This is kind of like what happened with Nitter. X changed their API so anonymous connections were no longer allowed. The ability to get tweets via Nitter never went away, it's just that how you do it had to change. Now you can still run Nitter, but you have to rotate between several burner accounts, use IP proxies, and be really careful about not making requests too quickly or in a pattern that looks like it's automated.

The changes did drive away most of the Nitter userbase, which X probably appreciated, but there are still highly specialized people running Nitter instances and similar tools. Some are even monetizing their work by providing scraping APIs-- pay them a fraction of what X charges for their API service, and get tweets using their fine-tuned scraping platforms.

We're not there yet with yt-dlp, but it does look like we might end up there. But here's the thing, when one door closes, another one opens. As long as there's a way to watch the video in a web browser, a hacker out there will find a way to download it without permission. Maybe it's a browser extension which extracts the necessary cookie from youtube and forwards it to yt-dlp. Maybe it's an AI agent in a headless browser that yt-dlp itself spawns and solves whatever JS shenanigans it needs. I'm very bullish here-- persistent data hoarders always win.