Karrie Karahalios addresses Text and Tie Strength at today’s Berkman Center lunch. I attended a talk she gave a MIT a few weeks ago, but it appears I did not blog my notes from that event. Being able to hear a different presentation about her research is fortunate. Using tools, she creates visualizations of conversations and interactions, breaking everything down into colored bars, sometimes with words. In some cases, looking at these kinds of images of meetings often helps people remember more than reading notes.
Karrie’s work centers on how people use communication technology, particularly differences in Internet use between people in rural and urban environments. She introduced us to her early experience with the telephone: the village where she lived as a young girl and one phone in the local pub where she would go each Sunday to take a call from her father. American rural people were more likely to have telephones than urban folks before 1920 and more likely to have a party lin—a shared phone line.
Since people in these two environments use technology slightly differently, is it worth developing different tools for people based on whether they’re rural or urban?
She showed a visualization of her Facebook connections which looks like a question mark with several tight clusters throughout and one disconnected triangle at the bottom. That triangle represents her three Greek relatives with computers, who are not connected to anyone else in her network. By discussing this picture, she moves into talking about Granovetter’s strong and weak ties. She observes how Facebook doesn’t adequately allow people to show the strength of ties. A woman she’s met once in real life and communicates with sporadically appears the same on her friend’s list as her husband.
Granovetter’s intensity, intimacy, duration, & services
Wellman’s emotional support
and two other people who suggested tie quality
social distance, structure, services
Karrie and her graduate assistant (whom she named several times, but whose name I’ve forgotten before getting to type it—my apologies) designed an experiment where they ask various questions about the strength of their relationships with the friends listed in their Facebook profiles. The questions ranged from the nature of their in-person relationship to whether they’d use the person to find a job to whether they’d bring the person with them if they switched social networking services. At this point, she gave a nod to Friendster and wondered aloud what happened when people stopped using it: do the relationships vanish or did people show up with the same clusters on a different service?
(She’s moving very quickly and I’m lagging a bit.)
Someone summarized her approach by saying she’s using the quantitative bits of Facebook to research the qualitative aspects of someone’s connections.
As she’s examined how people use Facebook, she’s found that many often begin by communicating with their strongest ties, but not always. While she likes to think she has a strong tie with her husband, Facebook isn’t their primary mode of communication, so she doesn’t regularly contact him through that service. Many people in her study of 35 users with more than 2,000 ties total communicate with people on their friends lists slightly more frequently than once a day.
(My computer keeps crashing.)
One criticism she has for Facebook is how it does not adequately allow users to indicate their ties to people. Everyone must fit into Facebook’s set categories. Karrie and her graduate student have created a model allowing people to easily adjust their personal relationships by dragging and dropping people’s photos into various groups.
The model she and her graduate student Eric Gilbert developed for Facebook, they tried applying to Twitter. They want a model that will work for various services. She suggests an interface where people can easily drag and drop people’s icons/avatars to different categories instead of having to edit each user via difficult-to-use screens. She suggests customizability so people can create their own number of categories and call them what they want. She calls the groups things like “Birds of a Feather,” “Research,” “We meddle,” etc. Icons on the left indicate the strength of her ties with the people with the list being sorted so stronger ties appear above those with weaker ties. People who tweet a lot might have one visible post with a number next to it indicating how many other tweets the person has made recently.
Their model wemeddle.com) automatically sorts people into categories and groups based on data in your account and your contacts. An audience member tried it and reports that it worked well enough to impress him. It accurately represented about 80% of his connections. It does not yet recommend people whom you should follow or to whom you should connect.
I don’t remember Karrie talking about this model and its application to Facebook and Twitter during the MIT talk. It seemed that most of that talk was about the presentation of people’s conversations and interactions during meetings—what she covered during the first few minutes of this talk. I’m glad I decided to attend this Berkman event so I could learn more about her research. This talk was a bit more relavant to my interests than the MIT talk. The visualizations and how they work are interesting in their own right, but I’m more interested in the social software aspects and how to improve existing software.
Ethan Zuckerman made some comments about how his wife often carves up what she wants to blog among various audience members/friends. Karrie said there’s been a lot of research into information flow and some research into relationships, but not much in the space of what information people send to which ties.
Karrie asks how people use Twitter: Tweetdeck, reading off twitter.com, going to the top of their Twitter list … ? Someone asked whether folks read as much as they tweet. Karrie says she reads more than she tweets. Ethan describes it as his “ten minutes of serendipity,” so he uses a completely new browser window and closes it when he’s done. People generally use hashtags and retweet interesting posts.
SwiftRiver sorts through tweets in order to try to find the most accurate tweets. There’s a problem with trying to figure out exactly what’s happening with all the social networking posts after a disaster. Some research indicates the number of real people actually tweeting or communicating from disaster zones is far less than what they seem to be because a number of people retweet or repost the information, even folks who are not located near the disaster.
Karrie mentions how doing research on Facebook is difficult because their terms of service indicate people who keep Facebook data on their personal computers for longer than a few hours violate their terms of service. That makes research very tricky. Is that part of the terms of service just a bit of an oversight not necessarily meant to trap researchers or is Facebook deliberately trying to discourage researchers from using the tool?
A lady asks about trust in these kinds of environments. She heard about an experiment where people went into isolation (a French farmhouse) with nothing except Twitter and Facebook and maybe other social software. After a month, folks will see how accurate what they know about the world is.
Weak trust ties are like what we have with police officers, firefighters, and other public authority figures in service roles. What about someone at your religious institution? Trust doesn’t map to strong ties the way people think it does. Karrie believes there is a bigger variety of strong and weak ties than what we discuss in relation to Granovetter.
Karrie’s ultimate goal is in information flow. If you want your information to get dispersed, make sure a bunch of garrulous, gullible people are in your network.
Wendly Seltzer makes the case for analyzing how people shift members of their communities around once they begin using the model to manage their friends. Karrie doesn’t want to destroy relationships by publicizing too much about them. Someone else makes the case for relationships being fluid and being able to easily change and represent that online. What if someone moves away? What happens after a fight?
Context is key. In some spaces, if you want someone to respond, being rude to them is the way to force that to happen. In other places, being polite is what garners good responses (and respect).
I wonder how many other people were blogging, tweeting, or using Facebook during the talk.