Why I Built Gazenest

There was a specific evening that made the problem impossible to ignore. I had opened YouTube to watch one video, a conference talk I had bookmarked during the week. I found it, watched it, and then kept watching. An hour and twenty minutes later I closed the tab. I could not tell you what I had watched after the first video. The time was real. The choices were not.

I had experienced this before, but that evening I was also tracking my time for a work project. The session appeared in the log with a duration I had to check twice. I had accounted for those ninety minutes as "a short break." The gap between that mental accounting and the actual record was what moved me from vague discomfort to wanting to understand the mechanism.

What I tried first

The obvious first step was a blocker. I set up restrictions during certain hours, then extended them, then began resetting the timer when something came up that felt genuinely necessary. Every workaround I granted myself was reasonable in the moment. Over three weeks the block had so many exceptions it barely changed anything.

The problem with a blocker is that it treats all YouTube sessions as equally bad and addresses none of them. It cannot tell the difference between two hours of carefully chosen tutorials and ninety minutes of autoplay drift. It fires the same restriction for both. That bluntness made it easy to rationalize around.

I also tried screen time reporting. Most tools will tell you that you spent 2 hours 47 minutes on YouTube this week. They will not tell you how much of that was a session you opened on purpose and how much was a tab that appeared in your hand while you were doing something else. The number alone is not enough to act on.

The question that changed the approach

At some point I stopped asking "how do I watch less YouTube" and started asking something different: how often am I actually choosing what I watch, versus how often is something else choosing for me?

That reframe matters. Two hours of watching can be a good session or a bad one. The variable is not duration; it is whether you were steering. A session that started with a declared intention and followed through on it is a different event from a session that began with a reflexive tab-open and drifted through algorithm-selected content for ninety minutes.

I built a small script that logged every video I watched, the channel, the duration, and whether I had opened YouTube by typing a URL or navigated there from a recommendation. After a few weeks I had enough data to see the pattern clearly. About 60 percent of my sessions had started without any declared intention. Most of the drift happened in the late afternoon and in the first thirty minutes after waking up. The channels I wanted to be spending time on were crowded out by ones I had never sought out.

That data changed my behavior more than any blocker had. Not because of willpower, but because the gap between "I watch a reasonable amount" and "you spend most of your YouTube time not steering" is visible and specific in a way that a restriction is not.

What I actually built

The script became an extension. The extension became a dashboard. The session logging became Intent Mode: a prompt that appears before you start watching, asking what you are there to watch. That declared intention becomes the baseline against which the session is scored.

Three scores emerged from the data as the most useful: Self-Control, which reflects the ratio of intentional sessions to drift sessions; Clarity, which shows how closely you followed through on what you declared; and Diversity, which flags when your content range has narrowed to the point where the algorithm is effectively programming your recommendations.

None of this blocks anything. Gazenest removes Shorts from the default path and disables autoplay, because those two mechanisms are responsible for most unintended viewing time. But access stays open. You can still drift. The point is that when you do, you can see it.

What has actually changed

I still open YouTube more than I probably should. That is the honest version. What changed is that most sessions now begin with some version of a stated purpose, because the extension asks. The Clarity score for the week is either coherent or it is not, and if it is not, I can see which sessions were the problem and what I was doing that day.

The tool that would have helped me on that evening exists now. Not to stop the session, but to make visible the gap between the session I thought I was having and the one I actually had.