How Digital Headlines Work

Here's what happened, in order.

I'm sipping coffee on a Saturday morning. I no longer subject myself to a daily dose of trauma-infused "news," so I take extra time on the weekend to stay informed instead. An algorithm hands me a headline. Not the best headline available that morning. Not the most important one. The headline the algorithm calculated I would click. I still have freedom of choice over what's on the menu in front of me — but the platform holds the far bigger power: defining what's on the menu in the first place. That's algorithmic power, and there's a lot to unpack in that one distinction. We'll come back to it another time.

I clicked. The story: a U.S. military officer, arrested earlier in the summer for a public protest calling for senior officials to be impeached, detained again this week after doubling down in a television interview.

My brain did not process this in the order I'd like to admit.

First move: this is a heroic story nobody is covering enough.

Second move: wait, is this even accurate — I think I recall something about this from earlier in the summer.

Third move: is this too political to write about. Will it trigger people? Will it cause a fight?

Fourth move: what would I actually be willing to do about an injustice like this?

Notice the order. I built the story before I checked the story. I decided how I felt before I confirmed what actually happened. Then I ran a cost-benefit analysis on whether to say so out loud. That sequence has a name in every framework I teach: confirm the narrative first, interrogate it never, self-censor last. It isn't a flaw specific to me. It's the default setting for a brain running on autopilot — fast rather than deliberate, System 1 rather than System 2, in Daniel Kahneman's terms.

This Isn't Just a Me Problem

If it feels like an oddly specific failure mode, it isn't. A 2016 study out of Columbia University and the French National Institute for Research in Digital Science tracked 2.8 million shared links across five major news outlets and found that 59% of them had never actually been clicked by the person sharing them. Six in ten people were opining on, endorsing, or attacking a story based entirely on a headline and a summary. Not because they're careless. Because the sequence I ran that Saturday morning — form the opinion, skip the verification, decide later whether to say it out loud — is the default sequence for a brain moving at feed speed. The algorithm didn't invent that shortcut. It just built a delivery system fast enough to run it thousands of times a day.

The Part I Almost Skipped

Here's the part that turns this from a hot take into an actual look in the mirror: I never sat with the strongest version of the other reading.

Uniform nonpartisanship isn't a partisan invention built to silence one officer under one administration. It exists so that no administration, of any party, can use active-duty personnel as a political prop. An officer protesting in uniform doesn't just voice an opinion — he borrows the institution's authority to amplify it. That's a strategic choice with real institutional consequences, whether or not it's also a courageous one. Whether it's grounds for punishment is a separate, genuinely open question.

You can distrust the administration in question and still think the enforcement here is doing exactly what a nonpartisan-uniform rule is designed to do. You can support the administration and still think this looks like a pretext to punish a critic. Both readings survive contact with the facts. I only wrote one of them down.

That's the actual bias at work. Not left or right. The instinct to resolve the tension the moment it appears, because resolution feels like clarity, and clarity feels good — regardless of whether it's earned.

The Thirty-Second Pause

So here's where I'm landing this, differently than I would have an hour earlier.

Notice how fast you move from headline to hero, or headline to villain — whichever direction your particular wiring runs. Then ask what would have to be true for the other reading to be the accurate one: for this to simply be an institution enforcing a rule that predates this administration and will outlast it. Sit with both versions for thirty seconds before deciding which one you believe.

What did your thirty seconds tell you about the thinking you don't notice?

This is Anchoring and Availability bias running exactly as designed, at the speed a news feed now runs. The algorithm supplies the anchor. Availability bias fills in the rest from whatever fits the story already forming. The correction — the thirty-second pause, the deliberate construction of the opposing case — is the only part of this sequence that isn't automatic. It's also the only part that's actually optional.

Frequently Asked Questions

What is algorithmic power?

Algorithmic power is the ability of a platform to control which headlines and stories a person is even offered, before that person makes any choice of their own. A reader retains freedom of choice over what to click, but the platform retains the far greater power of defining the menu of choices in the first place, typically optimized for engagement rather than importance or accuracy.

What does "confirm the narrative first" mean?

It describes the default sequence the brain runs on a contested headline: forming an emotional conclusion about a story, then checking whether the story is even accurate, then deciding whether it's safe to say anything out loud, and only then, if at all, considering the strongest opposing interpretation. Naming that sequence is the first step to interrupting it.

What is the thirty-second pause technique?

Before accepting a snap judgment about a contested story, deliberately construct the strongest version of the opposing interpretation and sit with both readings for thirty seconds before deciding what you believe. Most contested stories support more than one reasonable reading; the bias is not which side a person lands on, but resolving the tension immediately because resolution feels like clarity.

How many people share news without reading it first?

A 2016 study from Columbia University and the French National Institute for Research in Digital Science tracked 2.8 million shared links across five major news outlets and found that 59% had never been clicked by the person sharing them — meaning six in ten people were reacting to, endorsing, or arguing over a headline and a summary alone, without reading the underlying story.