Hacker News new | past | comments | ask | show | jobs | submit login
Tim Berners-Lee: we don't need arbitrary new TLDs (wired.co.uk)
89 points by leejw00t354 on April 20, 2012 | hide | past | favorite | 71 comments



Making the root domain a free-for-all effectively promotes what .com was to the global root. We can only do this once, and then the root namespace is gone forever.

We will never again be able to create new top level domains that have special restricted meanings, such as .arpa, .gov and the concept of country TLDs[1]. Any sensible name that we want to use will probably already be taken. Even it is not taken, a new special purpose name would never be able to differentiate from squatters using the same level of the namespace.

Not only that, but the special casing of existing TLDs such as .gov could get diluted to the point that the majority will not recognise them as official any more.

There is a place for free-for-all on domain name registrations, and it is .com.

[1] I presume that they will reserve two letter registrations for the specific case of new countries coming into existence, but my point is about the concept of new uses, rather than this specific case.


If only we had gone with the other order for domain names, that is com.disney.www instead of www.disney.com.

Then browser TLD defaulting would have handled most cases and we could have had additional conventions (such as *.store). (The "www." subdomain is basically useless.)


In that case there is still a TLD. It's probably worse because the differentiating part of the domain is buried halfway into the domain name. Imagine having domains like com.disney and com.donald. They look the same.

The web has no hierarchy so it doesn't make sense for domain names to have one. It would have been nicer to ditch the TLDs, use a sane protocol name and have implicit slashes. Like web:disney/some-path.


> The web has no hierarchy so it doesn't make sense for domain names to have one.

Actually, the "web" does have lots of hierarchy.

For example, Disney has hierarchy that the current scheme makes it inconveient to express. That's why we have disneystore.com, disneycruise.com, etc. instead of com.disney.store, com,disney.cruise, etc.



The web may not have a hierarchy but URLS do - but it doesn't work as well as the earlier X25/x400 version. So we have to have an address that increases left->right, then switches halfway through to a different format decreasing left->right

www.group.department.university.edu.country/person/topic/page

Instead of the more logical

country.edu.university.department.group.person.topic.page


> (The "www." subdomain is basically useless.)

Tanguy Ortolo answers 'Why www?' better than me.

http://tanguy.ortolo.eu/blog/article47/why-www


And who would control the .www domain? US?


> And who would control the .www domain? US?

The same as today.

In www.disney.com, disney controls that www. In com.disney.wwww, disney controls that www.

However, in the latter, disney could do com.disney.store . And, with browser defaulting the tld, disney.store would work.

Disney isn't going to use store.disney. However, they might well use disney.store if they controlled it. With the current order, they don't. They would with the reverse order.


The Uk used to do that :-)


The US are claiming legal dominion* over all domains under a large number of TLD's (.com, .org, etc.). New TLD's may help mitigate that.

*http://www.wired.com/threatlevel/2012/03/feds-seize-foreign-...

"If you just add one character to the length of the domain name you have 26 times as many names you can choose from. There's no shortage."

True, but TLD's like .pepsi and .HTC are a lot easier to remember than random combinations of letter.


Purely anecdotal, but my experience watching mere mortals use the Internet has been that they almost always Google for the company/site they want, not blindly type <companyname>.com. I'd be shocked if Pepsi owning "www.pepsi" somehow changes this.


I wonder how chrome will handle it. Type in "Pepsi" and you get taken to search results that make google money. What if it becomes a valid URI?


It may work like with a custom host file, but with better UX. If you add something like domain.dev to your hosts file and just write it onto your search bar without http:// Chrome will do a search and while searching check if there's a DNS entry for that 'weird' domain, if there is it will show you a tooltip asking if you wanted to go to http://domain.dev.


You're probably right. But given the choice, isn't it better to go with something that's easier to remember, even if it won't matter most of the time?


pepsi.com or www.pepsi or site.pepsi or brand.pepsi. You've then just got a multitude of hostname parts to guess - which is more confusing.


Not really. If you don't know what you're looking for, you just go to http://pepsi/ and navigate from there.


I don't fancy hostname clashes.


