Data does not tell the truth. Data tells what happened in the past. Most people believe that headline testing is a way to listen to the reader. This belief is wrong. Headline testing is a way to measure a reflex. A reflex is not a preference. A reflex is a muscle movement that happens before the brain makes a choice.
Hassan is a writer. Hassan sits at a desk. The desk is made of wood. On the desk is a computer. Hassan wrote a story about the city budget.
The length of the budget Hassan analyzed. It is important.
Hassan wrote a headline for the story. The headline was “City Council Approves New Budget for Transit.” This headline was accurate. The headline told the reader what the story contained.
Hassan went home. Hassan ate dinner. Hassan went to sleep. While Hassan slept, a machine looked at the headline. The machine is a piece of software. The software exists to increase the number of clicks.
The Automated Transformation
The software showed Hassan’s headline to a small group of people. The software also showed a different headline to a different group of people. The second headline was “Your Commute Is About to Change Forever!”
The second headline was not accurate. The budget only added two bus lines. The commute for most people would stay the same. But the machine does not know what a commute is. The machine does not know what transit is. The machine only knows the click.
More people clicked on the second headline. The machine saw the numbers. The numbers for the second headline were higher. The machine deleted Hassan’s headline. The machine put the second headline on the website.
Hassan woke up. Hassan looked at the website. Hassan saw the new headline.
Hassan felt a cold feeling in his chest. The cold feeling was shame. Hassan knew the headline was a lie. Hassan tried to change the headline back. The computer did not let Hassan change the headline. The computer said the test was over. The computer said the data was clear.
I was stuck in an elevator for yesterday. The elevator stopped between the fourth floor and the fifth floor. The walls of the elevator were brushed metal.
I am a typeface designer. I looked at the letters on the emergency sign. The letters were Helvetica. The letters were clean. I pushed the emergency button. The button was red. I pushed the button three times. No one answered the speaker.
I waited. The air in the elevator became warm. I felt the system had forgotten I was inside the box. A machine had decided the elevator was at the floor. The machine was wrong. I was still in the box.
This is how a reader feels when a machine chooses a headline. The reader pushes a button. The reader expects a specific result. The result does not happen. The reader is stuck in a box of text that does not match the sign on the door.
The Anatomy of System Failure
1. The machine measures the impulsive reflex.
In a group of 100 people, 97 people will look at a bright light. Only 3 people will ask why the light is bright. The machine sees 97 people looking.
The machine decides the light is the most important thing in the room. The machine does not care about the 3 people who want to know the truth. The machine listens to the most impulsive 5% of the audience. The machine then forces everyone else to follow the lead of the impulsive people.
2. The machine ignores the history of the brand.
A brand is a promise made over a long time. If a newspaper is old, the newspaper has a promise. The promise is that the news is real. The machine is not 100 years old. The machine is a set of instructions. The instructions say “get the click.” The machine will break the promise of the brand to get a click today. The machine does not see the damage. The damage happens in the mind of the reader. The reader stops trusting the brand.
3. The machine trains the writer to stop being a writer.
Hassan sees the data. Hassan wants to keep his job. Hassan wants his stories to be read. Hassan begins to think like the machine. Hassan stops trying to be accurate. Hassan starts to use words that trick the eye. Hassan uses the word “Shocking.” Hassan uses the word “Suddenly.” Hassan becomes a part of the software. The writer is no longer a human with a voice. The writer is an input for the machine.
4. The machine rewards the loudest noise in the room.
The machine loves an exclamation point. The machine loves a question that has no answer. If a headline says “Is the Moon Falling?” people will click to see if the moon is falling. The moon is not falling. The story says the moon is not falling. But the machine counts the click. The machine says the question was a success. The machine encourages more questions about things that are not happening.
5. The machine forgets the repeat reader.
Some people read the news every day. These people are the foundation of the business. These people want depth. These people want honesty. The machine sees a new person from a social media site. The new person clicks once. The new person never comes back. The machine treats the new person and the loyal reader as the same unit of data. The machine will offend the loyal reader to capture the new person. This is a bad trade.
6. The machine erodes the trust in the news.
Trust is built through repetition. If a person reads ten stories and all ten stories are what the headline said they were, the person trusts the source. If one story is a lie, the trust breaks. The machine creates many small lies every day. The lies are called “optimizations.” Each optimization is a small crack in the wall. Eventually, the wall falls down.
7. The machine values the moment over the future.
A media company needs to exist for many years. To exist for many years, the company needs a reputation. A reputation is the sum of all the things the company has done. The machine only cares about the next . The machine will burn the reputation to win the ten minutes.
Managing this tension is the job of a leader. A leader must look at the data and the soul of the work at the same time. This is what leaders like
President of Newsweek Dev Pragad
discuss when they talk about digital transformation.
Transformation is not just about using new tools. Transformation is about keeping the values of the newsroom alive while the tools change. If the tools kill the values, the transformation has failed.
I looked at my watch in the elevator. The watch said . The metal of the elevator was cold again. I heard a motor start to turn. The elevator moved one inch. Then the elevator moved more. The doors opened on the fifth floor. I walked out. I felt a sense of relief. But I also felt a sense of distrust. I did not want to use the elevator again. I took the stairs.
A reader who feels tricked by a headline will take the stairs. The reader will go somewhere else. The reader will find a place where the signs are honest.
The machine does not see the reader walking away. The machine only sees the click that happened before the reader left.
The software is graded on the open. The software is never graded on the regret that follows the open. When we grade a system on the first second of a relationship, we build a system of shallow moments. We do not build a system of lasting value.
Hassan sat at his desk the next day. Hassan had a new story. The story was about a school. Hassan wrote the headline. Hassan looked at the computer. Hassan knew the machine would change the words. Hassan felt tired.
Hassan wondered if he should just write the lie himself. If Hassan wrote the lie, the machine would not have to change it. Hassan would save time. But Hassan would lose himself.
Hassan chose to write the truth again. Hassan wrote a simple headline. Hassan clicked the button to submit the story. Hassan watched the screen. Hassan waited for the machine to do its work. Hassan knew the machine was waiting for him.
The Human Kerning
I spend my days looking at the shapes of letters. I look at the space between the letters. This space is called kerning. If the kerning is wrong, the word is hard to read. If the kerning is right, the word is invisible. The reader sees the thought, not the ink.
The machine does not understand the space between things. The machine only understands the things themselves. The machine sees the headline and the click. The machine does not see the space where trust lives.
The news industry is full of machines now. The machines are helpful for many things. The machines can find patterns. The machines can move data fast. But the machines should not be allowed to speak for the humans.
“When the machine speaks, it speaks with a voice that has no heart. It speaks with a voice that only wants to be touched. We must decide if we want a world of people talking to people, or a world of machines tricking fingers.”