German argues that the easy dopamine of genuine social connection ran out around 2014, forcing platforms to inject high-pressure UX patterns — streaks, autoplay, haptics, infinite scroll — into ever-thinner reward seams. He frames this thermodynamically rather than morally: each technique yields less per shot and leaves behavioral residue, just like fracking accepts worse EROI and groundwater contamination as collateral costs.
Concedes the fracking framing and reveals that internally the metric that mattered was not raw engagement-per-session but engagement-per-session at constant tolerance. This is a private acknowledgment that the marginal user was becoming harder to move, validating German's thermodynamic argument with insider evidence.
Extends German's metaphor by noting that TikTok's For You page is a horizontal well in industrial terms — the same dopamine seam drilled sideways across thousands of micro-fractures (15-second clips) rather than vertically into longer-form content. This reframes short-form video not as a new medium but as a more efficient extraction technique on a depleted resource.
A short essay titled Dopamine Fracking by independent writer Ivan German (igerman.cc) climbed to 263 points on Hacker News, drawing the usual mix of seasoned skeptics and reformed growth-hackers into a long comment thread. The argument is one line long and one decade overdue: consumer software stopped drilling shallow dopamine wells and started fracking — pumping high-pressure UX patterns into ever-thinner reward seams to keep the per-session numbers from collapsing.
The metaphor is precise rather than poetic. In oil extraction, fracking is what you do once the easy crude is gone: you inject solvents and pressure into rock that would otherwise yield nothing, you accept worse Energy Return on Investment, you tolerate groundwater contamination as a collateral cost. German maps each piece of that onto the modern feed. The easy dopamine — a real message from a friend, a photo you wanted to see — is the equivalent of light sweet crude, and it ran out around 2014. What replaced it: streaks, badges, autoplay, haptics on every tap, algorithmically smoothed infinite scroll, push notifications that interrupt sleep. Each is a fracturing technique. Each yields less per shot. Each leaves residue.
The HN thread broke along predictable lines but with unusually high signal. One top comment, from an ex-Meta engineer, conceded the framing and added a detail: internally, the metric that mattered was not engagement-per-session but engagement-per-session at constant tolerance — a private acknowledgment that the marginal user was becoming harder to move. Another commenter pointed out that TikTok's For You page is, in industrial terms, a horizontal well: the same dopamine seam, drilled sideways across thousands of micro-fractures (15-second clips) rather than vertically into longer content.
The reason this framing punches above the usual "attention is bad" essay is that it replaces a moral argument with a thermodynamic one. Moral arguments about engagement design have failed for fifteen years because they require platforms to act against incentives; the depletion argument predicts that the incentives themselves will fail. This is the same shift that turned climate from an ethics conversation into a balance-sheet conversation: once the numbers stopped working, the behavior changed faster than any campaign managed.
The depletion mechanic is real and measurable. Neuroscience has known since the early 2000s that dopaminergic reward systems exhibit tolerance — repeated stimulation downregulates D2 receptors, raising the threshold for the next hit. What the fracking metaphor adds is the economic layer: a platform's job is not to deliver dopamine but to deliver it per user-minute at a rising rate, because growth models price the user as an annuity. When tolerance compounds faster than acquisition, the well is dry regardless of how clever the next interaction pattern is. You can see this in the metrics platforms have quietly stopped reporting: time-on-platform per DAU, session count per user, the basic engagement durability numbers that defined the 2015–2019 era of attention-economy bragging.
The comparison to real fracking goes further than German pushes it. Real fracking is profitable on a per-well basis but a disaster on a portfolio basis — the decline curves are so steep that operators have to drill new wells constantly to maintain output, which is why the shale industry has burned roughly a quarter-trillion dollars in net free cash flow since 2010. The attention industry has the same shape. Every "new" engagement mechanism — Stories, Reels, Shorts, streaks, daily check-ins — has a shorter half-life than the last. Snap's streak was novel in 2015 and load-bearing within a year. BeReal was a 2022 phenomenon and a 2023 footnote. Threads launched, peaked, and flattened inside ninety days. Each successive extraction technique works for less time before the seam taps out, which is the precise definition of a depletion economy.
The HN thread surfaced one objection worth taking seriously: maybe the issue isn't depletion but substitution — users haven't lost their capacity for dopamine, they've moved their attention to whichever platform is currently drilling the freshest seam. That's plausible but doesn't refute the argument; it explains the churn. A platform that depletes faster than it acquires is just a slower-failing fracking operation, and the industry's growing reliance on cross-platform attention arbitrage (TikTok creators farming to Instagram, YouTubers cross-posting to Shorts) is itself evidence the underlying reservoir is getting harder to reach.
If you ship consumer software, the operational implication is that engagement design optimized for the 2018 playbook is now actively destroying lifetime value. Push-notification CTRs are down across the board; the industry-wide median sits around 1.5% for non-transactional pushes, down from roughly 4% five years ago. Streak mechanics, which used to lift D7 retention by double digits, now produce measurable churn spikes the day a user breaks one — a tolerance signature. The honest read of the data is that aggressive engagement engineering is now negative-sum for most products: it pulls forward usage you would have gotten anyway and accelerates the moment the user uninstalls.
The practical move is to audit your own product for fracking patterns and price them at their real cost. Every haptic on a non-essential interaction trains the user's nervous system to discount your haptics. Every autoplay trains them to mute their phone. Every "you haven't opened the app in 3 days" notification is a free coupon you're handing to your competitor — because once the user feels manipulated, the next click goes to whichever app feels less manipulative, even briefly. The B2B and prosumer products that have grown without these patterns (Linear, Cursor, Arc-style focused tools) are not virtuous; they're just operating in a market where attention isn't yet depleted, which is the same advantage a wildcatter has when they hit virgin rock.
For anyone building agentic or AI-first interfaces, the warning is louder. An LLM that can generate infinite personalized engagement bait is, in extraction terms, a horizontal drilling rig — and the same depletion math applies, faster. The teams currently optimizing for "AI-driven re-engagement" are about to learn what shale operators learned in 2014: the first wells are spectacular, the average wells are mediocre, and the curve is steep.
German's essay doesn't propose a fix, and that's correct — depletion economies don't get fixed, they get exited. The platforms that survive the next cycle will be the ones that switch from extractive engagement to whatever the attention-economy equivalent of renewables is: tools that produce something the user keeps, rather than experiences the user discards. The fracking metaphor's final gift is the timeline: shale took roughly fifteen years to go from miracle to write-down. The attention industry started its frack around 2014. The math says we're closer to the end of this era than to the middle.
Reminds me of Adornos "Dialektik der Aufklärung" and its take on what he calls the "Kulturindustrie". Almost 100 years ago he foresaw how the cultural offerings of society get commodified and chopped into bite sized chunks for each individual to receive theirs. He did not forsee
The strawberry example reminds me of the Instant Mashed Potatoes non-book review [0].> Since World War II and the large-scale industrialization it fully unleashed, a core method driving ‘progress’ across many different fields of human endeavor has been to shred something real and reconstitute it
As someone under 40 who never had any social media, I cannot overstate the negative impact it's had on my peers and their behaviours. Worst thing to ever happen to society imo, I feel for the younger ones who grew up with it.
I've found that this engineered optimization has a more pernicious side effect: killing curiosity.Lack of complexity stunts the desire to become curious - to give reasons to look closer, ask questions, compare experiences - and ultimately develop 'taste'.When everything is optimized i
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
I love this term - I think it beautifully describes the direction that at least, YouTube is heading towards. Take for example, this racket where a channel copies popular (non-kids) creators’ parody work, splits the screen in half with the content on left, adds a completely random DIY type video on t