Hacker News new | past | comments | ask | show | jobs | submit login
Top 70000 educational YouTube channels in 20 languages by category (limnology.co)
211 points by askytb on Nov 4, 2022 | hide | past | favorite | 73 comments



This is alright, I like the category breakdown, it makes browsing great, but its still subject to what _youtube_ considers good content. Which just results in the top content being short-form videos which provide quick-facts, not something I can actually learn with. Not necessarily a bad thing for different categories or if that is what you're seeking in the category but a consequential factor for someone seeking something outside the advertiser-friendly, profit maximizing algorithms.

I just love things the variety of things that fall outside the google-ads algos its so easy to silo our "content" from platforms like this. I browse marginalia [0] just to spice it up and see what else there is out there.

[0] https://search.marginalia.nu/


>Which just results in the top content being short-form videos which provide quick-facts, not something I can actually learn with.

Youtube also has tons of long-form 1+ hours content. There's a recent machine learning video that's 25+ hours long and it's not hidden away in obscurity. It's prominently listed on the 1st page of search results for "deep learning tutorial" : https://www.youtube.com/results?search_query=deep+learning+t...

As another example, I got a friend exposed to sewing channels on Youtube. Some content creators create hour-long "sew alongs" which she watches every week now to learn new techniques. It fills in a gap left behind by Public Television since they don't air sewing shows anymore.

Lots of long-form videos in car repairs, woodworking, etc.

EDIT ADD: the 25-hour deep learning video example is this one: https://www.youtube.com/watch?v=Z_ikDlimN6A

... which was also copied to freeCodeCamp.org channel (the 12th recently uploaded video) :

https://www.youtube.com/c/Freecodecamp/videos


>Youtube also has tons of long-form 1+ hours content. There's a recent machine learning video that's 25+ hours long and it's not hidden away in obscurity. It's prominently listed on the 1st page of search results for "deep learning tutorial" : https://www.youtube.com/results?search_query=deep+learning+t...

I would be careful with statements like this. I believe that the Youtube search results are tuned by your personal viewing history, and of course just on a general basis the results are subject to change. I don't see the video you are talking about when I do that search, but then I don't know what you mean by 'on the 1st page' since it is an infinite scroll.


Try digging deeper into the "Related Keywords" cloud at the top, the narrower is the keyword you pick, the more long-form & technical and the less pop-channell-y stuff you'll see.


I think this list shows that how overwhelming the amount of resources to learn but we have so limited time. I think Duolingo style learning paths are way to go because we need to minimize the time to learn something. These kind of dumps are of no use except for bookmarking and feeling good.


I had the opposite reaction. I think this list shows what is passable as educational nowadays. I would personally be very cautious to consider anything included as edutainment, never mind educational.

I took some time to peruse the list after topics for which I took formal education and could not find a single useful channel.

The only very few exceptions to the rule are MIT OCW kind of channels where universities include playlists with footage of undergraduate courses.


Can you get a job watching 1000 hours of YouTube?


Good point. Learning mostly occurs during the active parts where no YouTube video can provide. I think that's why universities will live a lot longer than people think.


This is exactly how I wish YouTube allowed me to browse for videos! Great resource. Does anyone on HN have additional resources that allow for content discovery?



I'm creating an encyclopedia based on Wikipedia, Wikidata, YouTube and many other sources of information. For each of the millions of topics in a language-wiki, the software renders a "topic card" with thematic sections, each containing links to other content. Link: https://conze.pt


This is actually pretty awesome! I think the ranking algo is doing a pretty good job. I checked a few categories I follow closely, and the top results match pretty well with how I would personally rank them. And now I'm off to explore other topics. Thanks!


Idea is great but there is a lot of overlap between the tags, e.g. there are six pages of tags for Chinese but half of them are all variants on the same tech topics (e.g. there is a separate tag for 苹果、手机、电脑、and "apple watch").


Can you also add greek? I'm always on the look for good content for my children but can't find much in my language :(


Unfortunately all the other languages have too few results, I just checked and Greek only has 0.7% as many channels as English in the db, so it'd only be a few pages of content


Yes it seems that too few people understand greek to make it worthwhile to produce quality content in my language :(


I second that "astronio" is pretty good.


Yes I've seen some episodes but I'm looking for something for younger children...


Don't hesitate to share if you stumble upon something. Kid is 1½ yo so could be useful too.


Amazing headline. It reminds me about a joke from a Futurama episode, from the late 90s, while Fry is channel surfing intergalactic cable, "Sheesh... 40,000 channels and only 150 of them have anything good on."


If 10,000 people say this but there is little overlap in the channel preferences, suddenly 40,000 channels seems a reasonable number. Point being that there are a wide ranging number of interests, just because it isn't in your niche, doesn't mean it shouldn't be there at all.


This is pretty awesome. One small feedback, I noticed how there are multiple tags indicating same kind of content. for ex - "Mathematics" and "math". Combining these would be useful.


TKT Talk under English > Music here, https://limnology.co/en/languages/en/keywords/music, is not in English.

Not sure how to report an incorrect listing on your site, would be good to better surface how to do that.


I'll add a report button, it definitely needs it. All of that data is from the the youtube channels themselves and _a lot_ of them mislabel things, even such basic ones as their channel's language.


Thanks!


