to hold a common wrong belief
People have the misconception it's our data being sold.The Social Dilemma · 2020 — Chapter 3 of 14 · 2'26" · C1
Tech Insiders Explain the Growth Dials, Explained: dial up the ads
On the surface, a speaker who was at Facebook explains how data becomes a model that predicts behavior. Underneath, it's an argument about power: the feed isn't tuned for you but to keep numbers rising, and the people running it can “dial that as needed.”
What happens
A speaker walks through what platforms do with user activity: the clicks and videos feed a model that predicts behavior, and the model decides what will keep someone watching. The explanation then turns to the three goals inside tech companies — engagement, growth, and advertising — and the idea that insiders can “dial that as needed.”
Listen for
misconceptionbusiness interestpull back ontend to trigger
00:00 / 02:26Space play · ← → lines · R repeat
Now
Phrase of the scene
misconception
a common wrong belief
At the start, someone corrects a common assumption: people think their data is being sold, and this line sets up the misconception.
People have the misconception that cats ignore you on purpose.
Also in this scene
- business interest
commercial advantage
- pull back on
reduce or scale back
- tend to trigger
tend to set off your emotions
- squeeze in
fit something into a tight schedule
- drive up your usage
push usage higher
Everything marked in this scene
25 notes across 31 lines
commercial advantage
The next line explains that handing over data would not serve the company's commercial advantage, said as plain fact.
It's not in Facebook's business interest to give up the data.hand over or surrender data
It's not in Facebook's business interest to give up the data.dummy subject: It is a dummy subject; the real subject is the infinitive phrase to give up the data.
It's not in Facebook's business interest to give up the data.In fast speech not in blends together; let the t pass lightly into in.
It's not in Facebook's business interest to give up the data.Do they is usually reduced; don't give both words full stress.
What do they do with that data?forecast what we will do
They build models that predict our actions, and whoever has the best model wins.subject clause: The whole whoever-clause is the subject: the person with the best model is the one who wins.
They build models that predict our actions, and whoever has the best model wins.typical time spent in one session
Nearing the end of his average session length.amount of advertising shown — Industry jargon for ad density; too technical for casual chat.
Decreasing ad load.reduce or scale back
A scrolling session slows near the average session length, and the system begins pulling back on certain content.
Pull back on friends and family.All of the can blur into one unit, with of reduced to a light vowel.
All of the things we've ever done, all the clicks we've ever made, allmake you continue watching
Where you'll go. I can predict what kind of videos will keep you watching.tend to set off your emotions trigger means to set off an emotional reaction, not to pull a gun
I can predict what kinds of emotions tend to trigger you.spectacular failures — Internet slang for joking about public mistakes; avoid it in formal writing.
The most epic fails of the year.fit something into a tight schedule
After a video test works, someone says 'Beautiful' and plans to slip a sneaker ad in before the next clip.
Beautiful. Let's squeeze in a sneaker ad before it starts.push usage higher
The film turns to company goals, starting with engagement: pushing usage up and keeping people scrolling.
There's the engagement goal: to drive up your usage, to keep you scrolling.the maximum amount of money
And then there's the advertising goal: to make sure that, as all that's happening, we're making as much money as possible from advertising.time clause: As introduces a simultaneous background: the money is made while all the rest is happening.
And then there's the advertising goal: to make sure that, as all that's happening, we're making as much money as possible from advertising.work out or solve
The explanation links each goal to algorithms whose task is to work out what to show next and keep the numbers rising.
whose job is to figure out what to show you to keep those numbers going up.relative clause: The whose-clause attaches the job to the algorithms mentioned just before.
whose job is to figure out what to show you to keep those numbers going up.adjust something whenever needed the dial is a metaphor for adjusting growth metrics
We often talked about, at Facebook, this idea of being able to just dial that as needed.increase the amount of advertising dial up means increase, not call
"Let's dial up the ads a little bit."In fast speech the t in little is light; let the phrase glide.
"Let's dial up the ads a little bit."degree of exactness
And so, that happ-- I mean, at all of these companies, there is that level of precision.Tech Insiders Explain the Growth Dials, explained
What's really going on
The scene reframes the product as prediction. Data isn't the commodity, the model is: “whoever has the best model wins.” The user experience is treated as a set of levers — “Pull back on friends and family” when a session is running long, then “squeeze in a sneaker ad” — and the moral distance sits in the language. Nobody says manipulate; they say dial. The calm, precise vocabulary is the point: “there is that level of precision.” What gets described is tuning at will: turn a number up when a goal needs it.
Theme
PowerWhoever holds the dials can turn what users see up or down on demand.
MoneyAdvertising sits alongside engagement and growth, with the aim of making “as much money as possible.”
Listen for
- not in — In fast speech not in blends together; let the t pass lightly into in.
- do they — Do they is usually reduced; don't give both words full stress.
- All of the — All of the can blur into one unit, with of reduced to a light vowel.
- a little bit — In fast speech the t in little is light; let the phrase glide.
They build models that predict our actions, and whoever has the best model wins.The value isn't the data itself but the model built on it: whoever predicts behavior most accurately has the edge.
I can predict what kinds of emotions tend to trigger you.The model is built to predict which emotions are likely to provoke a reaction in a person.
Beautiful. Let's squeeze in a sneaker ad before it starts.An ad gets forced into a gap in the session. "Beautiful" frames the placement as craft rather than as an intrusion.
We often talked about, at Facebook, this idea of being able to just dial that as needed.People at Facebook talked about adjusting things like a dial whenever needed — turn a number up when a goal calls for it.
Questions about this scene
Why isn't the company's product the data itself?
Because the value comes from the model built on that data. One line puts it plainly: “They build models that predict our actions, and whoever has the best model wins.”
What happens when a session nears its average length?
The sequence goes from slower scrolling and the end of the average session to “Decreasing ad load,” then “Pull back on friends and family.”
What are the three goals the speaker lists?
Engagement (drive up usage and keep you scrolling), growth (keep you coming back and inviting friends), and advertising (make as much money as possible from ads).
What do the "dials" refer to?
Controls that can be turned up or down on demand — “Turn the dial,” “Let's dial up the ads a little bit,” “Dial up monetization, just slightly.”
The Social Dilemma · 14 chapters
All chapters →
Up next · Chapter 4fall prey toPinterest President Can't Put Phone Down
Chapter 1rakes in moneyYou Are the Product
Chapter 2human futuresSurveillance Capitalism Explained
Watching · Chapter 3dial up the adsTech Insiders Explain the Growth Dials
Chapter 4fall prey toPinterest President Can't Put Phone Down
Chapter 5let the record showBen Takes the One-Week Phone Bet
Chapter 6conflateWhy Social Media Harms Teens
Chapter 7a mind of its ownAIs Plot Ben's Resurrection
Chapter 8reckon withThe Wikipedia Analogy
Chapter 9tilt the floorFake News Spreads Faster Than Truth
Chapter 10weaponizing social mediaFacebook Weaponized in Myanmar
Chapter 11digital FrankensteinsSpeaker Reframes the Tech Threat
Chapter 12willful ignoranceThe Civil War Interview
Chapter 13put the genie back in the bottleInterviewees Call for Regulation
Chapter 14set in stoneTech Insiders Explain What Went WrongChapter 3 of 14 · about 32 min to go