YABG (yet another blog/gallery)

Notes on the attention economy

This is an excerpt from my journal - with extra research. Part one of a three part series.

A setup

I’m on a flight from Vancouver to San Francisco. It’s an early morning flight - 7am to 9:45am. Everyone’s either cranky or sleepy (or both). It’s a 6 seat per row plane. I start to people watch, and notice patterns. 2/6 people on my row are on TikTok. 4/6 people on the row ahead of me are on either Instagram reels or TikTok.

I sneak a peek at the feed of a curly-haired, freckled, married guy. The “algorithm” has figured out that he likes black and white motivational videos. He scrolls, ignoring the flight safety instructions. A scantily dressed woman dancing pops up on his screen as a flight attendant comes by to check if everyone’s wearing their seatbelts. He snaps back into reality, turning his screen off. Once the threat is averted, he goes back to staring at his screen.

There’s another guy seated in front of me. An Apple ecosystem ad - iPhone, Apple Watch, AirPods. I observe what he’s scrolling through. He uses TikTok in landscape - streaming audio to his pods. A few TikToks of the Afghan cricket team show up - they chase 30 runs in 6 balls. He speeds through the video at 2x, quits TikTok. Replies to a text on WhatsApp - but quits midway once there are blue ticks on his messages. He opens TikTok - this time to create a story about him being on a flight - making sure to focus on the screensaver on his watch face. He scrolls past the WhatsApp status his friends have posted. All this - under 3 minutes.

The only two people who didn’t use their phones were older passengers - falling asleep from exhaustion as soon as they boarded. The two people I described above put their phones aside once the flight is too high to receive network. They lean back, close their eyes - as Vancouver shrank, revealing magnificent views of the Canadian Rockies.

The irony on a flight bound for SF was stark. You see Palantir ads at airport gates, people talking about AI agents, AGI timelines and funding rounds - Yet the most successful technology on that plane was the one reducing human attention to its most primitive state. Much of today’s tech is centered around capturing attention, and the strategy has become terrifyingly effective. Here’s a look under the hood:

The team deploys a whimsically named tool: the Distractatron.
It’s a small TV screen, placed a few feet from the larger one, that plays a continuous loop of banal, real-world scenes—a guy pouring a cup of coffee, someone getting a haircut—each lasting about 20 seconds. Whenever a youngster looks away from the Moonbug show to glimpse the Distractatron, a note is jotted down.

Overly saturated colors gets a baby’s attention

Overly saturated colors gets a baby’s attention

As part of the autoplay test, we tested how long the countdown should be between episodes. 5 seconds, 10 seconds or 15 seconds. 10 seconds caused the biggest increase in hours watched. We thought that it gave people time to digest what they had just watched, but wasn’t too fast (5 seconds) where it became jarring. Interestingly, Netflix recently changed the countdown between episodes to 5 seconds.

This marks a critical shift from capturing attention to actively harvesting it. The leap from “amazing product” to “doomscrolling” wasn’t an accident; it was the result of specific economic and technological pressures.

The Siren’s Song

How a series of disjointed events joined forces to get to where we are today.

Ads = free software

How is Google free? How is messaging on WhatsApp free?

The earliest form of ad-based monetization at scale came from search engines. No, Google wasn’t the first in this race. A lesser known search engine predating Google, GoTo pioneered the pay-per-click model. Advertisers bid for keywords in automated auctions - if a user searches “cat”, show my videos first - I’ll pay when the user clicks my content. While this might seem commonplace now, it was a radical approach - one that Google took and ran with. Eventually, Yahoo realised 20% of its revenue came from people clicking on GoTo links - and bought GoTo. GoTo disappeared from the market, but the impact it left on the web was massive.

In essence, this is what Google’s AdSense does, bringing in HUGE money for them. As of 2024, ads constitute roughly 75% of Alphabet Inc.’s entire revenue. This form of advertising is called pull based advertising. Users pulling the content they want by searching for it explicitly. Instead of creating demand through ads (as it happened in TVs and Radios) - they focused on using existing demand well. Being user driven, the conversion rates are higher - with sustainable RoI.

While all this was happening - Facebook’s user base was growing slowly. Zuckerberg completely focused on connecting people - monetization wasn’t a priority. To him, it wasn’t a business model, but a way to offset costs.

It might be nice to get some ads going to offset the cost of the servers.

- Mark, in an interview with the Harvard Crimson.

It didn’t take long before they introduced Facebook Ads - where advertisements occupied space on your feed. A form of push based marketing - it analysed usage patterns and demographics to recommend ads on your feed. A step back in the user experience, if you ask me.

A push based marketing scheme is based on how popular a platform is than addressing user demand. TV(which is a platform) ads are useless if no one gets cable. It focuses on funneling more volume - with lower conversion rates than pull based marketing. Volume. A depressing implication of this is: The more time a user spends on a platform, the more the company earns.

Today - advertisements have gone one level deeper. Instead of occupying space on your feed separately - They’ve become a part of the content we consume (influencer marketing). Humans tend to trust other humans recommending products more than looking at a banner. This translates to better engagement rates, while eliminating ad production cost entirely.

As the famous saying goes: If the product is free, then you’re the product.

The mobile explosion

We have an economic incentive now, but how do we reach distribution at scale? The mobile phone was the harbinger of the attention economy. A simple before-and-after comparison reveals how:

Pre

Attention was naturally batched. You watched TV in the living room, took a bus to the movie theatre, logged into your PC to check Facebook. There was no other way to catch KBC if you didn’t sit in front of the TV at 9pm on Sundays. Attention consumption required mindful transitions to certain places at certain times - what we call Appointment-based consumption.

