Using machine learning to curate content is not going to solve some of Twitter's larger problems.
When a technology company whose stock has been underwater buys a company called Magic Pony—as Twitter just did—the headlines pretty much write themselves. The acquisition, which has a reported price tag of $150 million, is supposed to boost Twitter’s abilities in the machine learning department so that it can get smarter about the photos and videos that it displays to users. But even having a magic pony may not be enough to help Twitter win this particular race.
As Fortune‘s Barb Darrow explained in a post about the deal, Magic Pony is a British startup with expertise in the visual end of machine learning, a branch of artificial intelligence focusing on algorithms that can adapt their behavior based on new data. The idea is that these kinds of algorithms will help Twitter come up with the kind of content users want. Twitter CEO Jack Dorsey explained in a post:
The Twitter post talks about adding Magic Pony to a group at the company building that it calls “Twitter Cortex,” a team of engineers, data scientists, and artificial intelligence experts who are building a product where “people can easily find new experiences to share and participate in.” In other words, a product that can boost the kind of user engagement levels that Twitter desperately needs in order to convince Wall Street that it is worth its $10-billion market cap.
Since he returned to Twitter TWTR as CEO a year ago, Dorsey has talked repeatedly about how he wants Twitter to become the first place people go for real-time content of all kinds. “Twitter is live,” he said during the company’s earnings call earlier this year. “Live commentary, live conversations, and live connections.” A big part of that has been a push towards video, with the company’s much-touted deal to stream NFL games being just one example of that strategy.
One problem for Twitter in betting on video is that Facebook has also set its sights on being the place where people go for live video and real-time content, and it has made significant gains even in just the past year, accumulating more than nine billion video views a day. Dorsey can complain that Twitter has been all about real-time for almost a decade now, but that doesn’t matter to users who want to see the latest Chewbacca Mom video clip or exploding watermelon.
As a number of outlets have reported, Facebook is also paying news organizations and publishers—including the New York Times—to produce live video for its platform, and they are doing so in great numbers. Is Twitter going to start paying producers as well?
Figuring out how to show users more relevant content, meanwhile, is something that Twitter has been trying to do for some time now, with features such as Moments and a timeline that is now ranked at least partly by algorithms (as Facebook’s is). It’s possible that Magic Pony’s technology will help with that, but showing users good content is only one of the problems facing the company.
An even larger problem is what appears to be the weakening of Twitter’s advertising business. According to a recent analyst’s report, advertisers—particularly the large ones that the company counts on for significant amounts of revenue—are losing interest in the platform. In part that’s because Twitter doesn’t have the kind of reach that Facebook and Snapchat have (the latter recently passed Twitter in terms of daily active users), but also because they aren’t seeing the return on investment they need.
Is Magic Pony’s technology going to somehow help Twitter repair the fundamental underpinnings of its ad business? Maybe if it starts surfacing videos that more users want to engage with, it could theoretically help convince brands that Twitter as a platform is still worth investing in. But the company appears to be rapidly running out of time to make that case.