The Social Aspect of Podcast Listening: Why Audio Is Becoming the New Shared Medium

For most of its 20-year history, podcast listening has been one of the most private acts on the internet.

You put headphones in. You walk by yourself. You hear a voice in your skull. Nobody around you knows what you’re listening to. Nobody at the office knows what episode you finished on the train this morning. Even your spouse, if you live with someone, usually doesn’t know what you’re hearing through your AirPods at 7am on a Tuesday.

This is a strange situation when you think about it. Television is social. Movies are social. Books are social (book clubs, conversations, “have you read it”). Music is social (concerts, playlists, “what are you listening to lately”). Podcasts, this medium that millions of thoughtful people consume more of than they consume any of those other formats, has been almost entirely solitary.

That is changing. I think it is changing fast. And I think the social layer on top of podcast listening is going to become one of the most important shared cultural mediums of the next decade.

This essay is the long argument for why.

The numbers tell a quiet story

A few data points to ground this:

• Around 545 million people globally listened to a podcast in the last month (Edison Research, 2025).
• The average weekly listener consumes around 7 hours of podcast audio per week.
• The average “core” listener (the kind of person who follows specific hosts, builds queues, talks about episodes) consumes 10 to 15 hours per week.
• The most popular podcast app, Spotify, hosts over 6 million podcasts.
• In thoughtful professional cohorts (founders, PMs, investors, scientists, designers), the average is closer to 12 hours per week, often replacing the time they would have spent reading trade publications.

Now compare that to print: the average American reads about 12 books per year. That’s roughly 70 hours of reading annually. Heavy podcast listeners are consuming the same amount of audio in 6 weeks.

The cultural weight of audio in the lives of thoughtful people has crossed a threshold. It is now the primary shared diet of certain professional and intellectual communities. The friends of mine who I think of as the most engaged with ideas are roughly twice as likely to mention a podcast as a book in a normal conversation.

The medium has scale. What it doesn’t have yet is shape.

The four phases of any new medium

This is the pattern. Watch for it.

Phase 1: The medium itself emerges. Someone invents the printing press, the radio broadcast, the LP record, the YouTube video, the podcast. The thing exists. People consume it individually.

Phase 2: Distribution gets built. Bookstores and libraries. Radio networks. Record stores. The YouTube algorithm. Spotify. The medium becomes accessible.

Phase 3: Social formats develop on top of the medium. Book clubs. Listening parties. Movie nights. Watch parties. Reaction videos. The medium becomes shared, not just consumed.

Phase 4: The medium becomes a cultural reference layer. Quoting from it. Remixing it. Building entire identities around it. “Are you a Beatles person or a Stones person.” “Have you watched the Sopranos.” “Did you read Sapiens.” The medium isn’t just content anymore. It’s the substrate of conversation.

Podcasts are deep into phase 2 and barely into phase 3. The medium exists. The distribution is solved. The social formats are missing.

Twitter (X) is doing some of phase 3 informally, in the form of “the new Dwarkesh episode is incredible” threads. Reddit subreddits do some of it. Group chats do most of it. But there is no Goodreads for podcasts. There is no IMDb for podcasts. There is no shared communal infrastructure for the thing that hundreds of millions of people now do for 10 hours a week.

That gap is the opportunity.

What “social podcast listening” actually means

Let me be specific about what I mean, because the phrase “social listening” gets overloaded.

I don’t mean live group listening (everyone presses play at the same time on a Zoom call). That format exists for music (Spotify Group Sessions, Apple Music SharePlay) but it’s tiny because the constraint of “all listen at the same moment” doesn’t fit how busy adults consume audio.

I don’t mean broadcasting (live AMA-style listening events with hosts). Those are events, not a medium.

I don’t mean comments under episodes (the YouTube model). Comments are anonymous, broadcast, low-signal, and they actively repel thoughtful people.

What I mean is: the asynchronous, intimate, peer-to-peer layer where the actual cultural metabolism of an episode happens.

Concretely:

• A small group of friends who listen to the same shows.
• The act of marking specific moments while listening (the modern equivalent of dog-earing a page).
• The asynchronous exchange of those marked moments inside the cohort.
• The conversation that develops in the margin of the episode itself, not in a separate forum.
• The cohort’s emerging taste, what they listened to last month, what they’re recommending this week.

That’s social podcast listening, properly defined. It’s the missing layer.

Why it didn’t happen until now

There were three technical and cultural blockers. All three are now lifted.

Blocker 1: You couldn’t see what your friend was listening to. Until recently, podcast apps were closed. Even Spotify’s API didn’t expose third-party access to “currently listening” data until a few years ago. Without that primitive, you can’t build a friend-graph around shared listening.

