Why ICANN's gTLD process leaves a lot to be desired | Hacker News Reader

Why ICANN's gTLD process leaves a lot to be desired

benwerd.com

59 pointsbenwerd68 comments

A few years ago when this was announced I seriously considered setting up shop to host these gTLDs. (I rebuilt the .pro registry, so I have the domain knowledge.) Then I realized I'd have huge risk from ICANN, if they dragged their feet I'd run out of cash. Glad I walked away, the extra couple years this took would've killed me.

Later I realized that, while the tech is easy for me, the sales cycle would be a long, painful slog. Every time I see a story about new gTLDs I feel like a dodged a bullet, as enticing as it is to want to build this product. :p

If other folks are considering this business, buy a beer for someone who's had to work with ICANN. This is one of those many businesses where the tech looks atraightforward but the everything else will crush a techie.

Thats one of the things that helped kill poptel who built the .coop domain :-(

>By failing to regulate usage, ICANN have left the door open for companies to have a monopoly on certain thematic addresses on the web.

ICANN reserves the right to revoke a gTLD due to abuse. That being said, I don't think anyone would have any form of hindrance from not having a .author domain. ever. Saying that amazon controls .author is like saying Lybia controls a suffix. It /really/ /doesnt/ /matter/. New gTLDs are cute, but thats about it.

Exactly the issue, they are cute but serve no purpose.

I disagree wholeheartedly. The purpose of the article was that the author believes that People could be blocked from registering .book or .love due to corporate interests - thereby causing a "clusterfuck" This is just silly. No one will be up in arms over this, at all.

Perhaps they won't be up in arms be but maybe they should be. What if American Express owned .com? Ten years from now owning .book might be seriously distorting.

I don't know the solution, but I suspect that having generic TLDs is better. This is like giving big companies the naming rights over cities and streets (the latter they already do to very limited extents). If Chicago became "AT&T Chicago" an that's what you had to refer to it as to get mail delivered or travel there...

Its a free market. If amazon acts like a clown with .author, authors won't use it. If they abuse their position once it is actually influential, they can be sued and have their right to registrant revoked.

simple as that.

I don't understand your Chicago argument. No one is forcing anyone to use .author, .book, or anything else. It is one of hundreds (and soon to possibly be thousands) of available TLDs.

This just seems like a steaming pile of FUD

Sued by whom? I only see another large corporation being able to take on Amazon. Then we're in the same boat as before.

These TLDs are supposed to be generic, not country-specific like .ly or .io or .uk. Giving private entities control over them is an all-around bad idea.

>Sued by whom

ICANN. They are the final say over domains, and can revoke any TLD at will.

>These TLDs are supposed to be generic, not country-specific like .ly or .io or .uk. Giving private entities control over them is an all-around bad idea.

What is 'generic' to you? Either way, you need to satisfy some form of requirement to the organization that owns the root server. Giving amazon control of .whatever is no more dangerous than giving north korea control of .kp.

If ICANN has the final say, they don't need to sue them then, do they?

> Giving amazon control of .whatever is no more dangerous than giving north korea control of .kp.

The TLD denotes some sort of category. Giving North Korea control over .kp is fine because .kp represents North Korea. Giving Amazon.com control over .books is not fine because .books does not represent Amazon.com, nor is Amazon.com representative of the concept of books. If they were giving .books to the American Library Association, I wouldn't have a problem with it.

Do you see the difference?

>If ICANN has the final say, they don't need to sue them then, do they?

Sue was a misnomer. The point being amazon wouldn't have ultimate control. They could lose it.

>The TLD denotes some sort of category. Giving North Korea control over .kp is fine because .kp represents North Korea. Giving Amazon.com control over .books is not fine because .books does not represent Amazon.com, nor is Amazon.com representative of the concept of books. If they were giving .books to the American Library Association, I wouldn't have a problem with it.

Yes, I just disagree with it. This is a slippery slope on a slippery slope argument.

Amazon is more representing of books on the whole that the US Goverment represent all of .gov(ernment) and ICANN represent all com(panies).

In order for your argument to hold water, you would need to show that having a .book domain is inherently valuable. If it isn't then it is of zero consequence. Google doesn't care that there isn't a google.kp becuase it doesn't matter. If we are to assume that we are several years in the future, amazon has been given control over .book, and they are now basically required socially for any kind of book tour/signing/press junket etc, /and/ if amazon starts turning people away who don't give them good deals, then sure, i'll be standing right beside you. And ICANN would most likely react with a request to change behavior from amazon, or they would take it away and manage it themselves.

But I can practically guarantee that

a. .books won't be that popular

b. amazon won't prevent a single author from getting a .book (they would just be turning away possible sources of income)

In the end, none of this will really matter.

