Video recommendation platform as a way to compete with YouTube?

Video hosting is actually a lot harder than it seems.

- For starters, you need the server space to host it. Videos can be large -- a 10 minute 4k video can be nearly 5gb in size.

- Then you need to the CPU to transcode it to various formats and bitrates (Youtube keeps more than 30 different formats sometimes)

- Then you need to bandwidth to actually serve all that data. Ideally you have servers geolocated close to the user.

As someone with a slow internet connection, I generally despise self hosted videos. If your video won't load because I don't have a >100mbit connection, I will likely leave.

I like going on Youtube and see what gets recommended to me, sometimes I discover useful/entertaining stuff that way. But most times, the recommendations tend to be terrible:

- repetitive, as if they insisted in the same specific video dozens of times

- not using the full breadth of a given channel I've subscribed to, e.g. just recommending a few videos out of 100s

- recommending stuff that has absolutely nothing to do with my interests, simply because it's going viral or it's been a previous big hit in the mainstream

- similar channels almost never get recommended to me. This would be super useful, since maybe I'm subscribed to 5 channels for a specific topic, but I could be subscribed to 20 as well - I just can't find them naturally

A good recommendation algorithm might be worth paying for, as it could provide endless entertainment.

>Why don't we just make a platform that crawls the internet for videos and build a recommendation algorithm. Therefore no matter where you host your videos you'll be easily discover?

Because there are fundamental tradeoffs of _information_ and _meta_information_ that a system has that factors into recommendations.

People upload about ~500 hours of videos to Youtube every minute.[1]

Yes, that means that centralization to that degree has negatives such as the guitar channel being deleted but it also has positives such as better (not foolproof -- but relatively better) detection of spam and mistitled content. Because Youtube servers actually have the real bytes of actual videos, they can scrub the content with machine learning algorithms augmented with some human oversight.

If you only have a platform that crawls for video urls and doesn't actually ingest the petabytes of video like Youtube, the "recommendations-only" service will be susceptible to gaming such as people putting "Emma Watson and Scarlett Johansson nudes" as the titles but the actual video is an ad for somebody trying to sell a used car. Or the first few days of a video url has "Tips to deal with COVID" which is real but the video hoster later switches it out to be a video for that same url to be something else.

Also, a centralized service like Youtube can measure actual user behavior such as "watch time" (because Youtube's servers know they are sending bytes to the client) to see if the video is actually engaging viewers and this factors into future recommendations. If web surfers are closing out a 10 minute video after 20 seconds, the videos are appropriately penalized because users are "voting" that the content is bad. I.e. the viewers "voted" without even having to press the "thumbs down" button.

A new platform to recommend videos pointing to decentralized video hosting has advantages but also has unavoidable disadvantages that lower the quality of the recommendations.

EDIT to add my previous comments about the financial incentives (ads) that help video content creators which many techies overlook: https://news.ycombinator.com/item?id=21506992

[1] https://www.google.com/search?q=upload+about+~500+hours+of+v...

We definitely need to decouple discovery from hosting.

I'm not even sure you would need to do any crawling. Maybe you could just let people submit links to videos hosted elsewhere. Kind of like you can submit any URL to the Internet Archive. But in this case you would use the frequency of duplicate submissions (along with in-house upvotes) to build out a recommendation platform.

Content creators don't want to leave the platform because they profit via ad-revenue, monthly subscriptions and superchats.

>Content creators don't want to leave the platform because of discoverability

Discovery is still incredibly hard on YouTube. YouTube is a search engine, adding another one won't fix that. I'm also skeptical that anyone could do search better than Google. When creators say "they can't leave because of discovery", they mean "I spent 10 years telling people my address is youtube.com/creator, and if I lose that, most people won't know where to find me."

Regardless, most Content Creators don't leave the platform because YouTube is/(was) the only platform paying people for content. While the type of content that is watched by tech professionals may do incredibly well with Patreon, that isn't true for several other types of content. This IMO, is why Facebook was unable to get people to create for FB Video (Unless you were a corporation that handled monetization already), it's a large reason why Vine creators complained about the platform and it's something Twitch and TikTok get right.

This, to me, is the real issue with starting a YouTube competitor. It's far easier to innovate on the format (Twitch = livestreams, TikTok = clips+ML), then it is to offer "video hosting" but "better". If you want creators to create for your platform, then you have to pay them, which means you either have to get advertisers (and deal with the same content cleanup YouTube does) or have people pay (and deal with that chicken and egg).