Blocker 2: You couldn’t easily capture a moment in audio. Compare to text. If I’m reading a book and want to share a line, I can screenshot it, type it, or use Kindle highlights. Audio has none of that. The native “share an audio moment” experience has been clunky-to-impossible. Without it, the unit of social exchange (the marked moment) doesn’t exist.

Blocker 3: There was no critical mass of thoughtful listeners. A decade ago, podcasts were either NPR-style mainstream or comedy. The “thoughtful business and intellectual podcast” category that supports cohort-style sharing (Acquired, Dwarkesh, Invest Like the Best, Lenny’s, In Depth, Founders, Knowledge Project, No Priors, 20VC, Cheeky Pint) barely existed. Now it does, and the audiences inside it are dense networks of operators who already know each other or share peer groups.

All three blockers are gone. Spotify’s API exposes playback. On-device transcription means voice notes can be captured privately and shared selectively. The thoughtful-listener audience is in the millions, and is densely connected.

This is the technical and cultural moment when the social layer should arrive. It will be built by someone. It is currently being built by several someones, including me.

What changes when listening becomes social

If you accept the thesis (podcast listening goes from solitary to social), the second-order effects are interesting.

Listening gets more careful. When you know your friends will see what you marked, you mark more thoughtfully. When you know an episode will be discussed Sunday, you don’t half-listen at 1.8x while doing dishes. Social pressure improves attention. The same way a book club makes people read more carefully than they would alone, a podcast cohort will make people listen more carefully.

Specific episodes become events. Right now a new Dwarkesh episode is just another item in your queue. In a social-listening world, the moment Dwarkesh drops a Demis Hassabis interview, your cohort coordinates around it. “Are we all listening to this one.” “Did you get to the bit about scaling laws.” The release becomes a shared event in the way an HBO series finale used to be.

Hosts get qualitative signal they’ve never had. Right now podcast hosts get downloads, ratings, and the occasional tweet. With a social-listening layer, hosts can see which specific moments in their episodes resonated, where the marking density was highest, where listeners dropped attention. This is the difference between the analytics a YouTuber gets and the analytics a podcast host gets today. The gap is enormous.

Discovery gets dramatically better. The hardest problem in podcasts is “what should I listen to next.” Apple’s and Spotify’s recommendation engines are bad because the underlying signal (downloads) is bad. A social-listening graph (your trusted friends finished this and marked these moments) is a much better signal. The recommendation surface goes from “algorithm” to “friend’s taste,” which is what every working recommendation system in the world eventually converges on.

A new kind of cultural artifact emerges. The “best moments” of an episode, marked by an interesting cohort, are themselves a kind of content. The annotated episode becomes a new format. Imagine reading the Acquired Microsoft episode the way you’d read a great essay, with the timestamps your favorite operator marked sitting in the margin.

Loneliness gets a little smaller. This is the smallest-sounding effect and possibly the most important. Listening to podcasts is currently a lonely activity. Even when 100,000 people listened to the same episode this morning, you have no sense of them. Social listening fixes that. The act stays solo (you with your headphones on a walk). But the thinking about the act becomes shared. That changes something about the experience of being a thoughtful person in 2026.

Why this is the right shape (and the shapes that won’t work)

A few formats have been tried and don’t fit. Worth naming them explicitly so we don’t drift back into them.

The reaction video. YouTube’s reaction-video format works for music and movies, where the visual reaction adds something. Podcasts are audio. Reaction videos to podcasts are awkward. This format will not become podcast listening’s social layer.

The infinite-scroll feed. Some apps have tried to build a TikTok-style feed of podcast clips. The problem is that the value of podcast moments is contextual (what was being discussed, why the moment landed) and feed-style scrolling strips context. The format also drives toward viral-bait clips, which is exactly the opposite of what the thoughtful-listener audience wants.

The public profile. A Goodreads-for-podcasts where everyone has a public profile of what they’ve listened to is a tempting design, but I think it’s wrong. The act of marking is intimate. Most people don’t want strangers seeing what struck them mid-podcast. Public profiles will push the marking to be performative, which kills the format. The right shape is private cohorts, not public profiles.

The single-show fanbase. Discord servers and subreddits work for the most rabid fans of a single show. They don’t scale to “I listen to 8 shows and want to discuss with my 5 friends who listen to 6 of those.” The right shape is friend-graph-centered, not show-centered.

The host-led community. Hosts who try to build community around their own show end up with parasocial-leaning fanbases. The dynamic gets weird when the host is in the room. The right shape is peer-only, without hosts present.