I think it's naive to trust Amazon won't abuse their authority over the .book TLD. It's not a slippery slope argument so much as it's a common sense argument. Most companies do not feel obligation to do anything unless it starts costing them money, and I don't see Amazon losing public favor (and subsequently money) because they don't keep the .book TLD open to everyone.

The only way I would feel remotely comfortable is if ICANN bound the bidders by contract to keep their TLDs open, but if that's the case, why open up bidding in the first place?

And it's difficult to assess how popular a TLD will be. Ten years ago, would you have said .ly and .nu would have become as popular as they are today?

What if amazon bought book.com and then sold subdomains? So harrypotter.book.com?

I wouldn't have a problem with that since it's a .com (which means a commercial website.)

I would really like to see the Library of Congress get .books and make each ISBN a domain.

If they make a request for it, they will most likely get it. ICANN has stated they will give preference to non profit/community programs. I would assume the LoC would fall into that category.

What if HSBC owned .bank? and these gtlds take off - would Barclays or Citi be happy to have their sites running on an HSBC-owned subdomain.

What if HSBC owned bank.com? Would Barclays or Citi be happy to have their sites running on an HSBC-owned subdomain?

No, they'd simply use something else. That's why something like .movie or .book will never become any some of de facto authoritative TLD; there are too many other options for those who feel shut out.

I am up in arms about this.

This might finish killing DNS as a human interface.

Thanks to Chrome's universal address bar people are already getting used to not caring about the domain - with a clusterfuck of tlds which makes the right address basically impossible to guess, why would they bother trying?

They probably wouldn't bother trying to guess.

And I honestly don't see that as a bad thing.

I rarely guess domain names. If I don't know the address, I search it. I expect most normal people search it even if they do know the address.

It's a bad thing - effectively it gives a few search companies control over the entire namespace. ie. if it's not searchable, it doesn't exist.

It's true that the decline of DNS further strengthens the position of a few gatekeepers - Google on the web, Apple on the phone.

On the bright side, this might kill ICANN or at least make them irrelevant. That is a good thing because ICANN is arguably the worst possible gatekeeper for internet names. They combine the worst aspects of the private and public world: they inherited a monopoly over a public resource which frees them from the constraints of being an efficient and trustworthy business, like so many state bureaucracies; yet they are a private entity not accountable to the public in any way.

That doesn't make it a bad thing. When did marketing depend on search engines?

Search Engines aren't the end all be all.

What, then, is the point of a domain name at all? You can get Google to index your IP. Save cash, protect yourself somewhat from USA or some other police state pulling your domain...there are myriad benefits.

Even when people stop using DNS to find what they're looking for, they'll still use it for a long time to confirm that they're in the right place. Numeric IPs definitely don't pass the smell test. In fact some browsers might even display phishing warnings.

Having more gTLDs just makes "finding the right place" even more difficult. Is mybank.com/.bank/.money/.finance correct? SSL will be easy to get on the other domains. With arbitrary gTLDs, it just makes the domain name even less useful for identifying than it currently is.

The point of a domain name is an easy to remember address. Not a guessing game. Not an easy to guess address.

I'm not even sure how you came to that conclusion based on what I said.

Multiple domain names (like book et al) will hopefully be easier to remember.

Things like "harrypotter.book" or "looking-for.love" won't be easy to forget if you see them once.

Domains shouldn't be a guessing game anyway.

This is really already the case for some people. When I watch my mom use the web, she'll type "gmail" into google rather than typing "gmail.com".

This + IPv6 addressing will be great. People who still think "www" is /required/ would be forgiven for just giving up.

This is a huge loss for UX on the Internet.

I hate the idea altogether and in my opinion, ICANN is doing this for a quick buck.

I have a hard time remembering 22 gTLDs as of now and they are just adding more.

Also, this is going to confuse old people. Whole internet is .com for them.

Whole process is broken and I just hope this whole scheme fails.

According to ICANN they are not making money off of it. The money is spent on the reviewal process. They emphasized multiple times during yesterday's announcement they were pricing it at break even, and if a profit were to be made, it would be given to the community for them to choose how to spend it.

Their financials are available here as well, if you wanted to follow up in a year or so http://archive.icann.org/en/financials/

> According to ICANN they are not making money off of it. The money is spent on the reviewal process.

Most of which is probably ICANN salaries.

My health insurer does this. "We're non-profit, our profit margin was only 2% last year." Meanwhile, we see headlines about the CEO's $4 million bonus.

I am probably being hardheaded, but what takes hundred of thousand of dollars in the review process?

(I'd read the link but it seems last updated 4 years ago with data from 8 years ago)

Business lunches and executive retreats?

Look at how many billable hours were involved in .xxx. Even non-controversial TLDs will generate mountains of legal paperwork.

Yes, the money is spent on the review process, which is another way of saying that ICANN spends the money on itself.

The review process is conducted by multiple outside firms that tendered in an open process for the right to review the gTLD bids.

This is, by all means, a terrible move.

I'm all for destroying the squatting "industry" by introducing an unlimited numer of gTLDs if technically feasible. However introducing them only to give some random corporation control over them seems absurd.

I fail to see how these companies could end up not being abusive with the TLDs they own. Better an Internet with open market .com that cost 200000 bucks than one with monopolies over many differen TLDs.

Is a monopoly over a TLD really that different from a monopoly over a normal domain? In the end, there's nothing that special about a Top Level, it's just a shorter domain.

I love the idea of city TLD. it's nice to have domains like sightseeing.paris or theme TLDs like amazon.shop this is what we already have for .xxx but giving one company the power over TLD like .book or .author is a big problem for me...

But which Paris? Paris, France? Paris, Texas? Paris Hilton? There are a lot of things named Paris. Who gets to decide? Few cities have unique names, so it's better to be specific. There's already a http://www.paris.fr, and US states have subdomains assigned under .us (and often regional subdomains under those)

Agreed. General concept is great but selling these TLDs to the highest bidder, idiotic. I can't believe this even got past the 'drawing board'. This needs to be protested in the same way GoDaddy was.

Off topic, but does anyone know why we use subdomain.domain rather than domain.subdomain? It's always felt backwards to me.

I think it would make more sense to use com.ycombinator.news or de.google.www. From general to specific, like IP addresses.

For those of you too young to remember and not from the UK... we used to do it that way here. It was called the JANET Name Registration System and we had addresses like UK.AC.OX for a machine at Oxford.

http://en.wikipedia.org/wiki/JANET_NRS

The DNS system from the beginning has the names the other way around (see RFC 882 from 1983). Since JANET and ARPANET were largely disconnected from one another the systems were able to develop separately. Ultimately, TCP/IP beat out X.25 and the Internet protocols took over.

One side effect of this is the the ccTLDs are drawn from ISO 3166 which would have the UK specified as .gb. But since we already had a whole addressing system based around UK. it was easier to switch to .uk and ignore the ISO 3166 standard.

Worth noting that for a while there were mail gateways between JANET and the Internet that did all sorts of fun flipping addresses so that someone at Stanford say could email foo@cam.ac.uk which would turn into foo@UK.AC.CAM at the gateway and vice versa.

Not only like IP addresses but also like the path part of the URL.

Interestingly, Tim Berners-Lee now thinks the same: http://www.w3.org/People/Berners-Lee/FAQ.html#etc

I believe UUCP mail did it that way.

At any rate, this is an address. There's a convention here. Do I write addresses as "1600 Pennsylvania Avenue NW, Washington DC, 20006, USA", or "USA, Washington DC, 20006, NW, Pennsylvania Avenue, 1600"?

However, that's not the only convention for postal addresses. In Japan they write them the other way around, starting with the prefecture and working down through city, subarea, block/building number and finally the addressee's name. I don't think it's the only country using major-to-minor order, although minor-to-major is more common.

And most of Europe does "rue Blah 1600"

Well... the building number appears after the street, but beyond that it follows the standard address convention, does it not?

Piazza di San Cosimato 63, 00153 Rome, Italy

Also, I'm don't know if Japanese addresses are reversed in Japanese, but when the Japanese express them in latin scripts, they are always in standard address order clear down to floor numbers and room numbers. Here's Sony Computer Science Laboratories:

3-14-13, Higashigotanda, Shinagawa-ku, Tokyo, Japan

Straightforward address ordering.

At any rate, the IP address scheme was invented by Americans, so it seems natural that it'd use an American (and generally worldwide it seems, at least in roman scripts) address ordering.

Yes, I skipped the detail that Japanese addresses in Latin scripts are written the other way around. SCSL's address in native script as quoted on their webpage is 東京都品川区東五反田3-14-13 高輪ミューズビル (ie Tokyou-to Shinagawa-ku Higashigotanda 3-14-3 Takanawa Muuze biru).

I've always felt like physical mail addresses were written backwards too. As it is, every word is meaningless until you have read what follows.

It would make a lot of sense with internet subdomains, too. Although to me, WWW seems far more general than DE, so it would we www.de.google. Which they will probably take when this goes live, so...

I suppose it's the way your brain maps it internally. To me, it makes sense to have the most granular things first, as those become what's important.

To use physical addresses as an example, say I'm delivering a letter (I'm delivering this letter end-to-end by myself in this hypothetical scenario). Once I know the country, I don't need to check that information again. Once I'm down at the city level, that then becomes useless information, and so on.

I'll grant that prepending "www" on every site doesn't make a great deal of sense, but when the "World Wide Web" came into existence as such, it made slightly more sense on the surface.

Sort of... UUCP worked by specifying the route the message had to take by a chain of machines that were connected to each other.

So you'd have something like mainvax!bigserver!otherserver!theirvax!john

I don't really feel this... I guess this might be a horrible thing to say but "I don't really care".

It costs 100k to get a tld. So if Amazon takes .author then what about .writer? Or what about biographer, columnist, composer, creator, essayist, ghost, ghostwriter, ink slinger, journalist, originator, playwright, poet, producer, prose writer, reporter, scribbler, scribe, scripter, word slinger, wordsmith, work-for-hire, writer?

Is amazon going to take all of them? So what if they do? Does that suddenly invalidate FirstNameLastname.com/net/me/us/co.uk or FirstNameLastnameAuthor.com/net/us/co.uk?

I'm worried about the opposite trend. That is to say Amazon (or any other company) REQUIRING people to buy a certain tld before allowing them to use certain services (want to sell in the apple store you must buy a .apple tld, want to publish a book you must buy a .book tld)

Interesting last point there. Ties into what I'm thinking will happen with certain gTLDs, namely that the gTLD will set rules and structure around how that gTLD can be used.

For example, .tel, domains in that namespace all have the same structure by design. I don't believe a .tel registrant can change their DNS in order to do otherwise.

This is a good thing to an extent - DNS for most people is a huge hurdle preventing them from establishing their own online presence, now more so due to IPv6. If new gTLD provide tools to get .author's or .lawyer's presence up and running in an effective and fair way, great.

But overall I am discouraged by this possibility as its a nail in the coffin of the open internet. It's akin to the review process in mobile AppStores, really stating what you can and can't do within the confines of a gTLD. Not something I'm looking forward to becoming comfortable with on the web.

I wish people would inform themselves before blogging on topics they don't understand.

ICANN has a dispute policy in place for this. Any community group can dispute a gTLD application that is targeted at them. So ".authors" can get together and prevent Amazon from getting the gTLD.

See: http://newgtlds.icann.org/en/program-status/objection-disput...

That absolutely doesn't contradict the point of the post. Yes, a group of authors could argue against .authors. But the process remains opaque, and the point is that the onus is on an authors group to make that objection, rather than ICANN making either a more open/market-like or more fair domain allocation process to begin with.

What process would you recommend? The process was pretty open market. Anyone was able to apply for a gTLD. Running a registry is not cheap or simple. If the price for application was lower, every Tom Dick and Harry would have been applying, and then they'd have to wade through thousands of applications per tld. So who do you propose is allocated and gets to runs the ".author" tld?

There does not seem to be any requirement to say how the registrar will work, ie if it is private, public or restricted to certain categories. I don't mind public tlds and some restrictions (like .book is for books but open for registration to anyone). And fully private ones if they are company names like .Samsung are perhaps OK by me if the words are not generuc. But we don't know what .book or .love will be yet.

I'd like to see the death / an alternative to DNS for resolving names to IP. Its become a little too political + an easy target for snooping/censorship (yes, techy people can get around blocks but that's not the point).

In terms of web, most people search rather than go to an address, and if they know an address at all it's google.com.

Domains are now more for machines than people, so perhaps a different way of resolving / finding information would be better - something peer to peer, distributed and self repairing.

I speculated a little about this when the whole SOPA thing kicked off: http://www.marcus-povey.co.uk/2012/01/10/dns-is-a-symptom-of...

so the author is upset that he might not be able to register http://benwerd.book when he can just as easily get http://benwerdbook.com only 3 characters longer?

People are failing to realize that if google weights these gTLDs then some could be screwed. imagine if .book is penalized over .author or .read ?

Which in turn would open a can of worms, because Google themselves have applied for their own gTLDs.

I'm worried about the long-term implications, not my immediate realities, fwiw.

That's a good point.

Even then, I wonder why would Google choose a gTLD over any other TLD. For Google, content only matters. It doesn't matter what your address is. If your meal (content) is good, then spidey is happy to munch it and send it's recommendations (traffic).

I don't think that's true. Google gives more value to .gov and .edu domains, for example.

While that's true, it's probably because gov and edu domains are strictly regulated.

I could imagine google giving higher priority to .book addresses, provided that (a) the user was searching for a book and (b) .book addresses required that the registrant prove that they're the author/owner of a book/book series.

In that case, it'd make sense to. I think Google thinks that way with domain names as well.

What would have happened if we had sold .com .net and .org to Tandy, DEC and Iomega?

.com addresses would be owned by "the Shack", and you would have to enter your zipcode before visiting a .com site.

HP would have announced the retirement of .net after the Compaq merger, and then change their minds.

EMC would raise the price of .org domains to $10,000/year, but you wouldn't be able to actually assign IP addresses until the next release.

.clusterfuck !

Reply on news.ycombinator.com