I imported 6 months of YouTube history. Here's the timeline of how my attention broke.

Before I looked

I had a general sense of my YouTube habits. I knew I watched too much some weeks. I knew there were periods of productive use, research, tutorials, content I'd genuinely sought out, and periods of something else. I had the usual mix of pride and vague guilt that most people carry about their screen time.

I didn't have data. Not real data. Not the kind you can't negotiate with.

Six months ago, I imported my full YouTube watch history into Gazenest. The file is available through Google Takeout, a JSON export of everything YouTube knows you've watched. It goes back years if you've kept your history enabled. I limited mine to six months because I wanted something I could hold in my head, a period I could actually remember living through.

What came back was not a summary. It was a timeline, video by video, with timestamps, with duration estimates, with channel information. It was a precise record of where my attention had been for 180 days.

Looking at it took me longer than I expected.

The shape of six months

The first thing I noticed was that my viewing wasn't evenly distributed. There were weeks where I barely used YouTube, a handful of deliberately chosen videos, clean search intent, short sessions. And there were weeks where I was on it for 2-3 hours a day, following recommendation chains, watching things I hadn't planned to watch and couldn't entirely explain afterward.

The pattern was not random. It correlated, with uncomfortable precision, to things happening in my life.

There was a week in Month 2 where the watch time spiked sharply. I knew, looking at the dates, exactly what had happened that week. A difficult conversation. A project that had gone poorly. I'd used YouTube the way you use a numbing agent, not for anything specific, just for the sensation of content moving.

The algorithm had responded efficiently. By Day 4 of that week, the recommendations had narrowed. I was getting served more of what I'd been clicking. The diversity collapsed. My Clarity Score for that week was the lowest in the six-month window: 28 average, with two sessions below 15.

The inflection points

There were three clear inflection points in my data. I'll describe them honestly.

The first was a new project starting. My watch history shifted visibly, tutorial content, research, domain-specific videos. The session structure changed: shorter, more search-driven, higher Clarity. This is YouTube being useful. I was using it as a tool and it responded like one.

The second inflection point was a stressful period I'd rather not detail specifically. The pattern is recognizable from the outside: watch time up, diversity down, Clarity collapsing, entry points shifting from search and subscriptions toward homepage recommendations. The algorithm found the channels I clicked when I was low and served them aggressively. I let it.

The third inflection point is the one I find most interesting because I hadn't noticed it consciously at the time. About 4 months in, there was a gradual shift, not a dramatic spike, just a slow drift. My average Clarity Score dropped from around 65 to around 48 over three weeks, without any obvious life event driving it. Looking at the videos, I think it was an algorithm shift. A few channels I'd engaged with started getting surfaced more aggressively in recommendations, and I followed them deeper than I'd intended. The drift was so gradual that I hadn't registered it as a change.

That is the one that concerns me most. The dramatic spikes correlate to identifiable life events, and they resolve when the event resolves. The slow drift doesn't have an obvious cause or a natural end. It just continues until something interrupts it.

What the data made visible that memory didn't

I have a reasonably good memory. I can recall significant life events within a few weeks' accuracy. I cannot reconstruct, from memory alone, the texture of my YouTube use across six months. I can tell you about a few videos I remember watching. I cannot tell you that I spent 74 minutes on a specific channel on a specific Tuesday afternoon, or that my session entry points that week were 89% homepage recommendations rather than subscriptions.

Memory edits. It keeps what felt significant and discards what didn't register in the moment. A lot of YouTube watching doesn't register in the moment. It passes through without leaving a trace in episodic memory, which is precisely what makes it easy to underestimate.

The import doesn't edit. Every video with its timestamp is there. The record is complete in a way that introspection isn't.

The most uncomfortable finding

I'd like to tell you the most uncomfortable finding was the stressful week with the collapsed Clarity Score. That would be a clean story: hard period in life, turned to YouTube, algorithm exploited it, now I know.

The more uncomfortable finding was the baseline. Even in my best weeks, weeks with no particular stress, weeks where I felt like I was using YouTube intentionally, my Clarity Score averaged 58. That means even when I thought I was in control, more than a third of my watching was low-intention, driven by recommendations I hadn't deliberately chosen.

I'd thought of my YouTube use as mostly intentional with occasional lapses. The data suggested it was partially intentional with a substantial floor of habitual, low-clarity watching that I'd been miscounting as choice.

That distinction matters. A lapse is recoverable. A baseline is structural. A baseline requires structural change.

What I did with it

I changed two things. I set up the watched-videos filter, which I should have done earlier. And I changed my default YouTube landing page to the subscription feed, removing the homepage from my automatic browsing flow.

My Clarity Score for the following month averaged 71. That's not a controlled experiment. I can't isolate the cause. But the direction is what I would have predicted from understanding what the data showed me.

The import is not a comfortable experience. It is the kind of honest that is hard to stay with. But it gives you something that six months of vague self-awareness couldn't: a precise account of what actually happened, in enough detail to act on.

The history import is available on Gazenest Plus.