Hacker News Reader

Comment by politician | Hacker News Reader

Telehash is the Open Source "firehose API" you seek. The proposal describes a DHT that routes JSON. However, it seems like I've been waiting for it for years. Maybe it's the obscure telephone lingo that the API is described in terms of that's discouraging would-be implementers, but it's nothing that a good old facade can't fix.

http://telehash.org/

View on HN

> However, it seems like I've been waiting for it for years.

https://github.com/jamii/erl-telehash is a reasonably complete implementation (it's only missing rate limiting and auth).

After working on erl-telehash I have some doubts about the protocol details (although the overall idea still appears sound). I described how I would modify it at https://gist.github.com/1910754 but I'm not moving forwards with it at the moment (working full-time for now).

I like to think of Telehash as the answer for p2p routing. It's powerful and there's working implementations.

The other piece of the puzzle is REST in the UDP context, for that, CoAP looks really well put together: http://tools.ietf.org/html/draft-ietf-core-coap-10

I have a weekend project going to blend the two and make magic, but I haven't put enough time into it yet.

Is there decent CoAP implementations? That's unfortunately one of the big stumbling blocks with universal p2p transport yet :/

Telehash is radically decentralized. How do you get a firehose out of that? It would be more like a point to point garden hose wouldnt it? I would like to be proved wrong, but I think Twitter/centralized and Telehash/decentralized will end up looking and acting like two very different services.

If the Twitter as an API group had won out the two best outcomes I can think of are either a buy out from Google, or a move to a foundation like Wikipedia. Google could afford to carry it, like it does Youtube, just so they could own a huge firehose of information and use it elsewhere in search ads, without trashing the stream with ads.

Needless to say founders looking to get rich wouldn't go for the low returns of the foundation model but it would have been the best possible outcome for the well being of the future Internet having and open, censorship resistant message bus and API, assuming you could get enough donors to support it.

Also analytics will be very difficult in a P2P architecture. That is the power of twitter: you only need to analyze the centralized firehose.

With a telehash style distributed messaging bus you'd need to have clients opt in to centralized reporting and analytics providers, like the library of congress, internet archive, even commercial firehose companies like gnip, etc.

Reply on news.ycombinator.com