I wish BitTorrent Live was made open source. There are a number of companies working on "P2P CDNs" for live streaming[1][2][3], but all of their work is proprietary. The companies' sales pitches are usually around efficiency: it allows providers to pay for less CDN capacity.
I'm more interested in the decentralization aspect. It makes live streams hard to censor and easy to distribute.
The closest open source equivalent we have is PPSPP (Peer-to-Peer Streaming Peer Protocol), or RFC 7574[4]. Unfortunately, the reference implementation, libswift[5], was seemingly abandoned a few years ago. In addition, PPSPP only deals with a stream of bytes, and while it was made with video in mind there is no implementation that handles video well: making different swarms for different quality levels, intelligent chunking, perhaps even deterministic encoding[6].
- Doing adaptive bitrate like YouTube does means doing many transcodes.
- Transcoding is either extremely CPU intensive, bandwidth heavy or low quality.
- Transcoding in realtime requires further compromise on the above, to the point where distributing the transcode over multiple machines is necessary.
- There doesn't seem to be an agreed upon, browser supported protocol for live DASH. Static is easy enough but live is more difficult, especially when you add seeking, pausing etc.
- The above means you need both a backend and a frontend system.
- Running a system like this is very expensive. The entities interested in running it and able to do so are usually willing to pay for it.
All this said, Emby[0] can stream a transcode and perhaps its stack would be a good place to start.
I'm actually an engineer on the BitTorrent Live product and we definitely hear that desire. While we can't exactly open source the protocol itself, some of the libraries we've developed are currently freely available under permissive licenses.
If you're interested in the tech and want to read some related stuff, we're hosting an engineering blog over at https://blog.btlive.io/.
There was one reasonably successful "http" company - Netscape. While most known for their browser, they also had a lot of products for the other end of the connection such as HTTP servers, Commerce servers, proxies etc.
Updaters/patchers for some popular games (Blizzard stuff, League of Legends, etc.) use P2P methods to reduce hosting requirements when pushing big updates to hundreds of thousands/millions of players.
I'm sure there's other use cases, but I'm not sure what they'd be.
Peer-to-peer only ever made sense for popular stuff that people won't pay for. By definition there is little business opportunity in markets where people don't want to pay.
>By definition there is little business opportunity in markets where people don't want to pay.
Counterexample: network TV, radio, facebook.com, gmail.com, pretty much all of the ad-supported web.
>Peer-to-peer only ever made sense for popular stuff that people won't pay for.
Counterexample: Skype (freeium model on a P2P network). The content wasn't "popular" it was private phone conversations and people paid for it.
Counterexample: Bitcoin (runs on a P2P network). People pay fees to miners so that the network relays their transactions.
Counterexample: The internet. A giant P2P inter-network with all sorts of different business relationships. Many of which involve paying someone to send a packet that is only meaningful to a particular party.
Yeah. The article is like "everyone at BitTorrent Inc said Cohen is brilliant, so he's brilliant". BitTorrent is a shit way to move files that only makes sense if your users are willing to subsidise your bandwidth for you. If you look at how big files are moved on the net in practice it's all professional CDNs. BitTorrent is hardly used outside piracy.
>170 million people used the protocol every month, according to the company’s website. Facebook and Twitter use it to distribute updates to their servers. Florida State University has used it to distribute large scientific datasets to its researchers. Blizzard Entertainment has used BitTorrent to let players download World of Warcraft. The company’s site boasts that the protocol moves as much as 40 percent of the world’s Internet traffic each day.
Doesn't sound like a "shit protocol that only pirates use" to me.
Exactly - I live in an area where lots of people have 50+ Mb symmetric connections, and BitTorrent gave us all a dream that we could all use those connections without everything having to be transferred over the backbone.
I'm not sure about that. A lot of what I've read about sounds like they were chasing Netflix with a lot of cash but no idea how to compete.
The article mentions that big cos are using BitTorrent for cluster upgrades. There's a lot of cash in enterprise networking. BitTorrent is basically a large blob multicast protocol.
BitTorrent Sync was a decent product too. What would have happened had they put $30 mil behind a push to take sync from Dropbox and Box instead of sidelining it? Seems like they spun out the best contender they had. It'd be funny if the spinoff succeeds.
It could be the classic situation of FOSS. They had a product people would've paid for if sold on its advantages. They gave it away for free. It got picked up by tons of freeloaders. With free version out & good enough, they weren't able to figure out how to make money off it. Messed themselves up.
For those not aware of the current situation: Dick Smith Electronics went bankrupt about 12 months after this article.
They were like the Tandy or Radioshack of Australia at first, and followed a similar path - killing off their electronics hobbyist roots, and became pretty much a generic consumer electronics store.
One good thing that came out of the disappointment of bittorrent inc is the open source syncthing which I believe was inspired by bittorrent sync's failings.
I wish they would improve the story around the official distributions.
I'm attempting to migrate away from an aging, single user, files-only OwnCloud instance to Syncthing, and the end user client is not nearly as painless. Syncthing distributes CLI binaries for a bunch of platforms, which is great. But there's no unified GUI for end users that's officially sanctioned. Instead, there's a bunch of 3rd party repos that one hopes they can trust. And, glaringly, zero iOS clients. (I'm working around this by adding a WebDAV server to my always-on system, but it's a workaround for sure.)
