you freeze. Important for visual designers. Especially important if you are asking them to do something.

Seeing Things

Wisom of Crowds provides a context that lessens what we’re each talking about. Your brain has to figure out less hard what’s going on.

At the end of the study they went back to the “out of control” group and told them to tell a story about something you’re passionate about. What is your life like? By getting users in touch with something they are passionate about, in control of, it takes away fear.

How can we bring these wisdom of crowd settings and in-control experiences that make them less nuts online?

Tags:   research · sxswNo Comments