Idea to immensely strengthen Namecoin

[MWD]

Idea to immensely strengthen dot-bit (Namecoin Domain system) with distributed trustless DNS-like system based on Bittorrent

How about distributing a copy of the block chain (and updating it), via something like Bittorrent for Namecoin. With a utility for your phone or computer to use it as a roving DNS server, without having to have the whole blockchain.

Including a function of checking and not bothering with addresses with no associated website for 3 out of the last six days or something like that.

Maybe it could be built on the code of uTorrent (With Namecoin wallet).

Also there could be third-party utilities like PeerBlock for Bittorrent that would ignore bunk addresses. Of course, that adds a central authority to it, but people could disable it if they can't get to sites they want to get to. PeerBlock does a kick-ass job with a continually updated list of blocked addresses for torrenting.

People could even make their own third-party censorship utilities. If Christian parents want to get to blocked gun sites but keep their kids off porn sites. It's not true "censorship" if you chose it rather than having it forced on you by some state backed by the threat of being thrown in a cage.

(There's talk in the gun blog world that a UN-ICANN is going to block gun sites, seems plausible, here's a video warning about the possibility:
http://www.youtube.com/watch?v=B2Ynelygvpk
But his solution is silly. If "GunsAndAmmo.com" is blocked, the same site under "FluffyPinkBunnies.com" would quickly be blocked too.)

Or if Arab-spring "dissidents" wanted political news to come in and out but wanted to keep out "decadent western pop culture", the market could take care of that (to an extent - no IP or name-blocking system is 100% perfect on white listing and black listing).

I really see no problem with that type of third-party solution, as long as its used in a voluntary way. Different companies could make different configurations, and people could chose what they want to block.

This would also make .bit slightly less of a threat to the nanny powers that be who would want to destroy a system outside central control. At least we could say "See? there ARE solutions if people don't want it accessible."

Of course, third-party utilities add a central authority to it, but people could disable it if they can't get to sites they want to get to. PeerBlock does a kick-ass job with an updated list of blocked addresses for torrenting.

Derrick adds: A trustable third party will likely be needed for some situations, I like the idea of independent auditors too.

Yes, I'm going to make an attempt at getting .net working on Windows Phone. I'm in a good position to do it.

But yes, post it up, I doubt there are going to be any other takers for Windows, but probably for Android. I think Terry Pearson from that G+ thread this morning is planning on working on it. I directed him to our forums for discussion.

I want to get dotbitns.com set up too; I need a network wide trustless resolver for my own local network so that All connected devices can surf dot bit, and so that my mail server can deliver dot-bit email. So I plan to develop a DNS server extension for windows server so that network admins can deploy do-bit on their LANs.


Michael W. Dean,
MeowBit CAO (Chief Anarchy Officer), cat herder, worm wrangler.

Our Forum's developer discussion of this topic is here:
http://forum.namecoin.info/viewtopic.php?f=2&t=1710

[domob]

I fully agree that light clients are necessary. However, I'm not sure what you propose (blockchain in a Torrent) is what we should do - actually, IMHO it solves a different problem: That of blockchain size in general. It could be useful to avoid any nodes storing the full blockchain at all if it was distributed in a DHT fashion, but at the moment that's not a problem.

As for light clients, there are multiple ideas already on the table. See, for instance, https://forum.namecoin.info/viewtopic.php?f=5&t=1388. One potentially relatively simple idea is to export the name-value database in some basic format, distribute it somehow (could be a set of centralised servers or via the P2P network itself) and use the the merkle trees of blocks to verify that the database obtained from somewhere is correct in a trustfree fashion. I think that this could be more or less easy to implement. Note that this wouldn't cause a hard fork of the blockchain.

The "better" proposal is to switch to tracking the UTXO set directly without separating the names from other coins. This seems to be harder to implement and more complicated, but would be a better solution going forward.

[MWD]

As for light clients, there are multiple ideas already on the table. See, for instance, https://forum.namecoin.info/viewtopic.php?f=5&t=1388. One potentially relatively simple idea is to export the name-value database in some basic format, distribute it somehow (could be a set of centralised servers or via the P2P network itself) and use the the merkle trees of blocks to verify that the database obtained from somewhere is correct in a trustfree fashion. I think that this could be more or less easy to implement. Note that this wouldn't cause a hard fork of the blockchain.

I like this. I think the usable addition from what I wrote above is this part:

Including a function of checking and not bothering with addresses with no associated website for 3 out of the last six days or something like that.

Most of the squatted domains have no associated website, not even a placeholder. Which is annoying, because people trying to even buy those domains have no way of finding who to buy them from. A few put an e-mail address in the blockchain, but most don't.

This is actually quite foolish of anyone trying to squat and sell domains, they have no way for a buyer to buy them. That's their tough luck, and their stupidity could be an advantage for actual usage of Namecoin. I think a way to make them lose interest in squatting would be to have a common system that simply ignores or doesn't download domain info for domains that don't resolve to anything.

MWD

[domob]

We could do that, but I don't think it would help. Even the full name-value database shouldn't be too big, and if you download only partial databases (where the set of names may even depend on the current network situation, i. e., an IP may not resolve for someone who's network is currently partially disfunctional or whatever), then there's no clear rule while database would be included in a block's merkle tree. If you format the database in a specified format and include every name, then in theory each full node can reconstruct it in such a way that it can prove to a light node that it is valid, using a current block's merkle tree. Thus every full node could be extended (without a hard fork) to handle requests for the current name-value database by light nodes.

Also, if the full database is provided, it enables mobile clients for other things like id/ or file verification as well. I think that's a huge benefit, and outweights the saved space. But I have not checked how large such a full database would be, maybe it would be interesting to find that out. On the other hand, we can probably ignore expired names and all history of names - while that may have applications (timestamps, potentially file verification), it is definitely non-standard usage to rely on a name's history, and light clients need not support that, IMHO.