Roll Your Own Television Network Using Bittorrent 252
Cryofan writes "Mark Pesce, lecturer at the Australian Film Television and Radio School (AFTRS) writes here and here about using p2p networks, specifically bittorrent, to create a grassroots television network. He cites as an example the BBC's "Flexible TV" internet broadcasting model using that as the core of a "new sort of television network, one which could harness the power of P2P distribution to create a global television network." Producers of video entertainment and news would provide a single copy of a program into the network of P2P clients, and the p2p network peers distribute the content themselves. Thus, a virtual 'newswiki' where the content is distributed bittorrent using some sort of 'trusted peer' or moderator mechanisms as a filtering/evaluation mechanism. So what is stopping anyone from doing this now? Awareness of the concept, perhaps? Lack of broadband connections? Lack of business models for content producers?"
Well, i did this. RSS + Bittorrent (Score:5, Interesting)
Huh? (Score:1, Interesting)
Ohhh, you mean legitimately!
They are doing it... illegally (Score:2, Interesting)
My first question was, 'why'? (Score:5, Interesting)
From the standpoint of news broadcasting, this could be really big, though. Set up a
P2P (Score:3, Interesting)
No No fundage necessary (Score:3, Interesting)
A friend and I produce a little 1/2 hour news talk show which we broadcast on local cable channel three. Now we are looking to get it on our local pbs station. costs are negligable. My friend who is a tech freak has the latest G5 with a DV card and a high end Sony Cam (about $5000 in hardware). Studio time is free based on cable regulations. (if your not aware FCC requires cable operators to provide free service and equipment to local users.) for us this included a studio with 3 mounted cameras, an editing room and post editing equipment. The hardest recourse is time. but for someone who is dedicated is the price we pay.
Torrentocracy (Score:4, Interesting)
Multicast (Score:3, Interesting)
Pesce a good speaker - knows his material. (Score:4, Interesting)
I attended this talk at the National Student Media Conference last weekend, ( for any other attendees, I was the NSMC volunteer managing the digital projectors... ) and it was interesting to see the ideas mooted here percolating out into the other panels that took place over the rest of the conference. I think the independant media needs to continue to forge closer ties with the tech community to allow things like this to come to fruition.
One thing that didn't get brought up was whether this will compete with or complement Indymedia's upcoming IVDN video distribution framework. I was hoping to chase Mark up on this after the conference, but lost his email address - thanks submitter!
YLFIP.S., Mark, if you're reading this, I crashed in your suite on Sunday night - thanks for the keys. :-P
Re:They are doing it... illegally (Score:5, Interesting)
Is this possible with BT considering that it sends out blocks in a non-sequential order and the .torrent file contains SHA-1 hashes of the blocks? eDonkey sends out blocks in random order, as well, in order to optimize against the rare missing block problem. I think this is a good optimization to take, especially on file distribution networks, but it sacrifices the ability to stream (as far as I know). Anyone know any more about this?
Peercast already does P2P video. (Score:3, Interesting)
I've had a go with it and its not too shabby.
With clients for Mac, Linux and Windows, availability is good. Unfortunately, Peercast doesn't advertise themselves too well which means there aren't so many video streams available yet (typically 5-15 video streams and 100 or so Audio streams.)
Slashdot TV (SDTV)? (Score:1, Interesting)
How about Slashdot TV! 24 hour Nerd News.
Slashdot effect the world!
~-~
Anonymous Coward - The one and only
Re:um they already are doing it (Score:4, Interesting)
BBC... (Score:4, Interesting)
Blogging is the news network of tomorrow. (Score:2, Interesting)
The obvious answer to my mind is bloggers.
Imagine getting your news not from CNN / Fox, but instead actually from someone on the ground living in an apartment in Baghdad while it's being bombed?
Get news reports on SCO vs Everyone not just from the media and court filings, but actually see image of the court building where it's all happening with bloggers telling us how they think the proceedings are going at the moment.
Blogging is the news network of tomorrow, and this is how it will be done.
Re:um they already are doing it (Score:3, Interesting)
Re:Bittorrent is not the right way to do this (Score:3, Interesting)
There's also other more immediate practical limits. Many users now are connecting via broadband, which is great, but the transfer speeds you get are asymmetric. So even though you have a very fast potential download rate, your upload rate is very limited. In a bittorrent setup where every peer's download rate is proportional to their upload rate, this means you are inherently unable to utilize your full download capability, because your upload rate cannot match it.
Also, a lot of users are attached to the same network providers. The routers in those networks are transmitting the same data over and over to each of the individual users. That's an unnecessary waste of bandwidth.
A sane approach would be to have a program guide (like TV Guide) published at a well-known URL that tells you what content will be available in what multicast group at what times. Your client software will join the multicast groups of interest at about those times. This informs the routers upstream from you that you're interested in a particular channel. When the multicast begins, the sender just needs to send one copy of the data to its local network, and all of the routers on that network will fan out one copy of the data per target network. This immediately reduces the network resource load from the NxN nightmare to a near-constant value based on the size of the network, as opposed to the number of clients. And in the common case where you have a bunch of broadband subscribers all connected to the same router, the data only traverses that network link *once*, no matter how many subscribers there are. No more wasting bandwidth with N copies of the identical data. And, everyone gets the data at their full download speed.
You may think that bittorrent/suprnova are successful *right now* but they can never hope to reach an audience of millions, the way a TV network does. The internet would melt down long before they could do so.
Any design whose network resource consumption scales based on the number of users is doomed to be a victim of its own success. But the approach I've described will scale efficiently because it's only dependent on the number of routers in the network, not on the number of listeners trying to receive the content.
Re:um they already are doing it (Score:3, Interesting)
Re:Bittorrent is not the right way to do this (Score:2, Interesting)
Yes, and then again, no. (Score:3, Interesting)
Firstly, it can only make educated guesses at the available bandwidth of the nodes. Nodes will lie/cheat/steal in order to get more packets, and you can't trust the clients. They're greedy.
Secondly, it doesn't really know the network topology. Again, you're only able to make educated guesses. If my neighbor and me are on the same torrent, then ideally the tracker would be able to tell us about each other, we'd connect, and share at very high speeds, being that we're both close to each other and on the same subnet and such. That case might be easy to recognize, sometimes, not so easy other times. Without full knowledge of the whole network, it's impossible to do perfectly in any case.
Third, even with the most efficent possible tracker, the grandparent is right. You have X users downloading, and they all are downloading Y bits of data. All data transfer is point to point, meaning that X*Y bits of data must be sent out for everybody to get the complete file. For every byte downloaded, there's a byte uploaded. You can make that fast by maximizing your throughput and managing it all into small sub-networks, but it still doesn't scale to everybody in the world.
A multicast setup does scale, even if it is a pain in the ass to do right now. One byte sent out from the source gets duplicated for each branch in the routing tree, and all users receive it. Upload rate is constant. If you ignore new users joining and old users leaving, traffic along each branch in the tree is only one copy of the stream, all the way until it reaches the endpoints (the viewers).
The problem with multicast is that it's confusing as hell because it requires cooperation of all the routers to handle the multicast traffic appropriately. But for any single source to many receivers, it's easily the most efficent way to do things.
And let's not forget that while torrent trackers *could* be more efficent, they are quite simply not that efficent. The torrent network is often highly connected instead of sparsely connected. Especially on larger files. A sparser network would be more optimal (read as: faster) in extremely large torrents, but it is rarely the case currently.
Good Content Idea (Score:4, Interesting)
If someone (PBS?) could release all of their educational content under a non-restrictive license then I'd happily pay for the dedicated servers to host and track the torrents. Math, History and Science programs would get even the adults involved but would be a great resource for people who are home-schooling or parents who want to keep their children occupied when home sick from school.
I don't know why we, Americans, have not done this already. I suspect that bandwidth is an issue but that is somewhat silly as it is otherwise wasted on illegal downloads and that sort of thing.
There should be a public education page that acts as an entry point for materials for students and teachers alike. Think "cable in the classroom" turned into "internet in the classroom". Why haven't a few public school teachers already gotten together and made this a reality? 30 minute shows aren't that hard to make. Take your lesson plan and turn that into a script. Read it, or hire someone to and viola.
The BBC and iMP (Score:2, Interesting)
Their own webpages are a little light on content and mostly aimed at helping out the Beta testers, but more useful information can be found on various [digital-lifestyles.info] sites. [theregister.co.uk]
iMP is P2P client that allows distribution of BBC programmes. There is a DRM component that stops a programme being watched 7 days after downloading. iMP is a great idea for the BBC as it has the potential to significantly reduce the infrastructure costs in terms of streaming and network bandwidth required. A big question for me though is how robust their DRM technology will prove to be.