Long ago, I created a test YouTube account to experiment with the recommendation algorithm. I’ve already put it through some wild scenarios, but now I’m trying something new: I’m subscribing to every channel that asks me to. No exceptions, even if the channel sucks.
Initial setup
I unsubscribed from all existing channels for a clean slate, but kept my watch history to avoid a completely useless starting point. Previous experiments have shown that if you erase your history, the feed resets to what the algorithm loves, not what you actually find worth watching. Just because cancer and heart disease are common doesn’t mean I should join those clubs. Same with trending channels and popular videos.
The method
- I only watch videos from recommendations or search results.
- If a creator says, “Subscribe!”, I hit that button, no questions asked. Makes no difference even if the video sucked. Even if the channel vomits a hundred trash videos a month and only one gem a year, I subscribe anyway.
- Showing a subscribe animation counts too.
Open questions
Will my feed turn into a waste dump of low-effort content? If not, does subscribing even matter at all? There’s only one way to find out.
Some background (if you want to know why)
When I started that account, I just wanted to find out how easily I can get the feed to show me useless trash if I click anything that looks interesting. During that experiment, I was still very conservative with my subscriptions, but I clicked videos very liberally. Turns out, YT has gotten better with this. In the past, a few clicks could ruin everything, but that hasn’t been the case in a long time. Nowadays, you can safely watch a few clickbait trash videos and they won’t haunt your feed for the rest of eternity. The feed doesn’t get instantly ruined any more.
Recently, I tried to be very strict by installing all sorts of browser plugins that allow me to control nearly everything. I used Unhook remove nearly all of the elements on the front page, and PocketTube to categorize my subscriptions just the way I like. That was a highly successful approach and I’m still using it with my main YT account.
BTW if you have to resort to trickery like this, it should tell you pretty clearly that there are some serious underlying problems with YT. Yes, I know YT is awful, and I’m already on my way out. However, before I go, I want to mess around some more.
The real question is, do I even need to be that careful with my subscriptions? Do they even matter? On my main account, I only subscribe to channels that reliably produce videos worth watching. The gem to trash ratio has be at least 95% for me to subscribe, but is that excessively strict? That’s why I have this test account where everything can go wrong and it won’t matter.


What you’re subscribed to has little to no impact on the algorithm anymore. It just pushes whatever it “thinks” is most similar to what you have watched over the last few days. It’s pretty unusable.
It’s only good at immediate recommendations. Like, if I see a music video that looks good, I watch it, next look at the home page is just music vids. I instead click on a climbing video, now home page is all climbing. It’s garbage at actually giving anything other than what you’ve just consumed.
Yeah, the effect of subscribing seems to be very subtle. So far, the experiment hasn’t lasted more than a few days, but I think I’m beginning to see a pattern.
The channels I subscribed to do show up a bit more often, even in the search results. I also see plenty of other channels in the main feed. Interestingly, there are a few channels I watch every now and then, but they never asked me to subscribe, so I didn’t. Somehow, those channels still show up in my feed as if I had subscribed, which makes me wonder what the role of subscribing even is these days.
However, this experiment needs to continue much longer. I wonder if it continues like this for several weeks. I’m really curious to see what happens to the channels I subscribed to. Will they all show up in the feed consistently, or will some of them disappear completely. Time will tell.