A technical look at ZeroNet | Hacker News Reader

A technical look at ZeroNet

znano.eu.org

74 pointsznano21 comments

> To eliminate the reliance on a central server, ZeroNet provides an internal plugin called BootStrapper, which allows peers to function as trackers.

Why use this concept of trackers when we have better (more distributed) alternatives in the form of distributed hash tables (DHTs), such as Kademlia, the one BitTorrent itself uses for trackerless torrents?

DHT still needs to be bootstrapped: https://stackoverflow.com/questions/1181301/how-does-a-dht-i...

One question on this topic interests me: In BitTorrent, do nodes that bootstrap using different bootstrap servers, ever see themselves in one single connected graph?

Since there are many different bootstrap servers for mainline DHT in BT, how is it ensured that nodes from two or more different bootstrap servers know about each other?

Is it possible (and how often does it happen) that a node that boostraps using server A will never find nodes that bootstrap using server B?

Yes, they should always see themselves in the same graph. Generally, bootstrap nodes will offer up addresses of further nodes that all belong to the same global network. This doesn't have to be the case, but it is the intention.

The nodes they know belong to the same network. Bootstrap nodes don't have to know about each other.

Yes. A network can split, all it takes is a node that knows nodes in both networks to join them.

But do such nodes exist in practice (in the current network). It would be interesting to research what clients connect to multiple bootstrap servers.

All implementations that I'm aware of use at least 4 bootstrap nodes. And you always bootstrap from all of them, there's no point just picking one. Here's my implementation, for example: https://github.com/anacrolix/dht/blob/master/dht.go#L106-L11....

Nice, thank you.

ZeroNet is now not in development but using DHT was on the list of things that were wanted to be implemented. It had many other problems like the .bit domains were kindof centralised by Tamas not using the namecoin blockchain.

ZeroNet is dead since the creator NoFish abandoned the project, the ”sucessor” is ZeroNetX.

It is a unique project and worth trying out, the community is somewhat active.

> It is a Peer-to-Peer Web-Like Network that cannot be censored or taken down, thanks to its decentralized nature.

Ah, so it's a child porn plattform, i see.

Why downvote something emotionally? It is that, and it is also for journalists and others to share censored info, and maybe censored software. There are many unsolvable problems in technology, in which the only solution is no abandon the technology altogether, abandoning also the benefits in the process.

Can you point to a single case where this was in fact used for that purpose? Cause for obvious reasons I cant prove my assertion though its fair to say we all know its true

From the article:

> It is a Peer-to-Peer Web-Like Network that cannot be censored or taken down, thanks to its decentralized nature.

That's - wrong. A suitably motivated attacker can or will take down your network, figure out the involved ISPs, find your precious server and your basement and shut it down.

Let's not kid ourselves. These networks are tolerated...until they become inconvenient.

Consider bot networks: They are taken down routinely. (just google https://www.google.com/search?q=which+bot+networks+have+been... )

Instead of basing a user-space network on these impossible ideals, find a economic incentive to operate them - centralized or not, federated or not. Someone has to pay for the Ramen.

And now downvote me for pooping all over "open, free, and uncensored network and communication."

The difficulty might be that an attacker would have to take down users all over the world to remove a page. Like BitTorrent, there is no server, the clients share the hosting effort all over the world. Not saying it would be impossible, only potentially very hard.

Can't parse this weird opinion. Do you think cryptography doesn't work? I don't understand.

The cryptography here is likely sound, but motivated actors have other tools, like traffic analysis (how much data and the timestamp at which it was downloaded), that could help narrow things down to few enough targets to investigate each more thoroughly. The more traffic with similar data sizes at the same time the harder it will be, of course.

And yet Bittorrent succeeded while MojoNation did not. Please, go and take your best shot at creating a distributed protocol that has the best chances for widespread adoption, as you see it. But no need to shit on someone else's attempt. Especially over some simplistic filler in some introductory writeup.

I did not know about Mojo Nation. I also discovered it was replaced by Mnet in 2002 : https://en.wikipedia.org/wiki/Mnet_(peer-to-peer_network)

One could probably argue that Bittorrent identified a more actionable economic incentive on the part of the swarm participants (wanting access to a very specific piece of media) rather than the more general objective in MojoNation of participating in a swarm that stored some set of arbitrary files that you or someone else may or may not come back to retrieve at some future point. Incentive engineering was considered for both systems, but only one of them hit upon a way to make the data pools self-sustaining.

Its the best that can be done when the underlying physical network layer itself is not decentralised.

There is a grain of truth in your post. But the "economic incentive to operate" is often misconstrued by crypto projects. I won't pay for what I already have for free, period. Economic incentives, you see. Doubly so if it's a service inferior to alternatives in every respect I care about. The final nail in the coffin is the necessity to buy your rainbow coins which I have no use of otherwise, a huge inconvenience and insurmountable hurdle even when that's just a few cents.

Reply on news.ycombinator.com