Slop Theater and the 37,000 Lines
37,000 lines of code per day. 380,000 lines for an issue tracker. Sentry's entire Python backend is 550,000 lines. When did lines-of-code become a success metric instead of a source of shame?
37,000 lines of code per day. 380,000 lines for an issue tracker. Sentry's entire Python backend is 550,000 lines. When did lines-of-code become a success metric instead of a source of shame?
In a bar, the bartender is legally obligated to stop serving you. Your coding agent will pour another round at 2 AM and compliment your architecture.
Armen Ronacher spent an hour arguing with a coding agent about a problem he'd already solved. The agent contradicted itself with identical confidence. He didn't notice for 60 minutes.
METR study: developers feel 20% faster but measure 19% slower. That 39-point gap between perception and reality is the most dangerous metric in your workflow.
47% of developers rank it #1. AI PRs have 1.7x more issues, 2.74x more XSS vulnerabilities. Here's a 5-layer review framework that goes beyond the checklist.
They look the same from outside. But BCG data shows brain fry sufferers actually have less burnout. Here's how to tell which one you have - and why it matters for recovery.
AI automated the exact tasks that historically trained juniors into seniors. Nobody designed a replacement. Here's the data, the risk, and the 'mental gym' approach.
Rick Rubin calls vibe coding 'punk rock.' But his own creative philosophy — subtraction, silence, listening — contradicts compulsive prompting.
Nearly half of workers hide how much they use AI. Addiction research calls this concealment. Developers call it Tuesday.
Your team ships faster with AI. Commits are up. Velocity looks great. Nobody's tracking the 41% bug increase, the review burden, or the burnout underneath.
AI tools don't get tired, don't close at midnight, don't suggest you stop. The human is the only thing standing between a useful session and a 4 AM spiral.
An AI infrastructure engineer shipped more code than ever — and burned out harder than ever. His experience, plus 30 developers on Hacker News, map the pattern.
Pro-AI developers and anti-AI activists independently describe the same compulsive patterns. A Reddit thread maps real user experiences onto research-informed addiction mechanics.
Boston Consulting Group researchers have a name for the mental fog from managing AI agents: 'AI brain fry.' Their findings map directly onto the OnTilt framework.
Your AI coding tool parallels the same 6 psychological mechanics as a casino slot machine. The behavioral patterns are analogous — here's what peer-reviewed research suggests.
Tab-complete parallels the behavioral pattern of a slot machine pull. The unpredictability of AI suggestions suggests an analogous mechanism.
I spent 6 hours refactoring a service that worked fine. The AI kept suggesting improvements. Each one made sense. None of them mattered.
AI tools create something that looks like flow but lacks its defining feature: agency. Understanding the difference between real flow and 'dark flow.'
You brush your teeth. You sleep 8 hours. What do you do about your AI habits? 12 research-informed practices for healthy AI-assisted work.