Awe Travels
In 2012, Jonah Berger and Katherine Milkman—both at the Wharton School—published What Makes Online Content Viral? in the Journal of Marketing Research. They pulled six months of New York Times articles, mapped each one to its share count, and ran the regression I wish more of us working in attention actually paid attention to.
Their finding was unsexy and load-bearing: virality is driven by emotional arousal, not valence. Both high-arousal positive emotions (awe, hope) and high-arousal negative emotions (anger, anxiety) drove sharing. Sadness suppressed it. Calm satisfaction suppressed it. What mattered was whether the content lit up the reader's nervous system at all.
When I run content teams, this is the study I think about. On the days when digital storytelling feels like the wrong place to be spending a life, it is the thing that keeps me at the desk.
The two strongest emotional drivers in the regression landed within a few percentage points of each other. Anger raised the odds of an article making the most-emailed list by 34%. Awe by 30%. Both worked because both are high-arousal. The algorithm doesn't prefer one over the other.
The practitioner has to.
This is interesting in a paper. It is consequential in a campaign.
The field test
In 2024 I ran the content operation for RFK Jr.'s presidential bid. We produced hundreds of pieces of content and achieved 2.5 billion organic impressions in twelve months. A meaningful slice of that work was a public, deliberate application of what the paper had described.
We knew Trump would post a response to President Biden's State of the Union on March 7, 2024. We knew it would be grievance. We knew it would be anger. We decided to produce a cinematic love letter to America—a clear-eyed, empathetic view of what the country was and what it could become.
His response crossed 6 million views.
Ours dropped the same week.
40 million views.
That is not a ten-percent delta. It is a 6.6× multiplier under broadly comparable conditions—same news cycle, same week, against an opponent with deeper paid amplification. Different teams, different audiences, different production values. The valence was the choice we controlled.
Why
The reason isn't mysterious. Berger and Milkman explained it twelve years before we shipped.
Anger pulls people in toward themselves. Awe pulls people together. People share what they want their friends to feel. Awe scales because awe is what social bonding looks like in the language of an algorithm.
Anger remains the most-built emotional category in civic content because it's the most-trafficked one in any given week. The engagement graph rewards it. The cable news rhythm rewards it. The aggrieved voices on every platform get the early traffic and the early reach.
More than anything, outrage is easy. Awe takes everything you have.
In the years I have spent in this work, here is what I have come to see. Rage is a brush fire. It erupts, scorches a field, dies by morning. Awe is a wildfire that starts small and keeps growing. It moves through people. It persists in the heart.
But that is a short-term signal. The long-run signal—whether a piece of content stays in the timeline, gets resurfaced, gets quoted in a conversation six weeks later—runs on awe.
Most communicators have spent a decade building for the news cycle because the news cycle pays in the same week. Anger is the default register on every side. Awe is harder on every side. And awe wins more often than the default suggests.
The awe piece outperformed the anger piece by 34 million views on the same news moment.
The data does not change. What changes is whether the practitioner builds for the longer signal—work that doesn't only catch a moment but is still being shared six weeks later. The choice the algorithm doesn't make is the one the practitioner has to. Awe is the harder choice. It is also the one that travels further than the data alone suggests.