The Internet likes to tease

The art of a good headline has been undermined by our reptile brains, and in the process, we're blinding our search engines by eliminating useful metadata.

Once, a headline conveyed a quick summary of the story: Fifteen dead in high-speed bagel tragedy.

Today, however, the structure of news has separated the tease from the story. Headlines dangle like lures, trying to trick our brains into clicking them: Why eating bagels on America's highways could kill you. Or maybe Ten things bagels do at high velocity.

This, friends, is the death of the headline. When I wrote for GigaOm a few years ago, editor Carolyn Pritchard admonished me to tell the story in the first paragraph. "The Butler did it; here's why." Don't bury the lede, goes the old journalist adage.

But we aren't writing stories to inform any more. We're writing them for likes, or upvotes, or retweets. We're building one-line Skinner Boxes that tickle our amygdala, trying to get readers to click through. Modern headlines are cryptic, staged. We bury the lede, because we make money when readers have to dig. And we encourage them to dig with the promise of life-changing advice, hidden on the other side of a mere click. "This gem will fix all your problems."

Teasers are a consequence of how the Web works

For starters, everything is analyzed, using analytical tools that track every click, mouseover, scroll, and back-button. Every action—or inaction—is grist for the analytical mill. A good headline no longer informs, it drives clicks. And a good story doesn't summarize, it keeps you on the edge, scanning for the answer, upping the time-on-page, ad impressions served, and other metrics that replace utility in today's free-as-in-beer press.

Years ago, I asked Gail Ennis, CMO of Omniture, about the automatic optimization algorithms some sites use to maximize traffic. "You know what drives the most views?" she asked me. "Big-breasted women." Turns out that cleavage gets clicks—from men and women. "The problem, of course, is that short-term gains lead to long-term brand erosion."

Headline optimization is hard science. Huffpo, for example, creates multiple headlines, tests them as soon as a piece is posted, then sticks with the one that works best. We use similar techniques—tuned for audiences—on every social platform: reddit gets one line; Facebook shows a paragraph.

There are even tools to help write headlines that get clicks. Sigh.

It's not just a matter of throwing multiple headlines at a wall to see what sticks. There is a lot of psychology at work, too. We live in an information-saturated world, so anything that sounds like it's been cleaned up, curated, and packaged into a quick fix is alluring. Just look at Buzzfeed and Cracked for examples. Today's fastest-growing—and most irritating—content networks rely on this. Unworthy Upworthy frequently says things like, "you won't believe what he does at 3:17," teasing us with the promise of content, but also requiring us to interact with that content to get the squirt of oxytocin we've been promised.

It's Chicken Soup for the Internet's lack of a soul.

Two problems with teaser headlines

Besides being annoying, this presents a couple of problems. First, legitimate news organizations have to stoop to the depths of their sullied, grey-market brethren to make a dime. It puts them squarely on the horns of a paywall dilemma: write shitty headlines and get clicks, at the erosion of utility to readers; or put up a paywall, and lose the visibility that comes from search engines and content sharing.

A lowered bar drives everyone towards mediocrity.

The second problem is deeper, and more fundamental. Search engines work by looking at the contents of a link ("This is a page about peanuts") and inferring that the link is relevant (the page is, in fact, about peanuts.) In other words, old-school search algorithms trusted that the headline (the text of a link) contained useful metadata about the content (where the link went.)

By writing teasers instead of headlines, we've blinded these algorithms. There's very little metadata in a teaser headline, so the search engine has to look at the destination and see what it's worth. Algorithms like Google's Pagerank can't rely on inbound links and headlines to understand content; instead, they have to score the page itself.

This undermines some of the fundamental democracy that existed at the dawn of the Web. Once, if enough people said a page was about something ("worst president ever",) then the page they all pointed to would become associated with that term (something called a Googlebomb.) Sure, sometimes this got Gerrymandered to great effect, but for the most part, it meant the Web "learned" about what was relevant organically.

Today, these massively optimized teaser headlines mean that the Web is increasingly curated by those engines. Known sources of data—Urbandictionary, Wikipedia, YouTube, Snopes—are safe choices, because the engines know they're valid. But this also leads to the centralization of the web, and the "hollowing out" of the Web's middle-class. Instead we have a few known sites, and billions of Tumblr, Pinterest, and Facebook pages, and nothing but platitudes in between.

Whither the Middle Class Web?

As a side note: I write about a lot of different things on this blog. It's a personal blog for a reason. The subscriber list has grown steadily but gradually. But the variety has an interesting side-effect. A few weeks ago, I wrote about teaching science to preschoolers. Then I wrote about how to run a debate. Then the ethics of torture using Virtual Reality. Each time, I gain ten followers, and lose three.

I find this fascinating. I could probably focus on one topic (as Ben and I do at our Lean Analytics website) and grow subscribers (we have 6,000 now!) But I'm not really trying to build a following. I suspect that the subscriber list for this blog is "people I would quite like to go drinking with." I'm triangulating an audience (or rather, the audience is triangulating me) because of the variety. That also makes this blog hard to characterize—the search engines can't put it in a specific box.

So if I had to guess, I'd say the future of this "middle class of the Web" is tribes, intersecting Venn diagrams of interest and topic, that are harder to define, more fungible, and more serendipitous.