If you've installed a website blocker, felt great for a week, then quietly uninstalled it, you're not weak-willed. You ran into a real design flaw. Blockers make a permanent yes/no decision about each site — and the way distraction actually happens doesn't fit that model.
The same site is both the answer and the trap
YouTube is a tutorial that unblocks your work, and a two-hour spiral. Twitter/X is where your industry talks shop, and where an afternoon disappears. Reddit has the exact Stack Overflow-style answer you need, three subreddits away from a meme feed. A blocklist has to pick one verdict for each domain, so you're forced to either over-block (and break your own workflow) or under-block (and leave the rabbit holes open).
Most people end up over-blocking, hit a moment where they genuinely need the blocked site, disable the blocker "just for a second," and never turn it back on. The tool punished legitimate work, so you stopped trusting it.
Hard blocking trains avoidance, not focus
There's a deeper issue. A wall you can't get through doesn't teach you anything about your own attention — it just removes the choice. The moment the wall is gone (weekend, different device, a session that expired) the habit is exactly as strong as before. You've outsourced the decision instead of getting better at making it.
That's also why the strictest blockers feel awful to use. They assume you can't be trusted, so they make themselves hard to disable. The friction works for a small number of severe cases, but for most people it just breeds resentment and workarounds.
What works better: context, not lists
The thing a blocklist can't see is what you're trying to do right now. "Am I allowed on YouTube" is the wrong question. "Is this YouTube tab moving my current task forward" is the right one — and the answer changes hour to hour.
A more durable approach has three properties:
- It goes by the moment, not the domain. The same app or site can be fine now and a problem in ten minutes, and the tool should treat it that way.
- It keeps you in the loop. A nudge — "hey, is this the thing you sat down to do?" — interrupts autopilot without removing your judgment. You correct course because you decided to, which is the muscle you actually want to build.
- It adapts. If it learns what your real work looks like, it stops crying wolf, and you stop ignoring it.
This is the gap between blocking and what you might call a focus companion. A blocker enforces a rule you wrote in the past. A companion reacts to what you're doing in the present.
A simple system you can start today
- Name the task out loud or in writing before you start. "Drift" only has meaning relative to an intention — without a stated task, everything feels vaguely productive.
- Catch the switch, not the site. The dangerous moment isn't the bad website, it's the unthinking tab-switch toward it. Notice the reach.
- Make the correction cheap. The goal is a five-second redirect back to the task, not guilt. Guilt makes you avoid the tool.
- Review patterns weekly, not minute-by-minute. Where does drift actually start for you? That's more useful than any blocklist.
Where AccountyCat fits
AccountyCat was built around exactly this critique. By default it goes by context rather than a list you maintain — it reads what you're actually doing (the active app, the window title, and only when that's ambiguous a screenshot) and nudges you when you've drifted from the task you said you'd do. You can still add rules if you want them, but you don't have to. It nudges first and follows up with a gentle check-in only if you keep drifting — one you can wave off or explain — and when you tell it something was fine, it learns. It runs on-device or through your own private cloud key, is free, and open source. If blockers have failed you for the reasons above, that's the itch it's trying to scratch.