Let's improve PeerTube – Help us define PeerTube's future roadmap
joinpeertube.org
joinpeertube.org
btw I think this is what Eugen, et al, should be doing with Mastodon gGmbH.
I live in Los Angeles. I want the city to archive its city council meetings on a video platform. Preferably, these would be available from a subdomain of lacity.gov instead of being available through youtube or another company of that sort (there are all sorts of namespace issues with today's approach).
The example here doesn't need to be limited strictly to council meetings either, the city could host all sorts of video on its own instance of the PeerTube site. I know there's currently a "LACityClerk" channel, but I'm sure you could find other Youtube channels concerning the fire department, LAPD, Department of Transportation, City Library, etc:
https://www.youtube.com/c/LACityClerk/videos
I (as a constituent) don't really care about the underlying hosting. Framasoft, seeing as how they are intimately familiar with how Peertube works, can sell that underlying hosting. The benefit for the city is they can centralize administration and control over this rather than just give all their content to Google.
* Youtube-like website where people post videos
* A hosted service on your own domain where they host the videos (like the parent was talking about)
* A self hosted version where you host the videos and pay for technical support
Your clarification about the data being portable is definitely key here.
Cities aren't the only ideal client to make this pitch to either, there are all sorts of art collectives, educators, museums, even commercial interests who would probably benefit to move off of YouTube perhaps simply for the privacy and data collection concerns.
Mastodon already has plenty of services like that, such as Masto.Host, but none are "official".
It's the exact same thing with most platforms that people think will kill youtube. The only reason why they avoided implementing all the crap we hate about youtube is that they are not popular.
this would effectively cut Google out of the rent-seeking loop.
Unless you're in a jurisdiction where that stuff isn't illegal I think the answer is, yes.
It's especially nice that the videos are optimized when you upload them and thus seeking them is way better versus just uploading an .mp4 to Nextcloud and linking that. You can even configure how many videos and threads you'd like to allot to processing videos, so even my old box with a 200GE CPU in it runs great (even if slowly), without the actual front end breaking or anything.
Though of course I assume that all of the federation features and whatnot are great, too. Here's hoping that the project has a bright future ahead of it!
My personal pet peeve in regards to dealing with video production would have to be how WebM-VP8/Vorbis (libre) feels worse than MP4-H264/AAC for encoding, at least based on the speed and file sizes. Maybe using kdenlive isn't the best integration of these, but I think it uses ffmpeg under the hood and therefore should be a vaguely accurate representation of both of them.
Examples of config (quality settings for video/audio, kdenlive options output, though I'm not sure about the full command it uses internally):
H264/AAC with 30/192: f=mp4 movflags=+faststart vcodec=libx264 progressive=1 g=15 bf=2 crf=%quality acodec=aac ab=%audiobitrate+'k'
WebM-VP8/Vorbis with 30/6: f=webm vcodec=libvpx acodec=libvorbis crf=%quality vb=0 quality=good aq=%audioquality max-intra-rate=1000
I'd love to hear about this business plan.
Now, can we convince Mr Beast to join this platform? No, but I don't think most people will notice a discernible dip in quality. Most exploitation-tubers would probably scuttle off to Facebook or Instagram, but I think most of the Patron-supported content creators would buy-in to a well made alternative. Key operand being well made though...
Servers and other fixed costs can be handled by donations or selling merch.
Tickets and high bandwidth cases may not be a good fit for the project, or may be on a "as available" basis.
IF this was true we would already have a Youtube competitor. We don't.
As torrents demonstrate it's not necessarily bandwidth that's a problem, it's mostly durability and discoverability.
even then it wouldn't be be able to handle the traffic or keep the lights on.
im sorry but you clearly have not thought this through. its just not sustainable with this donation model.
This unquestioned given of "where's the monetisation?" is almost everything that's wrong with the internet today. It's seen as "gown up" and realistic, but doesn't even question it's own wisdom.
We don't have to speculate with hypotheticals on what an unmonetised service would look like either, it's literally how things used to work before the corporate takeover of the web.
EDIT: Though I'd have to add that there are ways to get paid which don't have the same downsides (changing the content to make it easier to monetise). It's like how people prefer films that keep their integrity, instead apealling to the lowest common denominator.
What I'm saying is that monetization changes the way videos are made. They generally tend to be padded out, with intros, attempts at social engagement, ("Please like and subscribe, and let me know how you feel in the comments!"). This style is actively encouraged by YouTube.
Moderation depends on the instance. The sysadmins choose what videos to ban and what other instances to federate with. Most users choose a single homeserver that reflects their values on moderation.
Diode.zone is an example of a popular "heavily moderated" instance. It has some decent videos.
The less-moderated instances tend to federate more so I think a lot of the unpopular stuff gets drowned out with "popular" stuff (which apparently includes a lot of european TV news, linux stuff, liberal/libertarian rants, video games, public domain films, and "my little pony" videos). There is very little NSFW content at all because the default behavior is to blur the thumbnails and many instances just block it anyways.