This fundamentally created a different media economic model. Attention required deliberate capturing - so companies optimized for duration and satisfaction. Movie theaters invested in epic storytelling because audiences committed 2-3 hours of attention. They competed on content quality and depth because switching required physical effort— changing channel bundles, walking to the record store. The economic incentive was retention, not just fleeting engagement.

Another critical feature emerged - Information is relatively scarce compared to human attention. Distributing media across the world came with enormous costs. Movies and music often remained local to specific geographies. Not everyone could make movies - equipment & screen rights were expensive. Record labels would only agree to distribute your music to the public if it was exceptionally good. This created natural quality filters.

Post

Contrast this situation with today: Mobiles eliminated appointment based consumption, creating “interstitial” moments - riding elevators, waiting in line. Media could reach you during hundreds of windows throughout the day that previously didn’t exist as consumption opportunities.

Persistent connections with push notifications became the first systems designed to interrupt human attention on demand. Once these micro-moments existed, they created a natural market gap. Empty attention became wasted potential revenue. Platforms had to optimize content for these fragmented windows or lose to competitors who did. 30-minute articles simply don’t fit a 2-minute elevator ride.

Simultaneously, media distribution economics inverted completely. Anyone could upload music to Spotify or SoundCloud. Anyone could publish videos on YouTube. Digital distribution eliminated geographic barriers. Built-in cameras and microphones meant production costs plummeted.

Information became abundant - while attention saturated quickly.

As a consumer - you could have realised this abundance, curated your information diet immediately. If there had been large enough collective realization, we wouldn’t be facing today’s attention crisis. Unfortunately, vast market forces acted exactly in the opposite direction. Attention needed to be harvested - so companies optimized for retaining and capturing attention at any cost - an incentive fundamentally misaligned with content quality.

The mobile phone didn’t just enable the attention economy - it made it an inevitable, logical economic conclusion. What began as a communication tool became an economic model where platforms competed for quantity of distracted moments.

Lip syncing is all you need

Okay - now we’ve set the stage with:

  1. A monetary incentive through ads.
  2. A distribution vehicle in the mobile phone.
  3. An attentional market gap.

It wasn’t obvious how shallow, short-form content was optimal in this context. Two industries solved this in different, ingenious ways:

Musical.ly: Content Creation friction

In 2012, Alex Zhu thought he’d cracked EdTech with Cicada - short (3-5 minutes) educational content. The logic seemed bulletproof: MOOCs had < 15% completion rates, so break it into chunks. It didn’t take off as expected, but it taught Zhu that people want to be entertained, not educated.

Zhu pivoted to musical.ly, incorporating learnings from Cicada - Easy Creation with 15-second videos, pre-loaded music, one-tap recording. Most videos were of lip-syncing and dancing, paving the way for Easy Consumption. It grew organically among teenagers (mostly girls) - lowering the barrier of multimedia content creation.

Users flocked to musical.ly to create content, not to consume it.

Bytedance: Content consumption friction

Back in Zhu’s home, China - Bytedance’s CEO, Zhang Yiming, had made it explicit that it was an AI company in the search space. Its genesis came from a different problem - the Chinese audience were struggling to find the right information (Google search was banned). The existing search engine, Baidu, had undisclosed ads in their search results.

In 2012, Bytedance released its first product - Toutiao (Today’s Headlines). It functioned as a news aggregator and recommender from multiple sources. Bytedance was legendary for its A/B testing. They ran over 1.5 million total experiments, 2000+ new experiments per day, and 30,000+ experiments running at any point in time. Every algorithm tweak, every user interface change was meticulously observed. All features had to lead to an increase in user engagement, average watch time per session, or number of sessions per day. It’s rumored that even Toutiao’s name was A/B tested.

Users flocked to Toutiao to consume content, not create it. Algorithm was Bytedance’s moat.

Convergence

Bytedance soon released Douyin - cloning musical.ly for the chinese audience. The one crucial differentiator - it was backed by Bytedance’s infamous algorithm. Douyin tracked how long you watched a video, how many times you shared it, as indicators of “liking” a video. It unearthed your deepest desires within 40 minutes of scrolling.

Douyin also logged your usage patterns - to feed into the next iteration of the recommendation algorithm, to make it even better. There were multiple advantages at play here - each compounding in themselves, while holistically creating a virtuous cycle.

Epilogue

Bytedance bought musical.ly in 2017 for their user base. They swapped out its backend with douyin’s algorithm. Rebranded it to TikTok. Numbers grew. 315 million downloads in Q1 2020 - the highest quarterly download count any app had ever achieved. Average time spent on TikTok increased from 29 minutes to 58 minutes between 2019 and 2024. The average american teen spends 1.5 hours on TikTok in 2023 (this has likely increased now).

Other apps followed suit - Youtube Shorts, Instagram reels, Facebook reels, Netflix’s Short laughs, Linkedin reels, Snapchat Spotlight. Combined - they take up 4.8 hours of a teen’s day ON AVERAGE. In contrast - global users spend 2.3 hours per day on social media. Users reach for their phones around 100-150 times a day. Each session then lasts for one minute on average.

What happens when we point this tech at 3 billion humans? Can we get ahead? What happened to the curly hair guy and the apple ecosystem guy? - These questions each deserve a post of their own, in parts 2 and 3. In the meantime, during the interstitial moments in your life - elevators, waiting rooms, commutes, at the cafeteria - observe people around you. How many impulsively reach for their pockets? How many are running away from their thoughts? Are they controlling the phones, or are the phones controlling them?

Reply to this post by email ↪