Proposal for a new naming system (ELNS)


I am working on trying to develop a new naming system for IPFS. Here is the abstraction:

“Electric Light Name Service (ELNS) is a proposed utility to resolve IPNS hashs by mapping them
to human readable words. The mapping will function similar to the way DNS maps IP addresses to
domain names. However, ELNS will have one major difference, ELNS-names are not chosen, re-
quested, or purchased by the participant. Instead, nodes participating in the network will be awarded
a single randomly chosen ELNS-name. Nodes will then be required to participate on the network
to maintain their name. The core idea behind the system is to reward participation without financial
compensation or require participants to pay into the system to purchase names. We also want to
keep the barrier to entry as low as possible to allow an even playing field. This system should allow
anyone with a computer, able to run IPFS, to participate and be awarded a name.”

Please comment, shred or add to my idea. Link to full paper: /ipfs/QmeCw9s4iYAmmgB591QDwYhwqkttav9pu4hxJcEw5H5CcU

Better link for paper here seeded on cloudflare:


I’ve only read what you posted here, not the paper as a whole. I’d really like to know what you think the added value is for what you propose. I could be completely wrong but if i’m reading the quoted part correctly then you end up with a random - but persistent - name. How is that any different then a IPNS CID?

Also, in terms of the IPFS stack this ELNS would lay on top of IPNS (according to your quote). So the chain from IPFS CID to ELNS resolution would be like this:
→ IPFS CID (Qm…x) → IPNS CID (K…x) → ELNS random name (rand…om)

In strict technical terms the IPNS cid isn’t random (it’s a hash wrapped with some metadata to become a CID). But it is persistent!

It looks like they’re going for the “human meaningful” leg of Zooko’s triangle.

I guess the use case is something like: I follow a nice blog on IPFS. I don’t want to remember the IPNS key, obiously. Now, I can just remember : “en.beetle.argyle.resay.elns” and be fine.
Or: I work on a collaborative file with a colleague that is often updated. I can now just remember the ELNS address and type it in any (IPFS compatible) browser from any computer.
Something like that.

Exactly, trying to remember a bunch of CID’s is not human friendly. Not to mention sharing or promoting your site.

Also, think about something like a distributed database or some other type of app that requires some kind of common storage. You could do this with an ELNS address.

My 2 cents…

Personally, I would not seek to participate in a system which randomly assigns what amounts to a domain name.

The purchasing of domains is an important component of participating in distributing your information whether that be via IPFS or legacy Web. Regular domain names can be purchased for $5/year, and with _dnslink TXT records can be integrated quite easily with IPFS storage.

Secondly, there already exists a blockchain domain naming architecture, ENS.

So, the proposed ELNS doesn’t provide any new solution except randomly assigned names… which are not ideal… along with a requirement to “actively participate” in order to maintain that randomly assigned domain… doubly not ideal. I have several domain names in both ENS and DNS which are not actively used. These are not for hording or selling. I use them on and off for test purposes or running temporary data distribution in a convenient way.

In short, I don’t see the proposed ELNS as being useful for any problem that I currently have. Nor do I feel the proposal would have a large user base.

The biggest issue I’ve had with IPFS is the reliance on the public gateway system. And the proposed ELNS still relies on the public gateway system.

1 Like

On first look I tend to agree with @ipfsme .

@elns Could you explain what the key difference between ELNS and ENS would be and what use-case ELNS would cover that is not covered by ENS?

My first impression is I like the problem it’s trying to solve but not quite sure about the way it’s trying to solve it. I generally agree that it would be nice to have something more memorable and human readable than a CID. You have the use case where you want to share something with a colleague but you’re probably going to just email it to them so how is using an ELNS any easier than just copy/pasting the CID? If I take the CID out of context of the email it’s impossible to know what it’s referring to but is apple-bubble-head-puppy any easier to associate with what it’s about? I can see it being used for communicating over the phone but how often is that going to happen? In this sense it seems similar to a url shortening service.

I’m not sure I’d agree with the premise, “As in life, you don’t get to choose your name.”. Some people do choose their name. Actors and entertainers do it all the time. Reginald Kenneth Dwight => Elton John. Prince Rogers Nelson => Prince => The artist formerly known as Prince. People do it all the time when they get married.

I also don’t agree that “DNS names are valued mainly for their Vanity”.

I say cool, go for it. It would be interesting to see if people find it useful. I kind of feel like it hits an unfortunate middle ground of, “crappy names for everyone. We can’t find a way to fairly distribute names so everyone gets a bad one”