What a Slow Browser Taught Me

In January I switched my primary browser to one I had specifically chosen because it was, by design, slower. Not broken. Not unusable. Just slower — it loaded pages with slightly more latency, rendered JavaScript more conservatively, and blocked by default most of the tracking infrastructure that modern browsers tolerate in the name of compatibility. The difference in raw performance was something like a second or two on a typical page load. It should not have mattered. It mattered considerably.

I want to be careful about what I'm claiming here. This is not an argument that slow browsers are better or that friction is inherently valuable. It is an account of what a specific experiment revealed about my own behaviour, and a tentative set of conclusions I have drawn from it. Your situation is almost certainly different from mine. The patterns I discovered may not be yours.

The experiment

The browser I used was not, in itself, the point. Any tool with a similar friction profile would have produced similar results. What mattered was introducing a consistent, small delay into a behaviour so habitual that I had stopped noticing it as a behaviour at all.

Within three days, I had noticed something I found genuinely surprising: a significant proportion of my browsing was not goal-directed. I already knew this in an abstract sense — everyone who has thought about internet use knows that browsers enable a kind of ambient wandering that can persist for hours without producing anything coherent. What I had not understood was the precise texture of my own version of this. I was not, for the most part, following interesting trails or reading things I had planned to read. I was opening tabs as a response to micro-moments of uncertainty. When I didn't know the next sentence, I opened a tab. When a line of thinking became difficult, I opened a tab. When I needed a moment to decide what to do next, I opened a tab.

The slow browser did not stop this. It just made the act visible. A two-second delay is long enough to create a small gap between impulse and action. In that gap, I could see the impulse for what it was.

What friction does

There is a useful distinction between friction that impedes and friction that reveals. Friction that impedes makes a task genuinely harder to complete — it slows the work itself, introduces error, creates waste. Friction that reveals does not change the work; it changes your relationship to it. It surfaces information that was previously hidden, usually because the relevant behaviour was happening below the threshold of conscious notice.

The slow browser was, in my case, friction that revealed. The work of loading a web page did not become harder. What became visible was how often I was attempting to load web pages for reasons that had nothing to do with needing information from a web page. This is useful information. It is the kind of information that is essentially impossible to gather without introducing exactly this kind of interruption to an otherwise seamless process.

Speed, in digital tools, tends to eliminate this category of information entirely. Speed is optimised for task completion, which means it is optimised for the assumption that every use of the tool is a legitimate task. When every action is instantaneous, the distinction between purposeful action and restless habit disappears. They are phenomenologically identical. You cannot tell, in retrospect, whether you opened a tab because you needed something or because your hands needed somewhere to go.

Habit versus preference

The more interesting finding was not the volume of aimless browsing but what it said about my relationship to the browser itself. I had believed, before this experiment, that I preferred fast browsing — that speed was a genuine value I held about this tool. What the experiment surfaced was that this preference was mostly rationalisation. What I actually preferred was the sensation of momentum, and I had been using browser speed to manufacture that sensation regardless of whether any real forward motion was occurring.

This is a common feature of digital tool preferences: they are often preferences for the feeling a tool produces rather than for the work it enables. Fast response times create a sensation of competence and control. Extensive tab organisation creates a sensation of command over complexity. Elaborate note-taking systems create a sensation of comprehensiveness. None of these sensations are the same as doing good work. They are emotionally related to doing good work, which is exactly why they are effective substitutes for it.

I am not claiming that browser speed or tab management or note-taking systems are inherently problematic. I am claiming that the preferences we hold about our tools are frequently preferences for their emotional products rather than their functional ones, and that this distinction is worth examining. It is possible to build a digital environment that produces an excellent sensation of working without requiring very much actual work from you. Many people have done this. It is deeply satisfying in a way that becomes difficult to explain to anyone who has noticed what it costs.

What I kept

I ran the experiment for five weeks. At the end of it, I returned to a faster browser but kept two things from the experience. The first was a practice I have continued: before opening a new tab, I name what I am looking for. This takes about one second. It has not eliminated aimless browsing — nothing eliminates aimless browsing — but it has made the choice visible enough that I make it less often by default.

The second thing I kept was a revised understanding of what I am actually doing when I use a browser. I had categorised it, loosely, as a research and communication tool. It is more accurately a tool for managing my own uncertainty about what I should be doing at a given moment. That is a legitimate use. It is worth knowing that it is what you are using the tool for, because it opens the question of whether there might be better responses to that uncertainty than an open tab.