What a crazy twist for a startup that was doomed from the start. But the article doesn't really do a good job explaining how Johnson and Delamar were given control of the company. I get that Accel wanted out, but why just hand it over to a couple of bros with a plausible plan? I feel like there had to be deeper relationships involved, or else one hell of a pitch deck.
Whenever I read stories that involve guys like these that come in swinging and think they can change things and end up just spending everyone's money, I've always wanted to label this behavior and you got me thinking ... "Broxecutives".
Klinker vouched for them. He called Jeremy Johnson a "product visionary". Then as quid pro quo, BitTorrent (with Delamar and Johnson in control) funded Klinker's Resilio spin out (to the tune of about $5M of BitTorrent's cash). Oh yes, BitTorrent is still paying Klinker his CEO salary even to this day.
Normally the BitTorrent's general counsel would have objected to all this but Klinker (the CEO) and BitTorrent's long time general counsel are married to one another. How the board ever allowed the CEO to hire his wife as general counsel is totally bewildering.
Well that's fascinating innuendo. I'm not equipped to evaluate if it's true, but it makes sense. I wonder how the other equity owners feel about the outcome.
I wonder what's happening with Bittorrent Live. Maybe the economies of scale of the large companies have made P2P streaming a hard sell?
Why install a weird temperamental app when you can just stream video from Twitch (Amazon) or YouTube or Facebook? They've all invested tens of billions in networking infrastructure and have the necessary advertising structures to recoup the costs.
The economics have definitely changed. CDNs are dirt cheap and are technically simpler and more reliable than P2P. And for live video you probably want transcoding which is also easier to do in the cloud.
As former eng lead at a live streaming company, I second the idea that transcoding is a big deal.
Streaming bandwidth is super pricey, and the money is enough to really matter. Even a startup can easily get to five or six figure expenses. Bittorrent Live gets that right.
However transcoding is critical, given that every OS and browser has its own format quirks. BT Live can't remotely do this.
Check out Zencoder for the real competition.
Maybe BT Live will find a niche with use cases that just produce no money at all, or where servers get shut down for political reasons. Marches, samizdat, and, uh, questionable copyright.
P2P streaming solved a problem when bandwidth cost a lot. Now that it's so much cheaper everyone just chooses to serve it themselves. Even for small operations there's tonnes of cloud options like DO, Scaleway, OVH, et al that offer cheap bandwidth. 10 years ago I paid $12/mbps for dedicated servers just for streaming, now I can get ~200mbps for $5/month (0.025/mbps).
BitTorrent was, and is still, remarkable technology. Not focusing on the ethics and costs of piracy (that affects one part of what BitTorrent is used for), it has helped all kinds of content (entertainment and otherwise) have a massive reach globally (and across country borders) like nothing else has. I only feel sad for Bram Cohen in this whole series of events, and more so considering his condition (Asperger's syndrome) and how much he would've just put up with (in silent frustration) while others ran the show.
In this article, I was surprised to see no mention of uTorrent, the BitTorrent client, and how it changed from the time it was bought by BitTorrent, Inc.
P.S.: Off topic, but I dislike these Medium hosted sites with large blurry images on page load that come into focus later.
I think it would've worked if they had started off in a SAAS model, like, pay $1 for exchanging 1GB of data, $10 for 10GB of data and so on. At the time it launched there was still requirement for moving data across sites , but no reliable way. For ex: most software/trailware/shareware came in CD ROMS bundled in magazines !
Kind of a similar story happened to Diaspora Inc. It didn't manage to become a product or a company. But technology was good and open and it lives on today.
Meta, but did backchannel change the title, or did the submitter? It seems like a weird editorilization, from the current "The Inside Story of BitTorrent’s Bizarre Collapse".
"Unbelievable story of..." feels very click-bait to me.
Backchannel could have changed the title, wouldn't be the first time a title's been changed somwhere. The URI includes "the-inside-story-of-bittorrents-bizarre-collapse" which makes me believe otherwise however, unless they've been flip-flopping the title, or A/B testing.
BitTorrent Inc was ethically shady from day one. Call me old-fashioned, but the investors got what they deserved by trying to profit off of piracy while pretending to be legitimate.
I'm more interested in the decentralization aspect. It makes live streams hard to censor and easy to distribute.
The closest open source equivalent we have is PPSPP (Peer-to-Peer Streaming Peer Protocol), or RFC 7574[4]. Unfortunately, the reference implementation, libswift[5], was seemingly abandoned a few years ago. In addition, PPSPP only deals with a stream of bytes, and while it was made with video in mind there is no implementation that handles video well: making different swarms for different quality levels, intelligent chunking, perhaps even deterministic encoding[6].
[1] https://www.peer5.com
[2] https://viblast.com/pdn/
[3] https://www.streamroot.io
[4] https://tools.ietf.org/html/rfc7574
[5] https://github.com/libswift/libswift/blob/devel/TODO
[6] http://www.ndsl.kaist.edu/~kyoungsoo/papers/mmsys14.pdf