Twitter launched a new initiative Thursday to find out exactly what it means to be a healthy social network in 2018. The company, which has been plagued by a number of election-meddling, harassment, bot, and scam-related scandals since the 2016 presidential election, announced that it was looking to partner with outside experts to help “identify how we measure the health of Twitter.” The company said it was looking to find new ways to fight abuse and spam, and to encourage “healthy” debates and conversations.
In a series of tweets, Twitter CEO Jack Dorsey acknowledged that his company didn’t “fully predict or understand the real-world negative consequences” of how the platform was designed, like harassment, trolls, bots, and other forms of abuse. “We aren’t proud of how people have taken advantage of our service, or our inability to address it fast enough,” he wrote.
Twitter is now inviting experts to help define “what health means for Twitter” by submitting proposals for studies. The company will provide grant winners, who need to apply by April 13, both “meaningful funding” of an unspecified amount and access to its vast trove of user data. It’s uncommon for a social network to ask outside experts to help it define a new metric, but the resulting “Twitter health” studies certainly won’t represent the first time academics have tried to quantify the quality of interactions online.
‘What will determine the power of a move like this is how it will factor into Twitter’s business objectives.’
Katherine Lo, University of California Irvine
“Often when I see graduate students present on this topic they act as though they’ve just discovered something that was a topic of great discussion for 20 years,” says Susan Catherine Herring, a professor of Information Science at Indiana University who has studied online behavior since the 1990s. Herring teaches Computer-Mediated Network Analysis, or the study of how to analyze conversations in online communities. Which is to say, Twitter has literally decades of existing research it could potentially lean on to help inform its new study of health.
Twitter’s size and specific problems, though, may benefit from a fresh look. This may also be one of the first times that a social network has pointedly compared itself to a living, breathing body. And yet both Dorsey’s tweets and the official Twitter blog about the new project disclose that the company doesn’t know what “health” exactly means in this context. What would a healthy Twitter look like? What even is health for a non-sentient website?
It might mean curbing abuse entirely, or eradicating bots. Or health might look more like symptom management, curtailing things like harassment and propaganda without completely curing them. No one really knows. But Twitter does focus on four key metrics in its call for proposals: shared attention, shared reality, variety of opinion, and receptivity. Which boils down to: Are people talking about the same things using the same facts? Are they crafting a variety of arguments, and are people open to listening to new ones?
The social network borrowed the four “indicators” from Cortico, a non-profit research organization affiliated with the MIT Media Lab. Dorsey said on Twitter that these indicators weren’t necessarily the ones Twitter would end up using to measure the platform’s health, but they demonstrate one possibly approach.
Herring says that most of Cortico’s indicators are reasonable, and that it would be possible to craft an empirical study in which they could be measured. For example, one group of researchers could craft a simple study to test whether people are generally talking about the same things by analyzing the key words they use. Researching other questions, like whether users are open to hearing new ideas, might be more difficult to quantify. Herring acknowledged that the social network would need to eventually be more specific in scope, but that it was fine to start with fairly broad questions. She did say, though, that one indicator might present problems: “shared reality.”
“Are we using the same facts? There’s already a bias in that perspective because we assume we know what the facts are,” says Herring. “What other groups consider facts might not be the same. You’re approaching that from a biased perspective.”
Another thing to consider is exactly whose health Twitter will measure. “When they refer to it in terms of a bodily metaphor, the health of what body? The health of Twitter’s body?” asks Whitney Phillips, a professor at Mercer College who focuses on online culture. “I don’t think as a corporation Twitter necessarily has my health in mind.”
‘We are in deep shit on the internet in so many ways.’
Whitney Phillips, Mercer College
Twitter also hasn’t yet mentioned how the project will factor into its broader business goals. Traditionally, calling a company “healthy” means that it’s profitable—but that’s not necessarily the case when it comes to Twitter’s new initiative, and it’s not clear that steps to advance one definition would necessarily benefit the other. Other social media platforms, like Facebook, have acknowledged that making decisions based on users well-being may impact metrics that help lure advertisers, like time spent. What happens if Twitter finds a way to make itself less toxic, but shareholders reject it?
“Developing forms of measurement that deviate from engagement and growth is important for the industry, but what will determine the power of a move like this is how it will factor into Twitter’s business objectives,” says Katherine Lo, an online community researcher at University of California Irvine. “Many anti-harassment initiatives by social media platforms have become effectively toothless because factors like growth metrics, or commitment to advertisers, ultimately trump safety and health in day-to-day product decisions.”
Twitter’s effort to evaluate its health seems like an earnest effort to combat that trend. But identifying the right metrics is only half the battle. The hard part will come when Twitter tries to translate the results of an empirical study into a meaningful change for users. “I wish them luck, because they may discover a lot of information, but to try and implement change and create a healthy environment, that’s much more challenging,” says Herring.
For now, no one really knows what, exactly, a healthy Twitter would look like. There’s at least a consensus, though, that the social platform is certainly not there yet. Twitter asking outsiders for help could improve its platform for millions of people. Alternatively, it might be a sign of just how bad things have gotten. “We are in deep shit on the internet in so many ways,” says Mercer College’s Phillips. “This is an example of the depth of that shit pile.”