I had 863 subscribers. I'm walking away from all of them. For the last two years, I ran a YouTube channel under @j14wei. It started as a tech channel. Then I went traveling. Then I kept vlogging. Then one day I looked at the channel and realized the algorithm had completely forgotten what I was there to do. The tech videos were dead. The travel videos were doing fine. And anything new I uploaded got pushed to people who came for Bangkok food tours, not AI agents. So I had two choices. Slowly rewrite the channel by uploading ten or more tech videos in a row, hoping the algorithm re-classifies me. Or start a new channel from zero. I'm starting from zero.

The algorithm is a memory

YouTube's recommendation system is not watching your videos. It is watching who watches your videos. Every upload trains a profile of "people who like this channel." Once that profile is set, the algorithm keeps serving your content to that audience. Even if you pivot. Even if you beg. YouTube's own documentation says the system compares your viewing habits with similar users and uses that to suggest content. It processes over 80 billion signals daily. The primary inputs are watch history, search history, and engagement patterns. That means every travel vlog I uploaded was teaching the algorithm to find more people who like travel vlogs, and to show my next upload to them first. If your audience is travel vloggers and you upload a video about Claude Code, the algorithm will still push it to people who want to see Bangkok street food. They will not click. Your CTR tanks. YouTube decides the video is bad. It buries it. Half of all channels on YouTube have a CTR between 2% and 10%. When you push content to the wrong audience, you are not just getting a low CTR. You are actively training the system to believe your content is not worth recommending. Every bad signal compounds. The video was not bad. The audience was wrong.

Starting fresh is faster than pivoting

I did the math. To pivot @j14wei, I would need to upload at least five to ten tech videos back-to-back before the algorithm starts to reconsider. During that time, every video gets shown to the wrong audience first. My CTR craters. My watch time drops. The channel looks like it is failing, even though I am doing the right thing. A new channel does not have baggage. Upload one tech video. YouTube has no prior audience to push it to, so it tests the video on strangers based purely on the title, thumbnail, and early signals. If the video is good, it finds an audience quickly. If it is not, the algorithm moves on. That is the cleanest possible starting position. No legacy. No wrong audience. No history to unlearn. A fresh channel can outgrow a pivoted one in a month. YouTube even published official guidance on pivoting. Their advice is to prepare several videos in advance, communicate the change to your audience, and be patient because results take months, not weeks. That is good advice for a channel with 100,000 subscribers. For a channel with 863 subscribers and a mismatched audience, months of patience is just months of bleeding.

The 863 is not real

Here is the harder truth. The 863 subscribers on @j14wei do not actually belong to me. Most of them subscribed during a specific moment of the channel, for a specific kind of content. A tech video goes out to them, and they do not click, because that is not what they signed up for. A subscriber who does not click is worse than no subscriber. It tells the algorithm "the people who know this creator best do not care about this video." It is an active signal against you. The only subscribers that matter are the ones who will click on your next video. Everyone else is a number. I would rather have 50 real subscribers than 863 ghosts.

Niche clarity compounds

The mistake I made with @j14wei was not choosing. I wanted to do tech, and travel, and lifestyle, and food, and whatever I felt like that week. YouTube punishes that. Not because variety is bad, but because the algorithm is optimizing for one thing: does the next video land with the audience it built you? The more focused you are, the more aggressively the algorithm pushes you. Fireship is never going to post a travel vlog. Theo is never going to post a cooking video. Matt Wolfe is never going to post a fishing tutorial. That is why their channels compound. I did not understand this until now. The old channel had to die for me to understand it. There is a reason YouTube's 2025 algorithm changes started boosting small, focused creators. Channels under 500 subscribers now get prime placement if their early signals are strong. The platform is rewarding specificity more than ever. That is a tailwind you only catch if your channel is clean.

What I am actually starting

The new channel is about AI, building, and what I am figuring out as a solo founder. Ryu is the product I am building. Update Night is the newsletter I write. The blog is the idea factory. The YouTube channel is the place where I talk through the things I already believe. Every video has to pass one test. Would someone searching for AI tools, AI workflows, or AI founder advice click this? If yes, post it. If no, hold it. That is the filter that would have saved me two years of confusion.

The last thing

There is something uncomfortable about archiving content you spent years making. The tech videos on @j14wei are unlisted now. Not deleted, just hidden. Future me can decide what to do with them. But archiving is not failure. It is a clean slate. It is admitting that what I uploaded two years ago is not the thing I am trying to build today, and that the kindest thing I can do for both versions of myself is to separate them. The old channel gets to be what it is. A travel vlog. A nice diary. A Singaporean in Asia eating food. The new channel gets to be what I actually want. Zero is a fine place to start.

References