The horrible history of the headline, from print to social media, with a shout-out to Preet Bharara

There's a question floating around on Twitter: "What‘s something that seems obvious within your profession, but the general public seems to misunderstand?" One answer that I've seen several times is that "the writer of the story doesn't write the headline."

And then there's a tweet from @preetbharara, who declared "The worst members of the press are the people who write the grabby headlines for stories they don’t seem to have read."

I want to address both of these issues in some detail, discuss why I think writers should write their own headlines, why they often don't (some, in fact, do), and mostly why headlines aren't as simple as they first might seem.

First let's take a look at the eternal question: How did I get here? If we turn the clock back to the era of print, when everything was simple and pure except that it wasn't, we find that the headline never was very simple.

The earliest newspapers didn't even have headlines. Ben Franklin's Pennsylvania Gazette simply began each story with a dateline: "Vienna, Feb. 21." But headlines proved to be useful, and as printers' collections of metal type expanded to provide the necessary larger and bolder fonts, they came into widespread use, and not just as a single line atop a story.
This is the New York Times, reporting the assassination of President Lincoln:

President Lincoln Shot by an Assassin.
The Deed Done at Ford's Theatre Last Night.
The President Still Alive at Last Accounts.
No Hopes Entertained of His Recovery.
Attempted Assassination of Secretary Seward.

In the original print version, each of these lines was in a different style, each larger than the body of the story, and each was (this is important) constrained to fit in a physical space.

Early newspapers generally set type in multiple, narrow columns. A long story might wrap across multiple columns, but headlines were one column wide. Toward the end of the 19th century, multiple-column headlines became more common. In the 20th century, the "functional design" movement led to simpler presentations, but there were still hammer heds, deks, drop heds, pull quotes and other typographical components in common use.

These components eventually were formally recognized as serving a "grazing" and "skim and dip" behavior when scientists attached eye-tracking equipment to readers in laboratory settings, but long before that, editors and print designers knew that "display type" was important. The names of these components often differed from one shop to another, but they all shared one overriding characteristic: the tyranny of physical space. In printing with metal type, words simply had to fit spaces.

Newspapers evolved a factory-floor solution, a Henry Ford production line of sorts, in which reporters would dig up information, writers (usually, but not always the same as the reporter) would craft a story, and a sequence of multiple editors would edit the story for style, tone, content and accuracy. Late in the workflow, an editor would assign a story to a page, specifying the headline (and other components) and usually a different editor would write those components to fit the specifications. In the rush to meet a print deadline, the editor writing a headline might not be the same editor who edited the story. (The potential for introducing error should be obvious.)

Editors learned to count variable-width characters to estimate whether a headline would fit, avoid awkward line breaks and unbalanced white space. A supervisory editor (called "slot" in the United States) might reject as many as six or eight headlines, tossing them back to a subordinate (on the "rim"), in an attempt to get just the right combination of information, language, and fit.

This was a teaching and learning process; even a dense rim rat like me eventually learned to write better headlines under the brutal guidance of a good, merciless slot. This process generally worked to filter out error, but it was never perfect, and while a copy desk might intercept hundreds of errors a day, it also could miss many and create a few of its own, both in body type and headlines.

In the era of metal type, nothing pleased a blue-collar printer more than catching an error missed by the college boys in the newsroom, and many an embarrassment was avoided when a printer brought a sizzling hot stick of type, wrapped in a steaming wet cloth, to the copy desk with a question and a grin.

But isn't all of that in the past? No, it's not. Not entirely, but there has been great change.

Print isn't gone and isn't going away any time soon. Technology has done away with metal type and blue-collar workers in the composing room, and now economic forces have led to centralization of both print and digital processes. These changes have stripped away layers of error-catching and introduced unfortunate disconnects between the author of a text and its many headlines. And lastly, the Internet has created even more variations and conflicting requirements for a headline, complicating the production processes and creating more opportunities for mistakes to creep in.

Centralization enables beleaguered newspapers to cut costs, so today a local newspaper might be designed electronically in a regional center hundreds of miles away. Local editors might offer suggested headlines, but even with the type-squeezing magic of Adobe InDesign, there's still that physical space requirement that often results in a headline written by an editor who barely read the story, was extremely pressed for time, and has never set foot in the community. Similar centralization on the digital side runs the same risks of inaccuracies, exacerbated by the stripping of layers of editing to save money and the rush to be the first to get an item online.

Then there's the growth of conflicting requirements.

There was always conflict between the journalistic missions and the commercial missions of a headline. I briefly worked for an unscrupulous publisher, a bit of a sociopath, who instructed me to punch up the presentation and language of a street-sales newspaper. His exact words: "Go for the crotch."

The journalistic mission is to inform, so a headline begins as a tool to do just that. The late, great Kansas journalism professor John Bremner described the lede (first paragraph) of a story as a precis, a distillation of the essence of a story, and a news headline as a precis of a precis.

But even without the injection of commercial considerations, there's mission conflict. There's a journalistic goal to get people to read beyond the headline, to discover more information than can be presented at the top layer. In the 1865 New York Times example, notice that the last line is "DETAILS OF THE DREADFUL TRAGEDY." It's not informative. It's a tease. It's an invitation. It's a line designed to create a need that can only be met by picking up the newspaper and reading beyond the stack of heds and deks.

This need to get people to read may be both journalistic and commercial, as best seen in the New York tabloids: "HEADLESS BODY IN TOPLESS BAR" (New York Post, 1983). No one is informed by a newspaper not purchased or a story not read, and the tabloids rely heavily on street sales, essentially impulse purchases.

So it was already complicated. Then the Internet made everything even more complicated.