I believe Twitch was early enough to not have to go the "pay creators on day 1" route, bu Musical.ly (what eventually became TikTok), Mixxer, and Quibi all had to open their wallets and spend a ton on marketing and exclusives to get people on their platform (and two of them failed hard).

I think the key here is a deep misunderstanding where YTs value to creator is: It's not just recommendations, it's the fact that they take care of ad pipeline which ends up as pure revenue for the creator. Enabling monetization for videos is very simple on YT and you get a paycheck at the end.

Any competing platform needs to compete with this value proposition: you upload video, you get money.

I think this a great idea! YouTube recommendations are extremely lacking and is leaving a ton on the table in terms of a great user experience. I don’t trust that I’m being recommended content that they think is best for me. Rather is it based on if the creator paid Youtube to promote their video?

I like this idea. There are going to be a million reasons why it won't work that people will bring up, but ultimately I suspect they're all addressable.

One concern that would have to be addressed upfront would be to avoid having this turn into another Youtube. For example, once you have a great video aggregation site, there will be a desire to start hosting videos yourself. At that point you can start monetizing them and will inevitably want to promote that content, turning yourself into another Youtube. Basically once you control the demand, the supply side is easy to displace. I guess making it open source could address some of this.

Also, should definitely call it opentube.com (already taken of course, but surprisingly not in use).

Interesting concept. So you propose to have a landing page with a selection of videos tailored to the taste of the visitor? And seamless combining results from YouTube, Vimeo and others?

You would be effectively relegating YouTube to a commoditized CDN for videos.

Again, interesting concept.

I think the topic you may want to explore is "SEO for video" and I'm not sure there's really a definitive guide.

Google has this guide: https://support.google.com/webmasters/answer/156442

Moz has an article from 10 years ago. https://moz.com/blog/video-seo-basics-whiteboard-friday-1108...

My first thought would be that need at least a text-based transcript for crawling, but something akin to audio description would probably be most valuable too.

Peertube is great, but discovery is definitely a problem. Search and recommendations are the way to go.

https://peertube-search.com/

> Why don't we just make a platform that crawls the internet for videos and build a recommendation algorithm.

You are just doing YouTube without the advantage of having to use YouTube's infrastructure. You'll still have a central authority deciding what to recommend and whatnot. The main issue with YouTube is not the fact that they host the content, it's that they also decide what gets shown and to whom.

This is a really great idea.

You can skip the complex mathematics to begin. You'll millionaires before it matters.

Is look at adult sites. Why one site has not taken over. Could be a local maximum. Maybe they get paid off by advertiser's before they get top big.

The hard bit is political. I want my Stormfront videos recommended (by default, like early Youtube) but not a dude shooting himself (gore). But I want to find things if I want.

For best user experience centralized is the way to go. Imagine a video goes viral hosted on a shared server. Not a good experience for anyone. You could build a cache layer on top of it but then how will you reward the creator?

Are you expecting people setting up, and maintaining their own servers to host videos? Those times are gone. Content creators want to upload where millions of users are already present.

You are describing Metacafe from back in the day.

Videos from independent creators won't be able to go viral. Your viewership would be bounded by your network throughput. Even if you're already popular, a disproportionate number of your views will come as soon as you upload a video, so you would be end up paying for a lot of bandwidth you don't need.

YouTube recommendation sucks. It sucks so much. But YouTube has so much content. If you can build a page to link to YouTube videos, you don't even need to host videos. I would use anything but YouTube's homepage, I think it is a good idea.

Not sure if you really want to crawl and make suggestions based on that. It might be better to let people self host videos and essentially provide your service with an RSS feed of videos with some metadata provided.

Remember, convenience trumps all other concerns in capitalism. So your distributed solution has to be as convenient, or more so, to use than YouTube for content creators. You also didn't mention monetization, and again, it needs to be convenient. Allowing "roll your own" monetization is great, but putting a default revenue method is essential.

I think it can be done. The product looks to me like a cross-platform thick-client application that manages your content, both serving it locally, but also putting it on various platforms. I think it could be a real winner if done right! You might even be able to give it away, open source it, if you can sell default space, like Mozilla does.

The obvious starting point for this product would be the content creation tool itself. I'm sure Adobe (and Apple!) would love it if this was a reality, just to stick it to Google. The enemy of my enemy is my friend, after all.

This sounds awfully lot like metacafe.com , or mamma.com from even before - the so-called metasearch engines. You perhaps should try and see if you can monetize it.

Content creators probably don't want to host the videos themselves.

Apart from YouTube, what is there? Dailymotion? Vimeo?

So, like Google Videos search?

openvideodata.org wants to bootstrap open recommendations by scrobbling videos

https://soundclot.com/