Hacker News new | ask | show | jobs
by Fice 3806 days ago
Another problem is why we at all need to rely on any third-party hosting services while we have our own computers connected to the Internet. In this regard, I have high hopes for content-addressable distributed systems.

Currently, IPFS [1] looks most promising to me. It's easy to publish a static website to it [2], and public gateways are available to make the site accessible to readers who do not run their own nodes. While IPNS (the naming system) does not seem to work reliably yet, hash of the current version of the web site can be published in a special DNS TXT record, and if that same domain points to IP address of a public gateway, the web site will be simply accessible via that domain.

[1] https://ipfs.io/ [2] https://github.com/ipfs/examples/tree/master/examples/websit...

2 comments

Been feeling this way about hosting for a while now... I'm actually anticipating (wishing?) that "p2p web" (decentralized LAN applications) is going to start taking off soon. I think it's a matter of people building software that makes LANs easier to use for ordinary people. BT Sync is an excellent example of something like this. So is WebTorrent.

I had fantasies when Web Audio and Web Midi first were announced of browser based digital audio tools that people would use in one anothers physical presence, over a small local network. There is no reason that every network application needs to depend on centralized services! I was lucky enough to attend Music Hack Day in Berlin and created this Web Midi/WebRTC chat room as an example. I have no idea if it still works or not: https://github.com/timbresmith/pandemonium

My computer is not always up, or connected to an open (even moderately) network. What's more, I sometimes have to delay for 2 days an update because it requires an update at the wrong time, if I didn't missed it... And that's without even speaking about the reliability of my internet connection - I don't want to try to reach even 3 9 on it. So yes I have a computer at home, but it's not a properly managed server, and I don't want it to become one !
In a content-addressable network your computer seeds the content rather than serves it. You do not need a properly managed server for that. It does not matter whether the original node is accessible, information can be retrieved as long as it is available somewhere in the network.

In IPFS, if you need to immediately ensure availability of your content, you can request it via a public gateway to force it into the gateways cache. Then the gateway will continue to seed your data even if your node goes offline.