I deleted Twitter. Then I doomscrolled Reddit.
I deleted Reddit. Then I doomscrolled YouTube Shorts.
I deleted YouTube. Then I sat there refreshing Hacker News.
Sound familiar? If you're a developer who's tried "digital detox" and failed, this post is for you. I finally figured out why every attempt failed — and what actually worked.
The Problem Isn't the App. It's the Feed.
Every productivity article says the same thing: "delete social media." And every time I tried, I just migrated to the next dopamine source. The apps are interchangeable. The algorithmic feed is the drug.
Here's what I mean:
- Twitter/X → The "For You" tab is algorithmically curated to maximize engagement
- Reddit → The front page and popular feeds are optimized for scroll time
- YouTube → Shorts and recommendations are designed to keep you watching
- Instagram → The Explore page and Reels are infinite scroll traps
- Hacker News → Even this has a ranking algorithm (albeit simpler)
The feed is what creates the compulsion loop. Not the app itself.
Why Website Blockers Don't Work for Developers
I tried every blocker: Cold Turkey, Freedom, SelfControl, hosts file edits. Here's why they all failed for me:
1. You Need These Sites for Work
I can't fully block Twitter — I follow framework maintainers who post breaking changes there first. I can't block Reddit — half my debugging solutions come from subreddit threads. I can't block YouTube — conference talks and tutorials live there.
Website blockers force you to choose between productivity and access. That's a false choice.
2. You Just Work Around Them
As developers, we're literally trained to solve problems. A hosts file block? I'll edit it back. A browser extension? I'll use a different browser. A DNS block? I'll switch to mobile data.
Every blocker is just a puzzle, and we're professional puzzle solvers.
3. The All-or-Nothing Approach Creates Binge Cycles
Block everything for 8 hours, then "reward" yourself with unlimited access? Congrats, you just created a binge cycle. I'd be MORE compulsive in the evening because I'd been restricted all day.
What Actually Worked: Feed-Level Blocking
Instead of blocking sites, I started blocking feeds. The algorithmic, infinite-scroll parts specifically:
- ✅ Can still access Twitter to check a specific person's profile or search for error messages
- ❌ Can't see the "For You" or algorithmic timeline
- ✅ Can still go to a specific YouTube video linked in docs
- ❌ Can't see Shorts, recommendations, or the homepage feed
- ✅ Can still visit a specific Reddit thread from a Google search
- ❌ Can't see the front page, Popular, or browse subreddits aimlessly
The difference is surgical. You keep the utility, remove the addiction vector.
The Results (7-Day Experiment)
I tracked my screen time before and after switching from website blockers to feed-level blocking:
Before (website blockers):
- Average productive hours: ~4.5/day
- Average "distraction" hours: ~3.2/day
- Number of times I disabled the blocker: 2-3x/day
- End-of-day feeling: guilty, frustrated
After (feed-level blocking):
- Average productive hours: ~6.8/day
- Average "distraction" hours: ~1.1/day
- Number of times I tried to work around it: 0 (no reason to — I could still access what I needed)
- End-of-day feeling: actually satisfied
The key insight: I wasn't addicted to social media. I was addicted to algorithmic curation. Once the algorithm was gone, the compulsion disappeared.
How This Connects to AI Development Costs
Here's a surprising connection I discovered: my unfocused days were also my most expensive AI coding days.
I've been tracking my AI token costs per request while building two Mac apps, and I noticed a clear pattern:
- Focused days (feed-blocking on): Average 2-3 AI requests per task, clear prompts, low token cost
- Distracted days (feed-blocking off): Average 5-7 AI requests per task, vague prompts, 3-4x token cost
When you're constantly context-switching between code and feeds, your prompts get lazier. You ask Claude or GPT-5.4 to "just fix this" instead of thinking through what you actually need. Each retry burns tokens. Each unfocused prompt wastes money.
Cutting distraction didn't just save time — it literally cut my AI coding bill by ~35%.
Tools I Actually Use
After trying 15+ different approaches, here's what I settled on:
Monk Mode ($15, Mac) — This is the feed-level blocker I ended up building because nothing else did exactly this. It blocks the algorithmic feed components of sites while leaving the useful parts accessible. One-time purchase, runs natively on macOS.
TokenBar ($5, Mac) — Real-time token cost tracker in my menu bar. Sounds unrelated, but seeing "$0.47 for that lazy prompt" in real-time is its own form of focus training. When you see the cost of distraction in dollars, you pay attention.
Focus modes on macOS — Free, built-in. I use these for notification management alongside feed blocking.
A physical notebook — For capturing thoughts that would normally send me to Google/Reddit. I write down the question, keep coding, and look it up during a dedicated research break.
The Counterintuitive Lesson
The productivity advice I wish someone had given me years ago:
Don't block the tools. Block the algorithms.
Social media platforms are genuinely useful for developers. The problem is they're wrapped in engagement-maximizing interfaces designed by teams of PhDs whose entire job is to keep you scrolling.
Strip the algorithm, keep the utility. You get the best of both worlds.
Try It This Week
- Pick your biggest distraction site
- Instead of blocking it entirely, find a way to block just the feed/recommendation layer
- Track your productive hours for 3 days
- Compare to your baseline
I'd bet money your productive hours go up without the withdrawal symptoms that full blocking causes.
What's your relationship with website blockers? Have they actually worked for you long-term, or do you work around them like I did? Drop your experience in the comments — genuinely curious if I'm the only one who went through the delete-migrate-repeat cycle.
Building in public: @_brian_johnson
Top comments (0)