I find it hilarious that "Indian English" is a separate category


I had to split it because the sheer size of Indian audiences breaks all the metrics, and yet those videos would not be super relevant in the west as a very large percentage of the Indian educational youtube consists of various kinds of local exam/test preparation channels


Why?

IIRC Indian English has almost 250m speakers, which is in the same ballpark as what we'd call American English and has its own, distinct features as much as AmE vs. BrE.

Edit: added context


seems like you are both right


Looks cool! curious how you sourced and tagged the channels?

We have a few popular anatomy channels in different languages[0], but I couldn’t find any of our videos when searching for “anatomy” and also results were quite mixed. Some videos were quite unrelated or very loosely related at best.

[0] main one in English https://www.youtube.com/channel/UCHn_K1zOBYZqtmIYkXLEIQw


It's in the database: https://limnology.co/en/creators/UCHn_K1zOBYZqtmIYkXLEIQw

And on the second page for the anatomy keyword https://limnology.co/en/languages/en/keywords/anatomy?page=2

I think it got somewhat downranked because of the recent views/subscribers mismatch


Thank you!

> I think it got somewhat downranked because of the recent views/subscribers mismatch

Can you explain what you mean? also what about higher ranked results that don’t match the topic that closely? are you relying on YouTube’s ranking for keywords or do your own?


Ok now create a vector space of similarity between videos in each category and extract the key points from each video into a summary of each category.


This is a great resource. One thing to watch out for if you're using this to practice listening skills in a foreign language, is that not all regions seem to be tagged correctly. I clicked on Spanish, then the Mexico tag, and the top video was of someone with a Spanish accent rather than a Mexican or Latin American one.


This is good: https://limnology.co/es/creators/UCjtUS7-SZTi6pXjUbzGHQCg

It would be cool to be able to find similar channels in other languages, but I can see how that would be challenging.


Yeah, that is something I'd love to implement but so far could not figure it out, even simply interlinking the keywords across 20 different languages is very non-trivial


AI has already digitized these videos a countless number of times why are we slow to implement this as an algorithm into the genetic code of a fish which we may then stick into our ear to perform any spoken translation to us.

We will probably have similar for eyeballs one day but they may be more invasive.


This isn't eligible for Show HN - please see https://news.ycombinator.com/showhn.html - so I've taken that out of the title now.


Because it's a "list"? I don't think that's correct, there's ballpark 200k different pages with unique content in that app, that's a lot more than just a list.


It's not really an interactive product that people can 'try out' or 'play with'. It's an information repository. Nothing wrong with that, but it's not really in the Show HN spirit. I admit it's a borderline call.


The Chinese version distinguishes simplified from traditional characters. It shouldn’t do that.

The German version is a little meh. Top categories are something like “howto”. It seems like you used a very simple word-based algorithm to create the categories?


This is missing something critical: It shows tags, but does not allow to use them like tags. Instead it only allows to use them as categories. One should be able to narrow down results by clicking multiple tags.


I did think about it, but my concern was that because I only take a small subset of top keywords for each channel, it's very easy to create a combination of tags that'll show no results. I'll play around with it and see if it makes sense to implement.


Great resource, found a lot of interesting videos already.

One suggestion is to include video length in the individual video display, as most of the very short videos aren't that useful.


I am outraged that the "English" option has the USA flag!

It should be the St. George's cross, or at the very least (and most practically) the Union Jack.


It makes sense if you go by size of English-speaking population.

…but then it might be replaced by an Indian or Nigerian flag in a few decades :)


Look at me. I'm the captain now.


Under English, you have both "diy" and "do it yourself", as well as "english" and "learn english".


Thanks so much! Appreciate you doing this.


This is great, two suggestions:

- search box on every sub page

- allow for reverse search of channels I know (so I can check out similar channels)


Thanks! Search box has been requested multiple times, so will implement, and reverse search is already possible here: https://limnology.co/en/channels


Ah, no idea how I missed that, thanks!


This is great and very useful to me!

What source are you getting the channels from?

And how do you choose which channel goes in?


As a YouTube addict, this is one of the greatest things I've ever seen. Thank you!


How difficult is it to collect user feedback to improve the categorization?


Awesome - wish it had Dutch!


Many of the channels are stale, latest video from one or more years ago.


I don’t think it’s a problem unless it’s about some cutting edge technology.


What is the point of 70000 educational channels for an individual?


You can pick the one you like in your languay


Looks cool, how is it curated? Manually or with an algorithm?


Great idea. Bookmarked. Thanks for building and sharing it.


YT is absolute garbage... we should call out the ACTUAL people responsible for just how terrible their UX is.

If anyone is hiring and someone says they did UX at YT - do not hire them.


>anyone is hiring and someone says they did UX at YT - do not hire them.

Unless you and your business care more about monies than what people say on social media, and your company is suddenly making more money then... I guess hire them?


Thank you for posting this. Moreover Google is a spyware company that holds users hostages for money. I wish Twitch can replace Youtube someday.


Yeah, I wasnt calling out YT content - specifically calling out the platform team internally...


Now I'm busy. Thanks for sharing!


Site seems down for me, and isup.me.


this is very useful, the first video I hit was a very educational one with no fluff.


This is great! Thanks!


Thanks


Where are all the Rust videos? Python on top: https://limnology.co/en/languages/en/keywords/programming


Very cool. Would it be possible to separate South American Spanish from Spanish... Spanish?




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: