Today saw the IDEA2009 Social Experience Design Conference kick off in Toronto. Luke Wroblewski, Director of “Product Ideation & Design” at Yahoo, gave the first session, where he presented the results of a number of different pieces of research which showed that the way social interactions and relationships are modeled can have a large effect on the way that people behave online. He also uncovered some interesting facts along the way about which types of social model generate the most active users, and what factors influence user behaviour.
He divided social sites into five types of social model:
- No relationships declared (e.g. Amazon reviews, blog comments)
- A community site (e.g. YouTube, Yahoo Answers)
- A group (e.g Yahoo Groups or a Flickr Group)
- Symmetric/2 way “personal” relationships (e.g. Facebook, LinkedIn)
- Asymmetric/1 way/fan/follow relationships (e.g. Twitter, IM, Flickr)
The focus was on the latter four – sites where users can interact with each within that site other through messaging or collaboration, leave visible traces of behavior and manage their identity.
(It was noted that even for the first type where no relationship is declared, some grouping of users can be done by techniques such as geo-location, technology (device/OS/browser/preferences) or even by looking at browser history.)
Here are some highlights from the observations and findings presented:
- 1.8% of Wikipedia users write 70% of the articles. There is a commonly cited model for community sites like these – 1% are creators, 10% are curators, and 100% are consumers.
- On Twitter, 10% of users produce 90% of the content. In contrast, 10% of Facebook users produce 30% of the content. This suggests that among active users, 1 way relationships lead people to produce three times more (perhaps because they know their audience is interested).
- One way follow relationships like Twitter are more flexible because they support multiple relationship structures
- On average, the YouTube community produces 58,000% more content per user than Facebook!
Luke also highlighted a number of other insights gathered from the research:
- Relationship limits exist in all models. The average number of friends on Facebook is 120. 92% of those on Twitter follow less than 100 people. This is in line with Robert Dunbar’s work showing people can maintain stable social relationships with a maximum of 148 people.
- Tight-knit circles flourish. An HP Labs study showed that 92% of Twitter users have 13 “friends” (defined as two-way @reply type contact occurring twice or more)
- Communication reveals relationships. The same HP study showed that Twitter “friends” (as above) relationships were 90% reciprocated. Studies have shown 95% accuracy for detecting real friends using mobile call logs and location records. Interfaces such as Motoblur and HTC Hero’s friend aggregation interface are starting to arrange communications around relationships.
- More attention, more contribution. The number of daily Twitter updates increases from 3 to 6 when followers increased by 1000. Yahoo Answers showed a 480% increase in contributions when the number of relationships increased to 20.
- Less attention, less contribution. Similarly, as interest(followers) reduced, contributions become less frequent.
- There is a limit to contribution in all models. More relationships than 20 on Yahoo Answers did not lead to an increase in contribution. Similarly gaining several thousand Twitter followers does not lead to additional increases in contributions.
- 1-way relationships optimize for reach. In other words people are more willing to commit to 1 way relationships that allow them to stay connected with topics or people of interest via a lightweight subscribe. Facebook Fan pages are another example of this.
- Real relationships drive production. People are more likely to collaborate online and produce content with “real life friends” thanks to something he calls the “0-1-2” effect.
- Better tools yield more content of a higher quality. MySpace gives poor control and so people invest less. The game Spore owes a great deal of its success to the tools it gives to users to create something attractive and interesting.
Luke also observed that current relationship models are crude. He expects that in the future we will be able to do things like “follow Todd but only his thoughts on prototyping, not his fashion tips”