Do we need more top level domains?

This article makes a few good (but separate) points.

1. Hostnames are backwards. com.ycombinator.news makes much more sense than news.ycombinator.com, because the rest of a URL goes from least specific (protocol) to most specific (path components, query, fragment). The only case where the current form of hostname works is email addresses (user at sub dot domain dot tld). IIRC, one of the founders of the WWW (Tim Berners-Lee, perhaps?) already admitted that they got this wrong. I'm all for fixing it, but how? There would need to be easily machine- and human-identifiable differences between old- and new-style hostnames, or there's massive potential for confusion and collisions. There's also an unthinkable amount of software out there that understands hostnames in their current form.

2. The TLD system is broken. Hierarchical hostnames make great sense when you want to refer to the host foo, owned by building four of the computer science department of Caltech. The TLDs we're using today are hacks, a valiant attempt by ICANN to conserve the domain namespace. I hate to admit it, but I think that registering TLDs based on trademark really could work. It'd mean the end of fun websites like http://unicodesnowmanforyou.com/, but would bring a huge amount of sanity to the top-level namespace. It would create to a huge new market for private subdomain reselling (whoever gets .www is going to make a killing) where resellers would have the power to be selective about registrations, and to pull domains out from under squatters. Spam-heavy TLDs would lose credibility.

It makes today's internet feel like the Wild West.

As just one example, who gets "era", and why?

http://www.era.com/ ?

http://www.era.mk/ ?

http://www.era.hk/ ?

http://www.era.pl/ ?

http://www.era.ru/ ?

I am starting to believe most new TLDs are simply a scam by registries to force trademark holders to re-buy their trademarks every time. They serve little other purpose.

The author is reiterating the complaints that major trademark holders raised over a decade ago during the initial generic TLD (gTLD) expansion process in the mid-1990s (see Milton Mueller's book "Ruling The Root" for details). More gTLDs meant the need to purchase one's trademark in each new gTLD in order to protect that trademark. The power that trademark holders/WIPO had in this process lead to "sunrise" periods where established trademarks were allowed to be registered before the registry opened generally, and I've heard claims that the dispute resolution process can tend to favor trademark holders too, though I can't vouch for that.

The problem with the claim that trademark interests trump all others in domain name disputes is that non-domain-name trademarks must be distinctive and are typically for a specific market. To use the example already given in the comments, Apple Inc. the computer manufacturer and Apple Records the music label can coexist because they aren't really in the same market and won't confuse consumers as such. However, domain names don't allow for this level of subtlety, and so whoever registers apple.com first with good-faith intent "wins", despite justified interest from computer, music, and fruit companies.

While this attitude towards new gTLDs makes sense from the perspective of existing second-level domain/trademark holders, what about those who aren't established yet. Is the person searching for a short or meaningful domain name out of luck because trademark holders are sick of new TLDs? There appears to be demand for new TLDs, and it's been shown that the root zone will scale, so it would seem to make sense to allow new TLDs for those whose needs aren't met by .com et al.

The deeper problem in all of this is that we interpret meaning in these unique identifiers, and use that meaning as part of branding and to judge authenticity of a site. If (for example) websites were identified by the public key of the certificate that signed its contents, and if we navigated the web using a directory system instead of typing domain names into browser URL bars, we wouldn't have this problem.

I've been saying we need a .app TLD for years!

If they fucked the squatters and parked domains off then we'd probably never need another top or other domain again.

It's just stupid trying to find an unregistered domain now.

You could argue that the main problem with domains is their exposure to end users; they are essentially an implementation detail of a protocol. Since they are relatively simple to type on desktop computers, they've worked well for a long time. Now, finally, with touch displays, they may be called out for being too cumbersome.

The problem to solve is not the look of domain names, but authentication. How do you know that you've reached the "real" web site for a product or service or group? If you could know that in some other way, you wouldn't necessarily need to see an ugly domain name.

I could imagine something much more interactive. For instance, an app that lets you first specify your favorite products and services, that can then pop up a grid of touchable corporate logos to bring you directly to the correct web sites (no URLs needed). The existing DNS registry could be used to either figure out what you need, or give you a short list of choices. In time, apps like these would become so standard that you wouldn't "need" to register all permutations of a trademark, because they would already be choosing the correct one.

I like the idea, I hate how he uses 'ridicules' instead of 'ridiculous'.

Why don't we just raise the price of domain names? If you doubled the price you'd make a lot of domain squatting unprofitable.

How is it that you need baekdal.dk but not baekdal.cn? Don't they pose the same security threat when it comes to phishing? It's really not necessary to own 20 different tlds to protect your brand. How many people mistakenly go to baekdal.biz?

Definitely agree. Java's packaging does it right. Unfortunately large changes like this are just not going to happen without some drastic event. Changing around the order of TLDs would be no easier than switching the US to metric system.

We already have a system where anyone can buy any name they want. It's called .com. If we open up the root space as a free-for-all, then we will never be able to extend the system in the future. We can only do it once.

This is interesting, because it seems to solve the problem of having to buy dozens of .tlds to secure your brand name.