91–100 of 172 posts
I think I'll start to use exclusively CLI tools for discovering and downloading of YT content. The entire experience which starts from typing "youtube.com" in the address bar and pressing enter is obnoxiously unbearable.
I love that a third party is stepping up here, but it's incomprehensible to me that Google doesn't do this themselves. They're a search company, and they own YouTube. The YouTube data — including the subtitle files — is already sitting there on their servers; they don't have to scrape it, they just have to index it. What are they even doing? Fun thing to try: do a Google search with "site:youtube.com" in it. You get…
I would posit that google has determined that it’s more profitable to keep viewers on YouTube via controlling the viewing experience vs whatever additional ad revenue they’d gain by making videos more easily indexable. I’m basing this hypothesis on YouTube’s consistent trend of removing features related to controlling your own viewing experience. For example, removing subscription collections.
This is exactly it. It's the same business logic that is employed in the regular Google Search.
Google doesn't make money from giving you correct search results. It makes money by keeping you searching for the results you want.
Earlier quoted context omitted.
Here's a (kinda) ELI5: you would use a language model to create "embeddings" of the text, which you can think of as a set of numbers representing the "meaning" of a set of characters. These numbers can be plotted as points in a space, and embeddings of things with similar meanings are plotted close to each other. So things like "exam preparation" would have embeddings close to things like "top study tips". Say you ha…
Do you have any resources that might guide one on doing something like this from scratch?
Very nice! FYI: sqlite ships with a full text search engine featuring a Boolean query language, highlight(), snippet() and scoring: https://www.sqlite.org/fts5.html I’ve not used it with enough content to know how much faster it is than LIKE ‘%my query%’ but it should be a lot quicker. (Also, in most cases you don’t need to create an id column — every table has one already in the form of rowid .)
In what cases is it unwise to rely on rowid over a id field?
I love that a third party is stepping up here, but it's incomprehensible to me that Google doesn't do this themselves. They're a search company, and they own YouTube. The YouTube data — including the subtitle files — is already sitting there on their servers; they don't have to scrape it, they just have to index it. What are they even doing? Fun thing to try: do a Google search with "site:youtube.com" in it. You get…
Social media companies are in a constant tug-of-war against the end users when it comes to controlling what the users see. The ideal is that the user has absolutely no control over on what they see and the social media company can fully dictate content. That is what makes the money. Allowing users to freely query content in their own websites is completely antithetical to what they are trying to do. YouTube is also v…
I love that a third party is stepping up here, but it's incomprehensible to me that Google doesn't do this themselves. They're a search company, and they own YouTube. The YouTube data — including the subtitle files — is already sitting there on their servers; they don't have to scrape it, they just have to index it. What are they even doing? Fun thing to try: do a Google search with "site:youtube.com" in it. You get…
So sure, google can say ‘here’s a standard way to provide subtitles for a video which we’ll index’, but then that becomes a complete SEO side channel - google needs to validate that the subtitles actually match the content. And that means their bots downloading the video itself. And google really doesn’t want to go out there and argue that video needs to be downloadable by bots, because that’s the whole YouTube-dl case right there.
> yt-fts is a simple python script that uses yt-dlp to scrape all of a youtube channels subtitles and load them into an sqlite database that is searchable from the command line. Critically, this is per channel . I wonder if we can optionally configure this to share the downloaded transcripts to a central repository so eventually a good proportion of youtube's transcripts could be downloaded as one big text file.
I love that a third party is stepping up here, but it's incomprehensible to me that Google doesn't do this themselves. They're a search company, and they own YouTube. The YouTube data — including the subtitle files — is already sitting there on their servers; they don't have to scrape it, they just have to index it. What are they even doing? Fun thing to try: do a Google search with "site:youtube.com" in it. You get…
Spotify and YouTube are the leading examples but there are definitely others.
Earlier quoted context omitted.
Actually, letting people search video text would enable less watching and that is probably the reason they aren't interested in it.
Exactly! Google is in the attention business, the more you stay the more they make
Earlier quoted context omitted.
Rumble will perish due to its controversial niche, which mind-numbingly is also its main moat
I kind of agree about the most and niche, but think you’re wrong about the conclusion. That niche has ~50% of Americans and a large portion of the globe fine with it. Put simply, as YouTube pushes out niches (guns, comedy, pundits, etc). They move to rumble. This increases the network effects on rumble and will help it grow rapidly. The more niches on rumble, the less of a need to go to YouTube. YouTube effectively k…
Someone already mentioned Odyssey and I too am more hopeful of that platform - a decentralized platform is the best positioned to avoid political interference and censorship.