Hacker News new | past | comments | ask | show | jobs | submit login
BitTorrent Inventor Demos New P2P Live Streaming Protocol (torrentfreak.com)
72 points by Uncle_Sam on Jan 20, 2011 | hide | past | favorite | 22 comments



I hope he builds cryptography and maybe better privacy features in the protocol from the ground up this time. If wikileaks proved anything it's that we need a more distributed web, that's probably one of the only things that could guarantee any form of net neutrality.


That is far easier said than done. If the streaming is point to multi-point then the crypto can get tricky, if you allow people to join and leave the group at random it gets tricky, and keeping it all working without providing a list of the participants to each peer is tricky. With BitTorrent Bram started out with a system that had strong crypto and privacy (Mojo Nation) and stripped most of it out because it increased the complexity for little benefit.

Strong crypto and privacy have costs, both in end-user resources and bandwidth implications. Getting it all correct is hard and if you screw up a little it is usually worse than starting from the position that privacy and a secure channel is something that the user has to bring to the table if they really want it.


There doesn't seem to be very much public information about how this protocol actually works. How does it deal with NAT? Will it work on mobile networks? Does it require a native application?


Any p2p requires a native app as the browser can't act as a server.


This isn't 100% true. You have unrestricted network access from signed java applets if you request the right permissions (and the user clicks allow), for example.

Also, Bram might have come up with a reliable way around the same-origin-policy. DNS rebinding 2.0?


So a Java applet can listen on a port? Interesting, although not very practical.

Adobe Flash is also supposed to support some form of p2p connections, but I don't know anything about it.


The warning dialog is hella scary, though.


That is not necessarily true. With websockets, you can write a full-functioning p2p client in the browser.


WebSockets is only TCP. Bram is using his own networking stack based on UDP.


You can't connect two browsers directly via websockets, you can only connect to a server. A browser can't serve an incoming websockets connection.


Only if you can find some way around the same origin policy.


Actually, p2p streaming is patented by Spotify (likewise, p2p voip is patented by Skype).


What the fuck? How is p2p streaming even patentable? Good thing nobody patented server-client streaming. We badly need patent law reform.


You don't happen to have a link to that patent?


Here they are (after a small googling):

http://v3.espacenet.com/searchResults?locale=en_EP&IN=eh...

Specifically take a look at these claims: http://v3.espacenet.com/publicationDetails/claims?CC=US&...

I'm not really sure whether those are actual patents or patent applications, but even in later case the chance of them being rejected are very low.


Priority date 2007, I'd be surprised if there wasn't any prior art by that date. I know I was working on a P2P streaming system in 2006, for example.


There are many p2p networks for streaming content - they seem to be popular in Asia for watching English and European football, and I think many of these existed prior to 2007.

I'm sure there's a mountain of other prior art covering the concept. It isn't novel to anybody with half a brain. I was streaming audio in the 90s and clearly remember thinking about a p2p network then, because of the expense and legal threats I was receiving. All I wanted to do was enable football fans living abroad to listen to the local radio station on match days.. I thought I was doing the BBC a favour by extending their broadcast range! Yeah, I was just a naive kid. :)


PeerCast.org comes to mind. April 2002. IIRC it was based on Gnutella.


I would not worry about this patent. There is prior art.

For instance, this paper I wrote in 2002 uses techniques which overlap with their claims: http://www.springerlink.com/content/ujrmkgp929v6je7p/


Sounds a lot like http://esm.cs.cmu.edu/about/index.html

btw, I hope he doesn't plan on making money on this. Rinera (now Conviva) tried and failed. Media company executives freak out at any mention of P2P. I bet they freak out even more when they hear P2P and Bram Cohen in the same sentence.


So is this going to be an open source competitor to Abobe's Octoshape?

http://www.octoshape.com/addin/about.php

http://en.wikipedia.org/wiki/Octoshape

Can it be made to work with Flash somehow? Because that would definitely be a winner.


It probably cannot be made to work with flash (without Adobe's help.) To build a p2p system like this, you need greater access to the OS than the flash sandbox will allow. The only way to get this is through a browser plugin. Octoshape had a special deal with Adobe where the flash player would install the Octoshape plugin, without the user having to do anything. This was because the code was signed by Adobe.




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

Search: