RFC - Decentralised Naming System V - No DNS

Here’s a topic to discuss a decentralised naming system with no naming system at all - only native addresses, private bookmarks, curated collections, webs of trust and search engines.

Here is how it would work: in order to create a website you

  • generate private/public keypair
  • build website.tar.gz and sign it with private key
  • upload website.tar.gz + signature + public key to SAFE

From this moment the 16 character hashcode of your public key (say 6sgjmi53igmg7fm7) becomes the “native” address of your website (“key” in Zooko triangle). If you somehow communicate that to your users they will be able to bookmark it under a name of their choice (“petname” in Zooko triangle) and access your website. No DNS. No domain name.

Now suppose somebody somehow gives you an address like 6sgjmi53igmg7fm7. How do you know if that is a legitimate website or a fraud? That is where webs of trust (example) come into the picture. The trust will be linked to 6sgjmi53igmg7fm7 not to some human readable name.

You may ask what other ways will I have to discover websites? Links on webpages will be of the 6sgjmi53igmg7fm7 variety. So by following links on blogs and web-pages. Also hopefully people will start putting out curated collections of links (example) and search engines will emerge. Webs-of-trust (I hope there will be more than 1 of them) will be an invaluable source of info for the search engines.

BTW each 6sgjmi53igmg7fm7-style website will probably offer a self-name like amazon or wikipedia. That name (“nickname” in Zooko triangle) is merely a suggestion for your bookmark name and a hint for the search engines. Nothing more. No claim of uniqueness.

Unsolved problems: Paper Napkin. Also what do you print on an advertisement billboard?

Update: perhaps billboard could point people to a page on the regular Internet which would in turn have a link pointing at a SAFE native address ( 6sgjmi53igmg7fm7). So we’re not killing of WWW we’re turning it into landing pages for SAFE :slight_smile:


no branding. no outsideworld bitching. no squatting. same treatment for everybody. no greed. no favors. no run for registering fancy addresses. it will be all about the content that the site provide, not about the brand. just like a phone number, you don’t have to memorize it. you keep it bookmarked somewhere. it will be the same for future generations, without the stress that ‘easy and good’ names are already taken. hard to advertise (i personally consider advertising a bad thing, because if you truly offer good services, the customers will find you, not vice-versa). freedom lovers will love


You can always use a url shortener for the paper napkin problem, which is essentially another form of bookmarking.


I’ve been coming to this site for about a year and it seems like its been forever. The above posts by Catbert and Bugsbunny have to be close to my personal favorites. I was thinking along the same lines but couldn’t begin to formulate how or really quite why. I mean literally thinking tiny pieces of the above “like a phone number,” but no clue how to articulate or name even step one. Love this idea, really hope it sticks! Completely agree about advertisement and why it doesn’t matter and why we need to get beyond it and the understanding (related) that true fully end user oriented search makes it obsolete.


Although I am a bit confused as to storing a website in a tar.gz archive.

  1. Wouldn’t that make the website exclusive to linux as windows and mac use different file encryption protocols?

  2. Why compress the website in the first place as everything on SAFE is just transmitted using pointers in the first place?

1 Like

Details that don’t have much to do with DNS, but important nonetheless.

1 Like

Okay but I’m still confused.

Hi, that was just a simplified example. Though Firefox could unpack a .tar.gz and show you the content on Windows, there is no technical problem here.

A pointer needs to point to some content right? And that content first needs to be uploaded. 6sgjmi53igmg7fm7 in my example is in fact a pointer and I explain how it could receive its associated content.

1 Like

I also would like to see something different than the current domain name system. It is so annoying when you want to register a domain name and find that all the useful ones are taken. It´s easier now when there are more TLDs, but using a Key address would level the playing field. As @Happybeing suggested there could be apps/companies that provide short address links that can be written on a napkin, e.g. the app Shortcuts with the printed address Shortcuts/MyBusiness.


That would imply randomness instead of domain names. A traditional DNS system allows users to choose their own name. Theoretically the domain address can be down to 10(?) characters. It’s just not customizable or human memorable.

TOR hidden services are currently using 16 character strings A-Z,0-9. Indeed writing 16 characters on a napkin is doable if you carry a bit of paper or a mobile with it around. You could also print it on your business card.

1 Like

Freenet is using all these very big and ugly addresses no one can remember. TOR is doing the same with all internal .onion sites. It’s awful! Thanks to people like David and the team, we finally have a decentralized system with easy to remember addresses. Just register your own private spot on SAFE Net and tell your friends. It should be that easy. Gotta love the DNS. It’s a great thing they’ve even implemented it before the network is live.


Also, not using specifics, you can always ask the people around you if they have a site that suggested “Google” for the petname. Whatever they changed it it, they’ll be able to return the ones that originally presented themselves as “Google”.

Choosing the most popular one would statistically result in the best one. Therefore, may the best app win.

Where has this been implemented? If it already has been, then why was @dirvine so interested in these discussions?

Hmm… the issue is that the network has got 23,000 sites presenting theselves as Google, isn’t it? There surely are clever ways to find the real Google among them. People could share their bookmarks, we can have web-of-trust services similar to mywot. These and perhaps other kinds of data can be harvested by websites or apps doing the job of search engines. These I hope to give the job of giving me the best match to Google or Amazon.


True, but Freenet And TOR don’t have integrated petname sharing, which would be the game changer.


I think the team is open to all ideas. If 80% of users would scream out loud here on the forum that they hate the DNS, they probably will delete is from code. But I think that over 80% thinks it’s brilliant, that’s why we should let it in :wink:

But not a one-to-rule-them-all DNS. Sounds good to me.


I don’t see why we can’t have both. I mean assume for a moment we have tor like random no dns as a default. Then beside it you have some dns system. You might have a couple dns systems. So you use a DNS system to get a url of some sort, you can do the same for an onion like address from the no-dns system Then if you can’t get the url you like from the dns you can still just forward to your tor address and use either or. www.myreallylong.annoyingand.cheapur.safe works just as well as a placeholder for tor links a 16 character string would. Or say you have a really inappropriate url like www.kinkycleaningservices.safe or something. Might be good in SOME situations but you might not want to advertise that in mainstream public so putting a tor link on a napkin might be a good idea, be discrete and such.

Ultimately as discussed before a DNS is basically just a long set of bookmarks. So really how is it different than taking the tor links and assigning urls to them? It’s basically a bookmarking service.


Web of trust is the basis for distributed search and a way to get around many DNS issues?

Sorry, not really technical, just want to understand what I can. I have this sense that search will be the killer component or app because search will always have that spot.