Posted by
Microsoft Research Blog
Microsoft Research Forum Episode 3: Globally inclusive and equitable AI, new use cases for AI, and more
In the latest episode of Microsoft Research Forum, researchers explored the importance of globally inclusive and equitable AI, shared updates on AutoGen and MatterGen, presented novel use cases for AI, including industrial applications and the potential of multimodal models to improve assistive technologies.
How many followers do you have on Twitter? You probably have a pretty good grasp of that number, too, don’t you?
That’s understandable. Social media lends itself freely and easily to counting things, and if we start with no Likes and no followers, then each time our numbers lurch upward, we get a corresponding feeling of social success. Our popularity—or lack thereof—now comes with metrics, and they’re irresistible.
But what, really, do they signify? That’s the discussion Nancy Baym of Microsoft Research wants to engender with her paper Data not seen: The uses and shortcomings of social media metrics, published Oct. 7 on First Monday, an openly accessible, peer-reviewed journal published on the Internet about the Internet.
The paper examines how different people define audiences in different ways and puts the use of social-media metrics to understand audiences into the historical context of audience measurement in the mass media. The paper then goes on to state the appeal and the drawbacks of such metrics—and that data-analysis systems are shaped by the values of their creators.
“Metric and big-data analysis,” Baym writes, “generally serve economic values, while other approaches to data may be more appropriate for assessing social and personal values.”
An original perspective, to be sure, but even more interesting is that, in examining such issues, she places a particular focus on musicians.
“I chose musicians because they are at the cutting edge of changes that affect a huge variety of professions and individuals,” she says. “Having and building a social-media presence is taken to be important to success, and yet no one really understands what that means. I was going to music-industry events and hearing musicians express a lot of anxieties about social media, and I realized that most of the people on stage and in the media, talking about what musicians should do, were not musicians and did not fully understand their perspectives.
“My goal in this work has been to understand these changes from musicians’ perspectives.”
The paper, Baym says, was inspired by a comment from American singer-songwriter Erin McKeown about how the need to maintain a social-media presence—to get the self-validation that artists crave—forces her to think of herself as having two careers, one as a musician and another as a builder of an online presence.
“I saw a related disconnect throughout my interviews,” Baym explains, “between what the musicians experienced as really meaningful in their interactions with audience members, which were usually these single moments that were profoundly affirming, and the contemporary discourse about big data that suggests it is data quantity that leads to insight.
“It’s about what we miss out on when we focus on the macro and forget to look at the small moments that give life its meaning.”
In that way, the paper poses something like a left-brain/right-brain puzzler. Musicians, by nature, would tend to be more right-brained, intuitive and thoughtful, but managing social-media metrics requires logic and analysis, classic left-brain traits. Baym’s use of musicians to shine a light on issues with social-media metrics thus becomes a matter of true insight.
“The core finding,” she says, “is that there is a big difference between massive amounts of data being available, thanks to new media, and being well-equipped to manage and interpret data. It’s important because, these days, people are putting a lot of faith in big data, often harvested from social media, as a purveyor of truth.
“With earlier data, people had shared understandings about how to interpret data—not perfect, but adequate. With these new kinds of data, we really don’t fully understand the strengths and limits of data, so they are often taken to mean things they don’t. For instance, someone with a lot of followers may be understood to be popular, when they really bought those numbers. More mildly, someone may be taken as having a lot of commercial potential because they attract a lot of people online, yet that may not translate into people buying their product.”
The paper, based on interviews with 37 musicians and three managers over the past four years, certainly is a fascinating read. But what, pray tell, do musicians ask Baym about how to interpret this new “data thing”?
“They tend not to ask me about making sense of data,” Baym says. “They are more concerned with things like how to deal with haters and which media they ‘need’ to use. The message I try to convey is that, no matter what they are told by experts, there is not a single ‘right’ way to use social media. It’s a question of understanding what kind of a person you are, what you are comfortable with, and what your boundaries are and what kinds of media use can fit into those parameters.
“For some people, social media—including analyzing the data—comes very naturally, but for others, it’s a burden.”