If VeriSign or any of the other US registrars ends up with control over the new TLD, the US has de facto control over them anyways. All it takes is some government lackey with a piece of paper and VS will bend over backwards for them.


pepsi@pepsi.com seems better than pepsi@pepsi.pepsi in my opinion.


Yes, but you're not forced to repeat "pepsi". It could be alex@customerrelations.pepsi or inquiries@distribution.pepsi


there's no need for pepsi.pepsi as the domain, it could be pepsi@pepsi


There's a ton of legacy code and email software that will make that not work.

At best, it would be pepsi@pepsi. (note trailing dot).

My buddy used to be root@ws. but only a few MUAs and MTAs could handle delivering mail there, and always with the trailing dot.


You have to put the dot at the end to make it a fully-qualified name otherwise the resolver will think it is a hostname and add the local domain.


moreover, it would make no sense to own www.pepsi without also already owning www.pepsi.com


Why not? Some companies only own <company>.net or <company>.co.uk, without owning <company>.com.


Fair point. But www.pepsi looks broken to me. If you're going to decide you care about people who type addresses in by hand, it seems counterproductive to go against the conventions they're used to.

But if new TLDs become common I guess that will change with time.


www.pepsi does look broken, but http://pepsi/ is fine. We just need browsers to stop ignoring bare TLDs, like they do now if you don't put the http:// before.


This opens up more problems though.

For example how would the Google Chrome browser work (and search to a degree), does a user who types "pepsi" into their omnibar want to go to http://pepsi/ or search for pepsi?

It also adds confusion for webmasters trying to parses URLs on their website. When a user types pepsi.com/coke into their blog for example that link can be made clickable to http://pepsi.com/coke/ where it can't for pepsi/coke


The problem of detecting URLs with weird TLDs could be solved by using relative URLs (RFC 1808 §4) with the protocol omitted. It would put that vestigial // to good use—//coca-cola is a URL just as @cocacola is a Twitter handle.


Why do I suddenly feel like buying pepsi..


These will be ICANN domains, thus still under US jurisdiction.


I'm not a fan too, but there's one huge advantage of extending the TLD space - we'll get out of the mindset of mainstream audience that only .com/net/org is a 'serious' domain and maybe there will be no more ridiculous bids like color.com in the future.


Why do you think that will be the case?

When ICANN introduced .cat, .aero, .museum, .coop, and others many years ago, adoption has since approached zero.


There's something to be said about ".com" addresses...

When you place "pepsi.com" in a can , people know it's a web address. Now, if pepsi has its own TLD, how should they refer to their web site? Just "pepsi"? "www.pepsi"? "web: pepsi"?

I suspect they will reserve their TLD, and then continue to use "pepsi.com" as before, thus rendering these new TLDs useless at best, and noise at worst.


they didn't have problems with "go to our facebook page" so it shouldn't be a problem to promote "win a ticket to justin timberlake concert - go to our website - justin dot pepsi"


it's no worse than go to our website at justin dot pepsi dot com, but most people will still have a problem with it

if there's one thing i took away from my time doing front line support, it's that users don't enter URLs. if you tell them to find the address bar, they're lost. if you give them a URL to enter, they have no idea what to do regardless of whether the TLD is .com or .pepsi. people get to where they want to go by searching, whether they're searching google for a website or searching facebook for a page. the domain name system may have technical meaning, but for most users pepsi.com is just as obscure a code as 199.187.226.158


I agree. I can't believe how companies promotes facebook.com/brand instead of brand.com. It's perfectly fine to have a facebook brand page, but you should promote yours first!


Why? If you are sure you need a facebook presence, and not sure about your own site helping the brand, then you just push the facebook one. It's simpler. No confusion. One place to go to find out about the product.



these useless domains didn't have any commercial backup, .pepsi will be part of campaigns worth millions.


The main disadvantage of extending the TLD space is that it is a tax on existing domain owners.

If you register a .com, you have to be safe and pick up the .net, .org, .info, .biz, etc.


>If you register a .com, you have to be safe and pick up the .net, .org, .info, .biz, etc.

This is a rookie mistake. Nobody in the history of the internet has accidentally typed ".biz" instead of ".com" and gotten to the wrong website as a result.


The perception is caused by the fact that the other domains are jokes. What is .info? .biz? Viable alternatives but totally useless.

