12.20.10 12:04 AM ET
Cracking the New York Times Popularity Code
The most-emailed articles list on The New York Times website is one of the Internet's key barometers of news and trends, an essential way for the world to stumble onto stories and ideas that might otherwise get lost in the ether of the perpetual news cycle. Thus, careful watchers might have been puzzled by a seemingly out-of-place story last week. Among the latest news, feature and opinion pieces was a three-week-old science section story about a soon-to-close exhibition on cuneiform clay tablets. What could have propelled a stale, bone-dry story to the top of the Internet's importance arbiter?
I can tell you: It was me.
More precisely, it was a group of people under my direction who all, at my request, emailed that particular story within a relatively short timeframe to learn exactly what it takes to make the most-emailed list.
How we did it—and how many people it took—reinforces a lesson of our viral media age: Even at the biggest newspaper website in the world, the content that is spotlighted as most engaging reflects the judgment of a group far smaller than the overall audience, and can even be gamed by those motivated enough to do so.
Popularity, of course, is considered a key metric of the information age. From entertainers courting the spotlight to entrepreneurs who want publicity for their ventures—and don't forget journalists slaving away at their keyboards—"going viral" is a cherished goal. Exposure begets more exposure. With the electronic fire hose pouring out torrents of content each day, many turn to what's already been deemed interesting by others—whether it's their Facebook friends or the collective readership of a major publication—to help them filter.
Which is what makes the most-popular lists at the Times so intriguing. The Times attracts some 50 million unique visitors each month online, far outdistancing rivals like The Washington Post or The Wall Street Journal, according to comScore. In fact, the Times publishes several different types of most-popular lists, including rankings of which articles have been viewed the most and which have drawn the most interest from bloggers.
But it is the most-emailed list that enjoys primacy of place on the site's home page and many other key pages. Thus, the most-emailed list implicitly represents which stories are getting the most buzz. Readers can even subscribe to an RSS feed of the most-emailed list.
So how many emails does it take to make the list? We started several weeks ago by recruiting a cadre of friends and colleagues who agreed to stand by and, when instructed, visit the Times website, view the designated article and use the site's email function to have the article sent to a friend. Next we selected an article that was unlikely to have many, if anyone, emailing it on their own.
By 7 a.m. the next morning, about 300 emails had been sent, and our story was climbing.
To start, we went with pieces from the Science page, which offered us an achievable gateway—the chance to crack an individual section's ranking, rather than the overall list, which is dominated by news, service and opinion. (Not that the Science list is a pushover. As the Times itself reported earlier this year, one analysis by University of Pennsylvania researchers showed that science stories do particularly well in email-ibility.)
Round one of testing got under way on November 23 around 5 p.m. Our volunteers were asked to email a story on research funding, " Rare Hits and Heaps of Misses to Pay For," which had been published online more than two weeks prior, to one or more friends—and to do so within a few hours of the call to action. As our volunteers reported back, the tally showed that over several hours, 48 people emailed it to a total of 135 friends.
By 7:35 p.m., the story popped up on the Science page most-emailed list, starting at No. 10 and rising to No. 6 by 9:30 a.m. the following morning. (The default most-emailed list reflects activity over the previous 24 hours, according to language on the Times' expanded most-popular page.)
But it left us with a question: did the results reflect the 48 people sending the article—or the 135 recipients? So on December 9, we launched a second test, asking the volunteers to email an old, rote story, "Justices to Rule on States' Emission Case," to one person only. Ultimately, 35 senders shot the story out within a few hours. And by 11 p.m., it too emerged on the Science most-emailed list—eventually peaking at No. 5, overnight, slightly better than our previous test.
That suggested that the number of senders, not recipients, was the key factor in scoring a slot on the most emailed list, understanding, of course, that traffic naturally fluctuates on the Times' site, so the lists could be more difficult to climb on busy news days. (We launched all our tests on weekday evenings.) Once our test story surfaced on the list, of course, it raised the likelihood that other visitors to the Times site would notice the piece, decide to read and then email it. So it's possible that our test story peaked higher on the list than could be accounted for purely from our test emails.
So if a few dozen people can make a Science article "popular" on that channel, what does it take to get onto the big board, the one noticed by millions of readers, and even ardently followed, I'm told, by Times editors and writers themselves?
Clearly needing more than a few dozen friends, I enlisted a small army using the greatest motivator I had: cash. I found people, mostly abroad, via an online labor marketplace called the Mechanical Turk, who were willing for a small fee to register at the Times, or even complete multiple registrations, and send a story out.
On December 14, my troops were given a new story: the aforementioned "An Exhibition That Gets to the (Square) Root of Sumerian Math," now three weeks past its publication date.
Based on our previous tests, we knew we wouldn't see movement in the most-popular list immediately, since the Times appears to use a 24-hour moving average. We started to see some impact at 11:43 p.m., 70-odd emails in, when the story popped up on the Science most-emailed list at No. 8. By 7 a.m. the next morning, about 300 emails had been sent, and our story was climbing—now up to No. 4 on the Science-page list, our new high-water-mark.
But as we checked the overall list and found no sign of our test article, we figured we needed to turn up the heat—saturating the site with more emails over a 24-hour period. Things started to pick up. Shortly before 1 p.m., the Sumerian story jumped up to No. 2 on the Science page—and, more importantly, made its debut on the overall most-emailed list. Though the Times only displays the top 10 most-emailed stories in its home-page box, it also offers a link to an expanded list showing the top 25. After 400 emails, our story edged onto the overall list at No. 22.
So we sent word to our global staff: more emails! As evening came, the story inched up further, hitting No. 1 on the Science list and ticking up on the overall list: 19, 18, 17. Around 12:35 a.m. on December 16, 36 hours into the blitz, it happened. Our story made it onto the Times' home page, scoring the No. 10 slot on the most-emailed list. And it wasn't slowing down. As the hours passed, it rose higher and higher. At 12:18 p.m., it peaked at No. 3, outdoing a Gail Collins column, a story about the Obama administration's plans for handling a nuclear attack and a handful of perennially popular writers. (Take that, Tom Friedman!)
So how many emails did we need to pull it off? The answer: 1,270.
Clearly, roughly 1,300 email senders is more than a handful. On the other hand, 1,300 people wouldn't be enough to sell out one performance of Wicked on Broadway. More to the point, it represents a tiny fraction of the Times' overall readership. If our results are accurate—and a Times spokeswoman confirmed that the list is based on individual senders and did not have any disagreement with this story's methodology—out of the 30-plus million Times website visitors each month, it takes only one out of every 25,000 emailing a particular story to secure it a spot, at least for a day, in the hallowed most-emailed list. (The Times spokeswoman did indicate that hitting the No. 1 spot would have required significantly more email senders.)
The fact that the No. 3 slot was attained with fewer than 1,300 emails also raises the question of whether others might seek to manipulate the list for their own gain, or may have already done so. Notably, the most-emailed list of virtually every other news site in the world surely requires a small fraction of that number, and thus would be far easier to game. That certainly includes The Daily Beast, which is one reason this site's homepage features the most-viewed stories more prominently than the most-emailed.
The Times told us that the paper considers it important to maintain the authenticity of the most-emailed list and that was a consideration in basing the list on the number of senders rather than recipients. "This minimizes the ability for one person to game the system, though we know it's not a perfect solution," the spokeswoman wrote in an email.
As for why the Times features most-emailed over most-viewed, the spokeswoman responded: "We think of emailing as a more engaged action than viewing or searching, since the user is taking a personal action by proactively sending the article to a specific person or group of people."
This wisdom of crowds so popular on the Internet was born partly out of necessity. A better, impossible-to-game solution may be emerging from social networks, which let individuals know which content is popular within their own network. From The Daily Beast to the Huffington Post, many news sites now offer readers a look at what their friends are liking. At least one major newspaper site does too: The New York Times.
Thomas E. Weber covers technology for The Daily Beast. He is a former bureau chief and columnist at The Wall Street Journal and was editor of the award-winning SmartMoney.com. Follow him on Twitter.