Post edited 23:26 – March 9, 2009 by UmbraLux

I can see several potential uses ranging from mundane off loading computations onto the 'client' to grayer apps tunneling or even redirecting requests (useful for both privacy and bypassing network level security) and on up to outright malware either pulling data from your computer or inserting other malware on it. 

One of the links you posted mentioned doing it for browser based games – think of the hardware savings if you can offload some or all of the game's computing requirements on to the gamer's machine.  Yet you still retain many of the advantages of keeping code on the central server, any client side code modification is temporary and corrected with the next download, code deployement and patching is centralized, etc.

I ran across this recently but don't really understand its usefullness.  Anyone have any ideas?  It is a short draft that basically describes a technique to allow a client to contact a server and in turn allow the server to become the client and contact the server (originally the client).  I can imagine a few ways this might be applied but not for anything practical.

http://www.ietf.org/internet-d…..ttp-00.txt

I have read a few posts about this which had basically rehashed (in more detail) what I have said above.  The best I can see is that this could give the “server” a way to inject data into the original client browser.  Maybe this is tantamount to a new type of browser interaction or richer web UIs.  I still don't understand it but I am intrigued.

Some reference posts (there aren't very many):

http://ulaluma.com/pyx/archive…..se_ht.html

http://wiki.secondlife.com/wik…..verse_HTTP