The bar for getting a new TLD should be that you're doing something new and different, not just opening up a new namespace. .name is one example where they're at least trying to innovate.

A new TLD like .arpa that's used for some kind of underpinning would be a good idea if there was a technical requirement for it.

Otherwise you're just cluttering up the root zone with garbage and making it even more complicated for people to know what a legitimate domain is.

The public suffix list (http://publicsuffix.org/) is already full of garbage if you want to see how bad it is.


do we need TLD at all?

It is not my idea, but really, should't we abolish the TLD in general? Instead of http://google.com, http://google.de, http://google.tv .... we would get just http://google .

The reason why we have geographical TLD is just political anyway.


The downside is that some countries have restrictions on names; in Australia, a .com.au must match up to a registered business, which is awfully useful to prevent squatting when you add trademarks into the mix. No such dice with the straight .com, .org, etc.

(I like that we have both "restricted" and "unrestricted" zones; I can go register beiberloveskittens.com, for example, but know that beiberloveskittens.com.au won't go polluting the .com.au namespace without at least someone going to the effort of setting up a business for it to be attached to.)


Allowing arbitrary TLD's is semantically equivalent to getting rid of TLD's (e.g. in your last example, "google" could be thought of as a TLD).


I haven't thought about it that way.

But when I try all the possible TLD's as a separate addresses, none works ( http://com , http://org , even http://xxx or http://museum , nothing works).

What is the reason... is it DNS servers that don't like this type of nslookup?


Are you sure its not the browser's own rules and not the lack of an actual website that address?

Try ending with a period.

Certainly http://dk. works. I'm sure there are others.


I don't think it's a technical issue -- I'm pretty sure I've seen a country that hosted something at the tld itself, but I can't find it now.


Good question.

I don't think we do. I think 1 global namespace would be enough.

Say we ditched TLD's and limited domain names to 10 alphanumeric characters. That would still be enough domains to give 450,000 to every person on the planet.

That's with just alphanumeric characters capped at a length of 10. Allow unicode and remove the cap and we would have plenty of domains to go around.

All in 1 namespace.

Edit: yes I understand that the vast majority of those domains would be nonsensical, however even eliminating 99.99999% of them would still leave enough meaningful domains to go around the world many times over.


Domains are, in the first place, needed so we don't have to remember ip addresses.


Yes you coudl just use somthing like a phone number to access websites.

I dont think you have thought this through who would rember that the local doctors surgery is XDFE123456!


that particular string is very memorable ;-)


Every particular string may be made memorable, but you'd have to memorize it. Instead of it being natural, as in: a natural extension of how people think about things.


Back in the day I worked for BT and had memorable mail addresses such as 018222211 (Prestel) and 80:BTG174 (Dialcom/Telecom Gold) at least the Telecom Gold ones where slightly memorable as the prefix BTG stood for British Telecom Gold.


What are these political reasons? I genuinely don't know.


ITU :-)


What a load of rot. You're better to advertise something generic such as 'visit the <brand> website.' or 'find us online', rather than a barrage of non memorable domain names and various social media urls. This is such an elaborate con.


Indeed. Why not just ditch user-facing TLDs in the next DNS overhaul altogether?

People don't recognize the difference between pepsi.com and pepsi.org and pepsi.com.hackers-taking-your-credit-card.tt

.org is interesting meta-data, but why not transport that data as part of a meta-data record and let browsers display it to users in a consistent way they recognize and understand?

The point of DNS is to map a human-knowable name to a network or provider. We not only don't need new TLDs, we frankly don't much need the ones we have. dot-com/net/country codes/etc and even 'www' and 'mail' are of dubious benefit to the average user these days.

Sure, it's useful for engineers. And we can keep that in a diagnostic window. But for user-facing concerns, simplifying DNS is the only conversation worth having.


What we don't need is a bunch of stupid TLDs that do nothing but make existing domain owners feel like they need to buy more.

What we really need is a single new TLD that is distributed so that it cannot be controlled by any particular government or corporation. Something like dot-bit.org, but with more momentum behind it.


So for example a simple .pepsi controlled by Pepsi rather than a whole bunch of .drink / .food / .crappy_sugar_stuff that pepsi have to buy 'their' name in?

And all controlled by a company whose income mostly comes from making up new ones of these?


I think that adding new TLD's is inevitable as the internet continues to grow. I really believe the crux of the issue here is actually a two part external one, not the inevitable adding of TLD's. First, there is the problem of how the law on TLD's which is ambiguous at best, will deal with cybersquatting and other legal issues related to TLD's. Additionally, the larger and second part of my argument is how the search engines will add this to their already confused algorithms.


How are search engine algorithms confused in your opinion, and why would new TLD's complicate it further?


Sorry for the delay, I am a newbie and just noticed the response. The answer to both parts of your question is that search engines factor in TLD's in ranking, for instance .com's are often the most relevant when it comes to search, especially EMD's with an exact match, especially with Google. And day by day the search algorithms are changing, and search rankings lately are all over the place with algorithm updates. Many SEO leaders will support the argument that spammers dominate and the algorithms are struggling to get it right. Something I agree with. Now if you add another element to search algorithms it seems very likely that search ranking will only become more confused as data is added.


I usually agree with TimBL on everything, but not this time.

New or newly popularized TLDs represent new business and cultural territory. Without this new territory, the consumer "domain name memoryscape" inhibits the adoption of new business and artistic ideas.

It might even be a good idea to make it as arbitrary as possible: every year, choose one random two letter and several three-letter domains. Then let creativity decide how to make use of .kfi and .iz.


Randomly generated TLDs are a brilliant idea. You unexpectedly make some set of people very happy. The problems come in (a) determining where to host the root zone and (b) if that name is later needed for something else (a new country)

At the very least, you get 26 new url shortening companies, and a stack of people with cool email addresses...


Will there be .news, .games, .tech, etc. and if so, who will own them? And will they will be available for registration by normal people? Will normal people trust the owners enough to actually reguster subdonanes?

Those would be logical TLDs to have, but somehow I don't see ICANN encouraging their creation.


>He said that dot org was interesting because it captures the fact that you know that any website with that suffix is a non-profit.

This is simply false. There is no requirement whatsoever for .org domains and there are countless examples of .org domains owned by for-profit companies.


I'm sorry to play this card, but this is the (Tower of Babel)[http://en.wikisource.org/wiki/Bible_(King_James)/Genesis#11:...] all over again.

Perhaps the analogy is unnecessary, or even harmful, to make. But typically I have been taught, or rather I have learned by direct method, that some Christians believe BIBLE is an abbreviation, albeit a cheeky one, for "Basic Instructions Before Leaving Earth".

Often I wonder what the "imperative mood" of this document is supposed to be. Obviously it is not descriptive qua scientific authorship. We shouldn't assume this, surely, since a sophisticated method of science had not been yielded to the Hebrews of that time, etc. But yet it struggles, and seemingly succeeds, at preserving its descriptive weight, as vacuous as it is, in a time of reason and science. So we claim that it is allegory and tale. And so what if is? What is it describing? Like Aristotle, I believe, it describes a cycle of civilization: humanity's response to itself.

Obviously "God" in the story of the Tower of Babel is nothing more than the height of political force of Shinar. It responds to what they have built. So here we see the "confounding" of their language; like in our time, a confounding of the DNS system. What motivates this story? To what end does it serve?

Surely the U.S. government seeks to bolster the economy through its will, by "securing" our economic creativity. For "now nothing will be restrained from them, which they have imagined to do."

As we can clearly see this is a response to Otto Neurath, the Logical Positivists, the adoption of ideal, scientific and pragmatic languages, etc. Neurath sought to liberate the proletariat of his time through a technological mode of thought applied to language: complex symbols efficiently characterized such that they might contain chains of propositions. The illiterate worker thus might be informed through a proximity of symbol to nature. Today, we see this with technology: big-face-hypermodern-informational-UIs-with-geriatric-buttons, usability research, cognitive-neuroscience centered on atypical neuropathologies, the mobile web, Helvetica, etc.

The U.S. will respond as it should, and I play this card because, in the end, the only argument we face from politicians is an uncritical one: _what we've been doing works, so it must work_. How much influence these governments have on ICANN is unclear to me, but given the list of possible TLDs; e.g., pepsi, etc., it isn't tough to reach for the assumption.


Okay, so the points about Neurath are good stock.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: