The closest analogy I can think of for what is mentioned on article page 3 is that of using rsync to create a redundant stream of your data, then streaming it as a rapid-fire set of udp packets. Set up some feedback loops to control the sending rate but as long as the percentage of lost packets is below some threshold the receiving end can still unpack everything.

Interesting topic, interesting researcher, lots of new information, but there's a lot of bogus here too.

Yeah, I don't get why this stuff is considered so revolutionary. It's like no one involved ever studied tcp/ip.

Nice article. Discover Magazine also seems to be a great place to find things to read.

Anyone have recommendations for other websites that have good articles (aside fron News.YC)?