The end of a single global namespace? | Hacker News Reader

The end of a single global namespace?

christopherkullenberg.se

52 pointsunicornporn23 comments

China has already added some Chinese-language top-level domains to their DNS namespace.

The underlying problem here is Zooko's triangle: nobody has demonstrated a global namespace that is, in Zooko's words, secure, decentralized, and human-meaningful. Centralized systems inherently have central points of failure. Approaches like dot-p2p and Telecomix do not tackle this problem and will therefore not improve the situation.

There are a number of projects trying to tackle the problem by means of petnames, using non-human-meaningful identifiers in the global namespace (so they can be secure and decentralized). Git is probably the most prominent, but others include Freenet and Tahoe-LAFS, and there are quite a number of similar projects that are not yet public. Unfortunately, none of them are currently a replacement for the World-Wide Web, a problem I wrote about at some length in this unfinished essay in 2006: http://lists.canonical.org/pipermail/kragen-tol/2006-Novembe...

Also remember that China's .com doesn't look the same as Verisign's.

That would be about time. The DNS/ICANN system is by far the weakest point of the internet. First, I don't see why you have to pay to own a domain. Second, what's the point of having tld associated to countries? Obviously everything behind DNS/ICANN is just there to leave government the chance to track people and shut down services. Time to get rid of this.

What would you replace it with that doesn't have the same weaknesses? DNS already has a robust distribution model (as long as most TTLs are set reasonably high). Its main weaknesses are trust (in the crypto sense, i.e. that lookups are reliably authenticated) and policy (it should be inconceivable that any person or agency could tamper with an established domain name; after all, who fights crime by erasing a suspect's name?). I agree that the current system is being mismanaged, but transferring authority to another entity doesn't necessarily address the inherent weaknesses.

BTW, I'm with you 100% on the madness of TLDs. Sure, reserve some for specific uses, but there's no reason you shouldn't be able to register an arbitrary string as a TLD, and do it cheaply (except you have to solve the problem of domain squatting).

A decentralized/p2p model would help. You could think of a system in which you register a domain associated to a GPG key and you need the key to take it down or change DNS record. This way you get the cloud resolving addresses and you avoid the obvious risk of DNS hijack.

What about the squatters who would make their own keys in the initial rush? There would have to be a central authority to bootstrap this. After the initial bootstrap, the free market could take over. (I'd trust Google or Amazon with this, especially if they knew users could switch over to a competitor if they were P.O.'d)

The purpose of charging fees to own a domain is to ensure that they are allocated optimally. Sure, fees are a crude way to achieve this, but they are also a relatively inexpensive and they generally work. If domains were free, squatters would just start registering names in sequence. Personally, I feel like the fees are too low and encourage speculation as it is. They should go back to the $100/year levels for second-level domains. Perhaps they should open up some very inexpensive ($1-$2/yr) tertiary domains.

This could be worked out by charging fees only after a certain number of domains owned.

For instance:

1-10 domain, free

10-100, $50 each

>100, $250 each

With a control on credit card or something unique enough. Of course the problem is: who is collecting those fees?

The credit card companies will charge you a fee to use credit cards as an authentication mechanism and using prepaid gift cards can get around this. You can buy prepaid cards in bulk. Other authentication schemes are too expensive. Even if you could make it free, it still doesn't keep people from indefinitely registering domain names they don't use. I might be convinced to along with free domain registrations if it wasn't a finite resource, but it is a finite resource. If I have a legitimate reason to use and own a domain, it's worth $100/year. Otherwise you end up with our current situation where $5/year domains have made squatting a feasible business model.

Well, then let's find a compromise. This new more anarchic system could come in parallel to the already existing one.

If you are a business or university you still follow the .com and .edu rules. If you are a broke student who doesn't want to pay for a domain or someone who is afraid of getting tracked/shutdown you can join the decentralized system. I doubt squatters would be too interested in getting those.

I suppose the .p2p thing they mention in the article must be something similar as a concept. They should just get another extension though because p2p sucks

Cool! Why should amazon.com be the only ones that get to have the domain amazon.com?

The reason domains cost what they do is that providing them costs (with the current infrastructure / policies). The current cost price, for the registry, to register a domain is anything from $3.50 (for TLDs with 10M+ registrations) all the way to $20 (for TLDs with ~50k registrations). A SRS, a backup SRS, all the extra services (thick Whois, support, billing, backups, administration, auditing, legal etc) don't come for nothing. Obviously scale makes a big difference to price.

Where are your cost numbers coming from? Ultimately, the cost of maintaining a domain registration record is a small text file in a filesystem somewhere, plus glue records copied into the parent domain. I don't know of any filesystem that costs $3.50 to store a text file under 512 bytes, even including backups and filesystem administration. Accordingly, there are non-TLD domains (e.g. dyndns) that let you register any number of subdomains at no cost.

Of course, support can cost any amount whatever. But it generally doesn't cost very much. Mike Lawrie handled the entire South African .za domain single-handedly from 1994 until 2002 in his spare time.

The GTLD servers also bear a cost of answering queries, but that cost is zero for domains that nobody queries (most of them!) and in any case is not paid by registrars or registries.

Let me offer an alternative hypothesis. Domains cost money because they provide a valuable service, and competition in registry operators is nonexistent by the nature of the DNS. To register a subdomain of a particular parent domain, such as .com., you must pay whatever the registry controlling that domain chooses to charge, either via an intermediary such as a domain name registrar or directly.

Because subdomains of popular domains such as .com. are valuable and there is no competition in the market for them, you get to pay monopoly rents, which have fallen somewhat since ICANN wrested control of the GTLD monopoly from Network Solutions.

Internauts... I think everyone should start using this word.

It's already common language in French: internaute. News anchors use it frequently, and there's 7,380,000 results on Google. It sounds a bit cooler in English though.

One approach to this problem is called the Handle System.

http://www.handle.net/

It's like DNS except individual identifiers can be managed independently (imagine if example.com/foo and example.com/bar were managed by two different companies).

One problem that it does not address is human-friendliness. Identifiers all start with a handle prefix that's a number like 10.1109/ or 10000.1/.

Though, the latest release of the software will map DNS names to handle prefixes.

It's interesting technology and quite robust. For $50 a year you can host as many handles as you want.

Is there any reason we can't create a new root.

Then have lookup servers who have both roots. So the lookup server sees 'example.com' and checks ExampleNIC. ExampleNIC doesn't have 'example.com' and so the lookup server checks ICANN Root. If the lookup server gets 'example.other' it checks ExampleNIC and continues.

This should be able to scale to a large number of alternative roots. If you move between root domains you will have to get new domain names but no one power controls the DNS system.

People have been trying this since the beginning of DNS, especially back in the 90s when domain names were expensive.

Unlikely. Most likely we will end up with something like Freenet. I can't decide if that is a good thing or not.

But you already decided: "most likely we will end up with something like Freenet"

I re-read tomjen3's comment 3 times and couldn't find a value judgement, only a prediction. Where is the value judgement?

Freenet is known as not exactly your ideal freedom-of-information situation, with most content reportedly being child pornography.

Ah, the implicit syllogism is "if rampant child porn and neo-nazi content is what we'll eventually end up with, that's bad."

It was too subtle for me, but it didn't deserve downmods; I'm upmodding it.

Reply on news.ycombinator.com