DISQUS

Martin King · 1 month ago

Could the who or what be Dave Winer?

Richard Cunningham · 1 month ago

For years it seemed company after company was trying to be the new Napster and they would get closed down and all the code was lost.

I suspect the reason that BitTorrent won is that is was open source so no one could find a company to shut it down. BitTorrent has the benefit of known protocol with multiple clients and multiple sites for finding .torrent files. I think in a sense, it is the HTTP of peer-to-peer, it won because not because it's the best, but because it's open.

So it stands to reason that bitTorrent of realtime will be someone who creates an open system that everyone can use.

eas · 1 month ago

The seeming deficiencies of bittorrent are interesting to consider. The other P2P clients I'm aware of, like Napster, Kazaa, Limewire, etc, bundled both content discovery and content distribution. BitTorrent only handled distribution, and even that wasn't totally decentralized, since the client needed a "tracker" to find other users with the content one wanted to download. As you note though, all the pieces were open, which, among other things, allowed the deficiencies to be addressed or worked around.

dai_vernon · 1 month ago

The classical way of content distribution via BitTorrent is through a tracker, yes, but it isn't entirely necessary. The DHT made the tracker optional (though very nice to have) - hence why the torrents listed on the Pirate Bay kept working when the tracker was under a denial of service attack

BrianSullivan · 1 month ago

I am thinking Flash or Silverlight would be prime candidates as the repository for the technology. They have the need and something like realtime delivery of Olympic coverage would be a spark. I thought I vaguely recalled that NBC/Microsoft did something like this with the last Olympics and Silverlight (or did I imagine that?).

Whether Adobe or Microsoft could get their act together enough to incorporate or build the technology and infrastructure necessary to make it happen is another question.

Constantine · 1 month ago

I think the best candidate for this right now would have to be Opera. They're trying something really crazy by making a server inside the browser. Granted, the Unite project is a bit half assed, and closed source, but that type of thinking, to finally enable read write is what will change things hopefully.

My biggest want right now is to make internet services like applications on my desktop. Meaning I have Photoshop, Lightroom, Irfanview, <insert name of image editing/viewing program here>, and they all access the same instance of one JPG file.

I should be able to tell Flickr, Posterous, Wordpress, whatever, that "hey look, my files are sitting on this server; go use that data to provide me an awesome service"

Data portability as a concept makes me slap my forehead. It isn't about making your data easy to export and then import into another system, it's about letting data be data, services be services, and having the two be totally separate entities. Opera's Unite concept of putting a server inside every browser is similar in concept to bit torrent since you're just another brick in the wall helping people commit piracy.

These thoughts are all out there, but I hope in 5 years you're right, and Facebook is dead. Best thing I did in 2009 was get rid of it.

eas · 1 month ago

I'm not sure how far Unite goes, but I think the concept of putting a server inside of every browser may get further in a more browser-agnostic incarnation. HTML5 local storage features are becoming more widely supported. The mobile browsers on iPhone and android support enough of it to enable browser-based offline versions of gmail and google reader, and on the desktop, I know Safari 4, and I think Firefox 3.5 have similar levels of support, and if IE8 doesn't, it can be added with Google gears.

The local storage features mean that javascript code, images, and html data can be stored locally, along with a persistent datastore. The app runs inside the browser when the user visits the page. Once started, it can reach out and request updates over HTTP.

The shortcoming of this approach is that it doesn't provide for long-running processes that can be accessed remotely, but I question whether that is a big strike against it, since firewalls make P2P communication across the Internet difficult.

The missing piece then is a way for peers to communicate. Some sort of generic HTTP accessible message-queue could provide that piece of the puzzle.

dai_vernon · 1 month ago

I agree more-or-less with Richard Cunningham, and think that the dominant service will be something open that just takes over because nothing can stop it. I don't know what that is going to be, but it sounds like what people keep saying about the successor to Twitter - whatever manages to supplant Twitter will take over the real-time web.

furbuzz · 1 month ago

* Containers

* Contents defined by users (e.g. blog / pictures / music / whatever)

* Open Source File Formatw

* Secure Encryption / Keys

* Automatic Sync / Directory Contents / File Contents

* Distributed Storage / No central processing / nodes aware of other nodes with Client Software able to propagate messages in near enough real time

* RSSCloud

* Persistent, Distributed Storage based on Open Standards in a way that is unstoppable.

Or I don't know what I'm talking about.

But RSSTorrent, it seems to my relatively uninformed self, to be a way forward. Particularly, if there is some (self-defined) rights model that confederates "private" data within a defined private cloud of machines (such that none have too much data to allow cracking, and for them for to be friends anyway).

Super secret stuff never leaves the owner's possession, of course, as always.

David Recordon · 1 month ago

Stepping away from who will get brand value for a minute...

I think it's also worth thinking about what the BitTorrent of real-time will look like from a technology perspective. Both RSSCloud and PubSubHubbub are built around a model where a feed has one (or more) hubs that it specifies. This means that the hubs need to be super scalable. BitTorrent changed the downloading model, no longer did the site you're downloading from need to have a giant pipe, but instead you get many smaller pieces from lots of sites each with smaller pipes. Is there a similar model that will need to emerge for real-time, many hubs with each giving you a piece of the content you're interested in?

dai_vernon · 1 month ago

I think that Twitter, with its constant fail whaling, has proved that yes, we will need that. I think something like BitTorrent with distributed distribution will suit those needs well. I wrote about an idea using the DHT (distributed hash table), a common piece of filesharing tech, in a general sense on my blog, here: http://taylorheffernan.squarespace.com/blog/200...