The MVP is not finished yet so the minimalist landing page that I made for now is just there to validate the idea with potential users.

When it's done you can log in and use it with your own personalized dashboard.

I am currently making tests with Youtube Data API to fetch the most currently popular videos on Youtube in my local computer and it works. Just to get the hang of the API a little bit then I will start the real developing :)

Thanks!

I can not wait when you will finish the project because in my mind you have describen exactly all the bad parts, I have nothing else to desire. Youtube's search is the poorest search I have ever seen despite of really cool ML which can range videos by huge amount of parameters. From my experience, currently the only way to get really interesting videos (for me this is a science) is not to watch anything except science for getting falled into the science content loop, and watch more of [1] maybe without really watching just for the sake of recommendations.

[1] https://news.ycombinator.com/from?site=youtube.com

Thank you for sharing your thoughts with me! Well the engineering team at Youtube is working hard but still I cannot understand how the algo can still miss huge parts. I spoke with so many people on Reddit and there are so many different problems they encounter.

- They are used to watch videos about a specific topic and then if they watch 1 video about something else (let's say headphones for example) the algo forgets about their previous preferences and keeps showing recommendations about headphones.

- Some people get recommended videos they already watched before

- Some people don't really like Shorts so they would prefer if there was a way to toggle off so they won't see them anymore

And there are also so many different issues.