What does work: small private groups (4 to 8), friend-defined cohorts, asynchronous, anchored to episode timestamps, intimate. That’s the shape.

How this connects to what people are already doing

If you watch carefully, the social podcast layer is already happening, just in the wrong tools.

• Group chats with friends sending each other Spotify share links to specific episodes (every founder I know has at least one of these).
• Twitter quote-tweet threads of “this Acquired episode is incredible” with replies from people who finished it.
• Slack channels at AI labs and venture firms with the same dynamic, professionalized.
• People texting their best friend a 30-second voice memo describing the moment that struck them on a walk.
• The custom-emoji reactions to "🎧 listening to” status updates in friend Discords.

All of these are people trying to do social podcast listening with tools that weren’t built for it. The friction is high. The signal gets lost. But the behavior is there. The behavior says the demand exists, and the demand is waiting for the right product.

When the right product arrives, the migration will be fast. The behavior is already in place.

What the social layer will look like five years from now

Predictions, for the record:

• A handful of podcast-social products will exist by 2028, with significant traction.
• The dominant format will be small private cohorts (not feeds, not profiles).
• The most popular product will probably integrate with Spotify rather than try to be a player itself.
• Capture (the act of marking a moment) will be a major new behavior, with millions of people doing it weekly by 2029.
• Podcast hosts will start citing “high-mark moments” in their show notes the way YouTubers cite “chapter timestamps.”
• A few podcast clubs will become well-known the way certain book clubs are (Reese, Oprah). Probably one operator-club, one design-club, one literary-club.
• The cultural status of “what podcast cohort are you in” will be a recognizable social signifier, the way “what book club” or “what subreddit” is now.

Some of these will be wrong. The general direction will be right.

What it means for the medium itself

The biggest second-order effect is that podcasts themselves will get better.

Right now, the feedback loop between podcast hosts and listeners is weak. Hosts make episodes. Episodes get downloaded. The host gets a number. The host has no idea which moments landed, which ran too long, which were the actual quotable beats. This is why so many podcasts are 90 minutes long when they could be 50, and why so few of them have the editorial discipline of a great magazine essay. The feedback loop is too coarse.

When listening becomes social, the feedback gets fine-grained. The signal “the audience marked moment X at minute 23” tells the host exactly what worked. The signal “the marking density dropped to zero after minute 47” tells the host where to cut.

This will produce better podcasts. It will reward editorial discipline. It will gradually push the medium toward the kind of careful craft that print magazines used to enforce when editors had control. The medium currently has no equivalent of an editor; the social-listening layer becomes the soft editor by aggregation.

What you can do right now

If you want to live in this future a little early:

• Find the 3 to 5 friends who listen to the same kind of podcasts you do. You probably already know who they are because they’re the people you currently text Spotify links to.
• Pick one shared episode per week and listen to it. Use any cadence that works. Sunday-night discussion is the default for a reason (it’s when book clubs meet).
• Mark moments while listening. Use whatever tool. Margin was built for exactly this; voice memos with manual timestamps work; the iPhone Notes app at red lights works.
• Share the marked moments asynchronously. Group chat is fine. The point is the exchange.
• Notice what changes. I’d bet a small amount of money that within four weeks your listening gets more careful, your retention improves, and your friendship with those 3 to 5 people deepens around an unexpected new ritual.

This is the soft launch of the social podcast era. You can participate now. You don’t have to wait for the products to arrive.

The deeper bet

There’s a bigger argument lurking inside all of this. It’s that the dominant shared cultural medium of any era is whatever medium the most engaged people consume in the largest volume.

For my parents’ generation, that was television. For the boomer professional class, it was print magazines. For early internet adults, it was blogs. For social-media-era adults, it was Twitter.

For the cohort that will define the next decade of cultural and economic life, it’s podcasts. Audio is the substrate of their thinking the way TV used to be the substrate of suburban America’s thinking. And the social layer on top of audio is going to be where the genuinely interesting thinking and recommending and discussing happens.

I’m building Margin because I want to be early to that layer. I think it matters more than the surface (a note-taking app) suggests. The deeper bet is that audio becomes the new shared medium of the smart internet, and the social cohort layer around it becomes the place where modern culture is metabolized.

If I’m right, the period we’re in right now (mid-2020s) is the same kind of inflection point that Twitter was in 2008 or YouTube was in 2007. The product that ends up owning the social layer of podcast listening will be one of the platforms of the next decade.

If I’m wrong, I’ll have spent a few years making a beautiful note-taking app, and I’m fine with that too.

Selinay

If you want to live this thesis with me, Margin is in TestFlight. The personal capture works today. The cohort layer is what’s coming next.