Websites aren't newspapers, where many stories appear on one page, with layers of display type beckoning the reader with hints and details. Stories on the web are fundamentally disconnected from one another, displayed on separate pages. If you reach a page, it's generally on purpose, and that changes everything about the missions of a headline. There usually isn't even an effort to address the scanning behavior of casual readers, although there are examples, such as CNN's "story highlights," of details being pulled out and featured in the manner of a 19th-century newspaper.

The h1 component, the primary title of a story, is as much focused on informing a machine as informing a human. Search algorithms and Google News algorithms and human search-engine usage require that key words be identified and carefully introduced into the title. (This is a point of failure for local websites, which usually assume location and don't overtly mention a town name in a headline.) A CNN story page is headlined: "Note to President Trump: The stock market has hit an all-time high in 30 of the last 54 months." Is it really a note to President Trump? Of course not, but working his name into the (fairly long) headline helps the story be discovered in a web search for "Trump stock market" and aggregation with other political headlines on Google News. It's not just a business investment performance story.

But websites also have homepages and section fronts, and feeds, and other aggregations where headlines may be presented to readers of varying focus and purpose. The death of the homepage has been greatly exaggerated. Again we encounter the conflicting missions of informing the reader and incenting an action: clicking and continuing to the story page. The h1 that worked so well for search engine optimization might not be a good candidate for either of these missions, and it might not -- dare we say it -- physically fit. For these reasons, the web publishing system that we built for the Morris Publishing Group supported an optional "front hed" that might differ from the story title.

As the most measurable medium ever created, the Web has introduced performance metrics to the act of editing and writing headlines. There are systems for testing the clickthrough performance of alternate headlines, helping digital producers dynamically tune presentations to get results for which they may be incented, perhaps even in their paychecks.

Then there's social media distribution to consider.

When you share a story on Twitter or Facebook, they reach out, grab the story, and attempt to find a headline and a photo to feature with your post. There are technical conventions to help this process along: the open graph tags, which can encode for machines only, in the invisible code underlying a page, a title, description, image URL and other information optimized for machines. Twitter has its own set of meta tags intended to provide similar information. If the info is missing, Twitter and Facebook make a guess by looking at the rest of the story, which can backfire, so it's wise to design production systems and editorial workflows with social sharing in mind.

Based on this information, the social services construct a little image-headline-summary "card" that is displayed below and works in tandem with whatever the user chose to write in a post or Tweet. This introduces yet another complication: What should the post say? How long should it be? Should it duplicate the title and summary in the card (the answer, by the way, is "no," and I'm looking at you, Guardian.)

Again we have that potential for conflict between journalistic missions and commercial goals. At best we have an informative experience, at worst we have frustrating and misinformative clickbait., "on a mission to change what the world pays attention to," arguably changed the way many editors approached headlines, as documented by Poynter's Roy Peter Clark in a post that notably is constructed as a listicle, a clickthrough-inducing format. He described Upworthy headlines as being "praised for being irresistibly attractive and attacked for being cynically exploitative." Certainly Upworthy sacrifices the journalistic goal of informing for the more commercial goal of clickthrough.

All of this complexity creates a need for special editorial skills that reporters generally do not possess.

But didn't I say that writers should write their own headlines? Yes, and here's why.

Reporting is hard work. Writing is hard work. Transforming a reporter's notebook full of scratchings and/or a pile of interview tapes and/or a stack of documents and/or a database into a compelling narrative is neither simple nor easy. Headline-writing, which might at first seem to be just piling work on top of work, actually can help a writer focus. Does the story go where I had intended it to go? Is the lede graf on point? Where is the nut graf? Does it bear out the promise of the headline? (Does it even exist?)

A story title written by a reporter may not always work particularly well as a print headline, or as an online or social media title. But let's go back to the issues raised by centralization and specialization. The editor or online producer, overworked and in a hurry, benefits immensely from the writer's viewpoint and choice of language. What does the writer think is the most important part of this story to learn first? And does the story deliver on what the writer thought it was promising?

Even if the writer's headline is ultimately discarded, it can be hugely valuable.

So why isn't it universal practice? In many cases the writer may be living with a production system designed in the last century, for print. If your experience with a content management is primarily with a Web-based system such as WordPress, it may not occur to you that thousands of writers are still stuck in the last century, using a content management system designed for print, where online publishing is just an afterthought. Fielded data for digital publishing might not be supported, or if supported, might not be required. (In the editorial workflow system we implemented at Morris Publishing Group, we called it an "online headline" and made it a required element before a story could be submitted.) Or the problem might not be technical at all, but rather just a case of institutional lethargy. Change requires leadership.

Now let me turn my attention to Preet Bharara, the former federal prosecutor ("Banned by Putin, fired by Trump") and podcast host, and his Twitter thread on headlines. I encourage you to click and read it, not just for his Tweets, but also for the responses and discussion by others. Bharara cites a couple of famous headlines, notably "Let Detroit Go Bankrupt" on a Mitt Romney op-ed in the New York Times, and "Ford to City: Drop Dead" on the front of the New York Daily News.

While I don't think his examples support his claim, I'll quickly concede that bad headlines do get written by people who may give less attention to the story than they should. Sturgeon's Law ("90 percent of everything is crud") doesn't excuse it. The hard truth is this: People are lazy, and I don't mean headline writers. Readers will always take the path of least resistance, and if a headline is inaccurate, misleading, raises unanswered questions, exceeds the facts conveyed in the story, or simply lands with a thud, most people are going to be misinformed.

The headline has great power, and with great power comes great responsibility. If there is one thing that hasn't changed since the introduction of the headline, that's it.