Peter Coy
The New York Times

For Elon Musk, Lessons From Friendster

For Elon Musk, the new owner of Twitter, the demise of Friendster is a cautionary tale. Friendster, a social network founded in 2002, gained millions of users soon after it started but lost them almost as quickly, later metamorphosing into an online gaming site based in Malaysia before closing down entirely in 2018. Already by 2009, The Onion satirical news site ran a story headlined “Internet Archaeologists Find Ruins of ‘Friendster’ Civilization.”

How did Friendster collapse so quickly and how can Musk, the world’s richest man, avoid making the same mistakes at Twitter, which he took over on Thursday?

An academic discipline called network science can shed light on the question, albeit without giving any firm answers. This week I interviewed a scientist in the field, David Garcia, a Spaniard who is an expert in computational social science and teaches in the department of politics and public administration at the University of Konstanz in Germany.

The risk for Twitter, as for any social network, is “unraveling,” Garcia said. Something happens to raise the costs or reduce the benefits of being in the network. It could be a failed redesign of the interface or some kind of flame war (or a takeover by a divisive gazillionaire). So a few people who were only marginally attached drop out. Some of their friends now have less reason to participate, so they drop out, too, and so on.

This snowball doesn’t roll all the way down to zero users. There are tight-knit groups of people who stay in the network because their friends are still in it. But there aren’t enough of these clusters to entice advertisers, so the network hemorrhages money and eventually shuts down, Garcia said. He described the process in a paper, “Social Resilience in Online Communities: The Autopsy of Friendster,” that compared Friendster to Livejournal, Facebook, Orkut and Myspace. He wrote the paper in 2013 with Pavlin Mavrodiev and Frank Schweitzer, two colleagues from the Swiss Federal Institute of Technology in Zurich, where he worked at the time.

Mathematically, a network will survive if enough big clusters remain to keep the network commercially viable after the marginal users get swept away. The key metric is what’s called the k-core. The “k” is the number of people a user is connected to; a “core” is a group of users. By definition the zero-core includes every user, even the people who aren’t connected to anyone else. The 1-core includes only people who have at least one connection.

What matters for a network is not just how many friends someone has, but how many friends the friends have. The 2-core includes only people who have at least two connections, and where each of those people also has at least two connections in the 2-core. The 3-core includes only people who have at least three connections, and where each of those people also has at least three connections in the 3-core. And so on. Picture a k-core as a bunch of people in a circle who are connected to people opposite them as well as the ones to their sides.

The big idea of k-core analysis is that a network with lots of cross-connections (a “mesh” network) is more resilient than a network with one influencer at the center and lots of people on the periphery (a “hub and spoke” network). Part of Friendster’s problem was a lack of cross-connections. (This was the case in the United States, anyway; it did considerably better in the Philippines for some reason.) “Rather than using Friendster to make dates, most of its users were simply cruising around and looking at the weird interests, pictures and blog-droppings of strangers (including so-called fakester profiles of Jesus and Burt Reynolds),” a retrospective in Inc. magazine said.

Friendster suffered from the rise of Facebook, which was opened to the general public in 2006, and from some interruptions in availability. Garcia speculated that an unsuccessful interface redesign in 2009 was the final blow. “One can never be sure with such observational and historical analyses, so I can imagine that it started for a mixture of reasons anyway,” he wrote in an email.

Twitter has defied repeated forecasts of its demise in part because many of its users have formed large and tight clusters organized around interests, whether economics, finance, football or ethnicity. Those webs of affiliation are hard to replace. While some people have proclaimed they’re leaving Twitter because Musk took over, others vow to stay. “Abandoning such places out of some misplaced sense of moral rectitude simply clears the field for more lies and mischief,” Tom Nichols, a staff writer for The Atlantic, wrote on Monday.

Musk himself in a tweet in April asked, “Is Twitter dying?” He wrote that many of the most-followed people on Twitter post rarely. That’s true. But Twitter doesn’t survive mainly as a broadcast medium. It succeeds based on interaction. You’re probably not going to get retweeted by Barack Obama or Taylor Swift, but you might get retweeted by your cousin, and that could be pretty satisfying.

Seen that way, Twitter is fairly resilient. On the other hand, it hasn’t turned a profit in eight of the past 10 years, and now the Musk buyout has saddled it with about $1 billion a year in interest payments. Its biggest threat is probably the new boss himself, who could drive away users and advertisers by allowing more offensive content on the site in the name of free speech.

What’s the chance of that? “I don’t like making predictions,” Garcia told me.

Since the spring of 2021, Britain has had three pro-Brexit prime ministers: Boris Johnson, Liz Truss and Rishi Sunak. But over that time the public’s misgivings about Brexit, the 2016 referendum to leave the European Union, have only grown. Why the second thoughts? An August poll by UK in a Changing Europe found that a plurality of Britons (46 percent) thought being outside the European Union had good consequences for Britain’s Covid vaccination program and 35 percent thought it was good for Britain’s ability to control its own affairs, but only 11 percent thought it was good for the cost of living. Consumer prices including housing rose 8.8 percent in the 12 months through September.

The New York Times