Edit: actually I just found some sort of vlog channel run by nudists that didn't mark their videos as NSFW. LOL
The idea is that you join or self host some home server, and the admin of that server makes sure to only show content from decent sources.
To this end the admins have a few tools. They choose the list of servers to show new videos from in the publicly available lists ("new videos" etc). So these will come from servers with a history of good moderation.
As a user, you can also subscribe to content from servers not specifically approved by the admin. If the admin finds it really offensive they can block that also. Then, as a user, you need to either give up on that content or find a different home server.
Obviously an admin is also responsible for the content they themselves host.
As I see it, the main blocker on adoption is the extreme difficulty of deploying this giant piece of software. The setup steps is long, complicated and requires implicit knowledge on production level systems, which very few developers have.
I've really smashed my head into this trying to get PeerTube to work and I just can't. Other software I deploy is always in a docker container. I don't care how complicated the software solution is, deployment is always a breeze if it's containerized. In fact, deploying docker contains can be considered a "no-code" solution for services like DigitalOcean and Render.com.
Docker PeerTube would be an absolute game changer.
Right now, PeerTube requires that I have a stable URL which can't change ever. Can this be changed? If it's docker, the actual url may not be necessary at all.
Please please please, make a docker deployment the top priority for PeerTube. It will multiply the number of developers that will be able use it and contribute back.
I’ve been running it in Kubernetes with very few issues.
The documentation is available here: https://docs.joinpeertube.org/install-docker
I would like to be able to host videos inexpensively, perhaps in an s3 bucket or on a raspberry pi. The free / cheap version needs to scale to at most 2-10 concurrent viewers (a more expensive version that scales indefinitely would be nice for commercial use).
It would be optionally password protected in a way that is client side encrypted (probably in a way that is transparent to the end user; such as via a key in a url, that loads a bit of javascript that fetches and decrypts the video).
That way, amazon or whoever won't accidentally decide my video is a pirated win95 iso or whatever. Also, the people in my videos won't (necessarily) have their images harvested for bulk survelliance.
I should have final say in whether the content is taken down (and own the domain name, in case I want to move to a new cloud provider).
It should be trivial to set up.
I get the impression that PeerTube meets zero of these requirements.
I get the impression that it's for personal video sharing. Technically competent users will host their own personal instances and host instances for their communities. These instances are federated into larger social networks where users can interact across instances.
>I would like to be able to host videos inexpensively, perhaps in an s3 bucket or on a raspberry pi. The free / cheap version needs to scale to at most 2-10 concurrent viewers (a more expensive version that scales indefinitely would be nice for commercial use).
I do not think something proprietary like an s3 bucket would be compatible with Peertube, but I have not maintained an instance. The system works with webtorrent (a version of BitTorrent that runs over WebRTC), so it depends whether the concurrent viewers are all watching the same video or not (because then they will seed the videos to each other using webtorrent).
>It would be optionally password protected in a way that is client side encrypted (probably in a way that is transparent to the end user; such as via a key in a url, that loads a bit of javascript that fetches and decrypts the video). That way, amazon or whoever won't accidentally decide my video is a pirated win95 iso or whatever.
The fact that your VPS host is untrustworthy and temperamental is not peertube's fault. Consider a better VPS provider perhaps.
>I should have final say in whether the content is taken down (and own the domain name, in case I want to move to a new cloud provider). >It should be trivial to set up.
These two things are in conflict. Merely getting ahold of a domain name is non-trivial, although I suppose you could operate it as a Tor onion service or something on the other end of zooko's triangle.
The US CLOUD Act ensures they're all untrustworthy (or on the wrong side of an iron curtain).
>> It should be trivial to set up.
> These two things are in conflict. Merely getting ahold of a domain name is non-trivial,
OK, then no more difficult than setting up a raspberry pi or hobby AWS account.
Peertube supports storing videos in S3 compatible object storage: https://docs.joinpeertube.org/admin-remote-storage
I get that it's not a big technical lift. I don't get why this doesn't exist in a polished form.
https://instances.joinpeertube.org/instances?sort=-totalInst...
A lot of my friends use diode.zone . That is like a more electronics-focused instance. There are views and likes, but mostly I just check my subscriptions. The popularity of a video depends a lot on traffic from external websites as well as how many instances federate it.
That said, I like this account for video tips: https://mas.to/@FediVideos
I believe that there are some services that collect videos for indexing etc. but I don't have experience with them.
Rock and a hard place. That's where I would focus on platform improvements though.
Platform-level ads are surely for everyone not big enough to attract sponsors.
Actually it has a huge reason to, and sometimes large advertising companies do get accused of such cheating. [0]
You're right, though, that getting accurate numbers is hard even for companies with large budgets to invest in stopping "fake" clicks, which usually involves all sorts of fingerprinting and surveillance tech.
Trying to recreate all that in a transparent and privacy-preserving way, across a network of nodes in different jurisdictions, and with no contractual obligations, is... well, if not impossible, then let's just say it's going to need a lot more blockchain to make it work. ;)
[0] https://www.marketing-interactive.com/facebook-sued-for-alle...
They have a pretty tight control over video hosting, video monetisation and video discovery, making it impossible for an incumbent to enter into one of those markets without challenging Google in all three.
You're kind of putting the cart before the horse here in asserting that those are three independent markets. I could just as easily say that restaurants participate in the "food cooking, table bussing, and patron seating" markets, and that it's way too hard for my upstart kitchen to compete in the food cooking market due to Cheesecake Factory having such tight control over the seating and serving markets. But the reality is, maybe my cuisine just sucks, and the singular market won't entertain it.
Isn't this comment explaining why Google isn't facing anti-trust issues?
And without those large advertisers your CPM will be extremely low which will just invite even more content aimed at capturing that market (spam, low quality videos).
People forget that decentralization doesn't work well for some industries, especially ones that rely on ad revenues (centralized entities) that rely on centralized control over audiences and content allowed on the platform.
There's a reason Youtube works so well and its because of its censorship and AI led moderation to appease the big advertisers. You can't rely on some regular joe advertising his Token on PeerTube to continue paying in crypto and neither can creators rely on PeerTube to be able to convert tokens on a third party exchange that is facing insolvency.
It's amazing to me how quickly people forget the on-ramp and off-ramp of money is still fiat, nobody is going to hold on to PeerTube's tokens for decades.
Likewise, cash for creators means better content and not needing YouTube.
None of this should utilize crypto may I add. Stripe and PayPal APIs are very comprehensive, especially for mass payouts, and split payment systems.
[1] https://www.bloomberg.com/news/articles/2022-01-13/paypal-su...
FWIW, I've used PayPal since the 90s, I run a biz that uses it, clients use it every day and have used it every day for over a decade without any issues, and I have personally sent several thousand PayPal Invoices over the years. Automated billing also works just fine, far more reliable than charging a card that will eventually expire.
Further anecdote: whilst you point to PayPal, I've heard of more people having their bank account frozen for whatever reason, e.g. someone stole their card, suspected fraud etc. We can point to any of these things as a worst case scenario, but likely 99% of the time, it's working as intended. PayPal's support has also been some of the best to deal with when it comes to API queries.
A lot of the instances are poor quality, with flickering buttons. They outright look like the player ui’s that porn sites use….
Is there a way to explore content across all peer tube instances?
Here is a pretty good search engine:
https://search.joinpeertube.org/
I usually try that if i'm trying to watch a video for free and I can't find it anywhere else. There are are servers that just re-upload documentaries and stuff.
https://kraut.zone/w/aWCVkSQTXKJ1yuR6o8zjDY
Kraut.zone is fetching this guy's vlog from his personal server. But you don't need to visit this random guy's site yourself or run any of his JS in your browser. But if you are really concerned about security you shouldn't even be using JS in the first place. "Le Walled Garden" is not an exception.
This isn't a problem with the search engine, exactly, but it'd be useful if it filtered out unavailable content.
https://kraut.zone/search?sort=-views&searchTarget=search-in...
You can find well-connected instances here, if you sort by "Following":
https://instances.joinpeertube.org/instances?sort=-totalInst...
You might also need to enable WebRTC. I know I usually leave it disabled for security reasons.
I think a client that is able to subscribe to remote instances client side (without an AP account somewhere) would significantly improve the state of discover ability across peertube servers.
It also gives some decentralization because the original source can go completely offline, yet as long as there is a copy of each fragment on somebody's machine, everyone can watch the whole thing.
They've got a great API, they've done fantastic work on live streaming, they have WebRTC implemented to distribute server load, and they've figured out searchability across multiple servers with sepia search. I don't see what needs improvement really.
Personally I think the 1 click monetization ad driven model is going to die, I don't think a lot of work should be put towards implementing something like this. People are moving towards a sponsorship model or patron model, and we are seeing all sorts of problems with the 1 click monetization with regard to patent trolling, content policing and the like, it just doesn't work out well. Of course with this transition comes a move away from monopoly on tube sites like what YouTube has, if someone is basing their revenue on patrons or sponsors they don't need to keep all their content on one website, peertube pretty obviously fits into this business environment well.
So I think as far as monetization, peertube should focus on tools for creators that help them accurately count their viewership and active subscriptions, and integrations that enable people to hide content behind a paywall for patrons, a plugin system for different payment methods would be fantastic.
Besides this I think peertube is pretty much feature complete, although some work could be done on decentralizing search.
Maybe they should curate that list.
No.