Hacker News Reader

Comment by pmlnr | Hacker News Reader

The whole decentralize everything! movement seems to overlook DNS. I get that technically the DNS system is decentralized, but in reality, you lease a domain from a random, usually politically charged entity, who can do anything with it, if they wish.

We need a real DNS system, one where an individual can request and have a domain for life and which is truly decentralized.

Unfortunately none of the attempts - .onion with tor, .bit with namecoin, etc - seem to be working. .onion is despised because tor and impossible to memorize them; .bit never gained any traction and namecoin, being a bitcoin clone, has it's own issues.

Anyone knows of any working approach?

View on HN

> one where an individual can request and have a domain for life

Annual payment is proof that you're still alive.

AlterNIC tried settint up alternate tlds years ago, and is now gone. The problems are an alternate root is very unlikely to get consensus, and name resolution without consensus isn't very useful; and you're not going to be able to have friendly names without an arbiter to decide who gets pmlnr.example.

> Annual payment is proof that you're still alive.

I don't want my domains sold after I die, and I don't want my bookmarks to stop working after someone else dies.

The only parts of a website that I want tied to someone's lifespan is:

- whether or not the site is updated, and

- how the site is hosted

If I could snap my fingers and magically figure out a way to make the site's hosting outlive the person, I'd do that as well. For me, extend GP's original statement to read

> one where an individual can request a domain and have it removed from the market permanently.

Your vision sounds pretty terrible. At a very cursory glance it seems fine - I don't want my information web to degrade.

Once you dig even a little bit into this, though... you get a quagmire of old and useless information. Outdated sites never go away. They just become a constant burden on the whole system. Information that's been refuted and invalidated years back is still alive and kicking because a domain exists and can be indexed, and there's no chance to ever update it, since the owner is long since dead.

The longer this lasts, the more and more noise to information you get. It's like having an incredible information web and then giving it dementia.

I think it would be absolutely incredible to peruse old blogs and websites from a century ago. Clippings from 75 year old newspapers or letters are interesting for plenty of reasons. The cost of things; The news of the day; The writing style. Imagine direct access to this sort of content from 750 years ago!

Someone's going to have to pay for hosting that data, though. The domain name is only a small part of the picture.

That's why we have wayback machine. Or we use another way to archive it.

The Wayback Machine is great, but it's basically a hack. Archiving shouldn't depend on a single centralized entity occasionally crawling the web and saving chunks of it to its archive (but only what it finds during the crawl, and excluding content with large file sizes, such as videos).

It ought to be built into the architecture of the Web, decentralized, immediate, and (at least for small file sizes) on by default. Oh, and censorship-resistant. Even for large file sizes, I think there ought to be some very easy-to-use mechanism to donate either hard disk space or money to publicly archive content of your choice.

Those are lofty goals, of course, but the current web has is quite vulnerable to bitrot as it is, and there's no guarantee the Internet Archive will continue to operate indefinitely.

> Archiving shouldn't depend on a single centralized entity

It doesn't. It's decentralized, with lots of archivists, it's just not federated.

Are there any other Web archives with scope comparable to the Wayback Machine? I have not heard of any. I guess there may be private archives which are not publicly known or accessible.

As the Wayback Machine currently operates, the present owner of a domain name can make the archives go away.

OpenNIC is still going strong and has alternative TLDs. Of course, the problem is that most people won't be able to resolve your domain.

Remember, there only 2 real hard problems in computer science. Cache invalidation and naming things.

P.S. I'll add, that on the issue of a personal identifier, we need to remember that not everyone wants a public address. In fact most people keep this kind of information quite private. A virtual address is no different, unless we explicitly make two kinds.

Handshake [1] does this exactly.

[1] https://www.handshake.org

How is this better than namecoin and how will it not end in same problem?

Also this really does not do what the person you responded to asked for - I can't get some personal domain name for life from this - it is market based AFAICT. Really seems to solve almost none of the problems and goes directly for what ICANN only now does with .org as no domain will have any price cap with handshake.

> I can't get some personal domain name for life from this - it is market based AFAICT.

The yearly renewal fee is fixed, and is a standard transaction fee, not market based [0].

[0] https://handshake.org/faq

Claim from the site you linked to is not the same as your claim:

> Renewals for names are annual and cost a standard network fee. Miners will receive the transaction fee as part of their block reward.

This could mean:

- Fixed USD fee (it is not this one)

- Fixed coin fee (don't think it is this one either) - but coin price is marked based - so fee is market based

- Price does not depend on domain name but miners basically decide the fee like it is with other cryptos (think it is this one) - still market based - and actually could be gamed to make it higher for some domain names

So ...

Can you explain how the miners can price discriminate against certain domains? Seems like the network fee would need to be identical for every concurrent transaction.

I don't know exact details, but if it works like bitcoin miners can refuse to put renewals in blocks if the fees is not high enough.

In which case you can just wait until there are less transactions waiting to be confirmed, driving the price up (it's a renewal, so you have plenty of time to complete it).

You can't wait to renew, or someone who's willing to pay the price will steal your domain.

I'm pretty sure you have a time period during which you can renew the domain.

Namecoin is a great project. One of the issues however is that it introduced a .bit which essentially meant it was a new system/tld. Furthermore, the existing stakeholders in the naming system (domain owners) did not have an easy way to join the system without risk of losing their name or being sold their own name at a high price.

Handshake mitigates this by replacing the root zone entirely and, in addition, allowing existing stakeholders (.com, etc) to continue to own their existing assets.

I read buzzwords - proof-of-work, coins. Also seems to be venture funding based.

What's the catch?

> We need a real DNS system, one where an individual can request and have a domain for life and which is truly decentralized.

I very much want this also - but this cannot be solved in the same way as the challenge of business domain name assignment - a different approach should be used:

[idx].[yyyyddmm].[given_names].[family_name].id

e.g.

023.19830210.john_smith_3rd.doe.id

given_names can have some standard seperator

And then if there are two people with same given names with same family name they get different indexes. And this will get everyone riled up because this means you will need a worldwide consistent database of people and you will only be able to get this if you give very good proof of ID and then this because your world ID number basically.

That bumps directly into "fallacies programmers believe about names". Not every culture uses family names the same or has them at all, for instance.

See also Falsehoods Programmers Believe About Names: https://www.kalzumeus.com/2010/06/17/falsehoods-programmers-...

Unless someone goes out their way to choose their own 'branded' name perhaps they could just receive a unique hash of their name for a domain? Tools can always be built to handle these names so you're not memorising such lengthy domains.

How do you decide which person get which name? This just gets you back to the whole ICANN problem. I can remember a couple of digits which makes my name unique. I cannot remember 1e873645-1f68-4b48-9eae-934ec717229b

That's just a technical problem that new infrastructure can solve by abstracting away the name. Instead of visiting 1e873645-1f68-4b48-9eae-934ec717229b.com/foo, you're just visiting "bar/foo", where 'bar' is their name. Click on the URL and the full hashed version is revealed for proper sharing.

In the same way that nobody memorised Facebook's UUID of a person, they can just remember their name.

And then we are back to the problem of deciding who gets what name

But then .onion - or at least the way it's done - is already there.

During the socialist system one received a personal identification number. It was very similar to this.

The good side:

- I like the idea and I have been thinking around the same lines

The bad side:

- how would changed names, like marriage, be handled?

- it shows too much PII - I believe knowing a domain like this would immediately fall under GDPR

- still not easy to remember

- doesn't allow pseudonyms - I know that pseudonyms might look like they go against request a domain for life, but they don't. In our culture, are name is given by someone else and/or inherited, but many would like to associate their presence with something they decide on their own.

> how would changed names, like marriage, be handled?

The same way it works with NI numbers in the UK and SS numbers in the US, you tell the agency about the change and that's that.

Why would a number being associated with your person have any issue about your name changing? the very purpose is to disassociate the name from being a unique identifier.

Norway still uses a very similar system and I think other countries also.

You do raise very valid problems though - but I feel like either you have at least some of those problems or you go to routes which is basically what .onion does where you have your_chosen_name + bit_of_not_memorable

Simple, don't have reserved names!

Nobody owns our language.

Martti Malmi, Satoshi's first Bitcoin contributor, solved this several years ago:

https://github.com/irislib/iris/blob/master/README.md#identi...

Another approach worth keeping an eye on is the Ethereum Name Service (ENS): https://ens.domains

Personally I find tor to be most usable out of the approaches you listed, but obviously not for memorable names.

Another project to keep an eye on is OpenAlias https://openalias.org/

DNS is probably one of the most popular subjects of the decentralize everything movement. Not that any of the solutions have really caught on though.

Reply on news.ycombinator.com