Hacker News new | past | comments | ask | show | jobs | submit login
If ICANN only charges 18¢ per domain name, why am I paying $10? (2014) (stackexchange.com)
293 points by rms_returns on June 18, 2016 | hide | past | web | favorite | 112 comments

An idea I've been tossing around: What if domains cost $100 if more than one person wants it? The idea here is to squeeze the squatters with 000s of domains.

Whenever I want a domain, most of the ones I want are taken. Not by people who are making use of it, but by people who are squatting in hopes of extorting anyone who would actually use the domain to produce value.

I'd actually be happy if Verisign or any other private company or any government extracted such an unfair price for useful domains because it would free up so many more useful domains for use.

Edit: I'd love to hear good arguments against this, I'm partly throwing this out there to see others thoughts.

I have doubts how this could work. Let's say, I register the domain name A for my personal use. So far, nobody else wanted it. One day I get an email from somebody claiming to want the domain as well. He makes me an offer: he would step down if I paid him $30.

I'm neither willing to pay $30, nor $100 dollars for my personal domain. $10/year is already enough, so I'd probably have to decline and risk losing my domain (because he could simply buy it for $100). I'm sure there are many scenarios where someone could even buy your domain (assuming you're not willing to pay $100), and damage your reputation (and since it is the Internet, sueing them can be impossible if he is outside of the us and getting back the domain name through a court would be way more expensive then $100).

Now, another scenario is when a domain broker bought domain B before company X does. The domain broker paid the usual price because nobody wanted it at that time. Now, company X wants to buy that domain and offers the register $100 for it. The domain broker is asked if they still want it for $100 by the domain register. They would agree knowing that they can sell the domains for a lot more than $100 to company X.

I don't see how this system could help anybody but maybe I just didn't understand your proposal right.

You're talking about game theory as it relates to pricing / auction games. This is a really well-studied topic, and indeed there are much more efficient ways to match value to demand.

F.ex. https://en.m.wikipedia.org/wiki/Vickrey_auction

Somehow, I had never even heard of auction theory. Thanks!

This has to be the only comments section on the internet where I actually get smarter by participating. :)

I'd love to hear good arguments against this

Squeezing out squatters also squeezes out anyone without means from domains they're not using for commercial purposes.

I already spend too much money keeping things online. Of the handful of domains I don't do much with, one is just for my primary email account. The same name is registered on other TLDs so there would probably be some demand for it. Am I making a "useful" enough use of it? There isn't even a web page on it. Am I "producing" enough "value"?

Or am I expected to lose a domain because someone else could use it to make money? Or simply because they could afford to spend more money keeping their stuff online than I do? No, that guy can go fuck himself.

It's a limited shared resource, and if someone else would do something with it that's worth the fairly reasonable sum of $8/mo, when it's not worth that to you, then yeah, I'd rather they have it, especially if it also means that good names aren't all camped on by rent seekers who do literally nothing with the domain. That said, I think we should just price all .com's higher, regardless of demand for specific domains.

We can leave other tld's for things that aren't worth $8/month to their creators.

The thing about seeking rent, is that's domain names are not a limited resource, even .COM. Domain squatters are seeking rent, but not in a traditional sense of a limited amount of land.

A creative individual can come up with a new, memorable name. If you have a service about pets, then pets.com doesn't have to be your name.

For example, a friend really wanted [firstname][lastname].com, but there happened to be a hugely popular singer with this name, so he went with 1[firstname].com, resulting in a short, catchy registration.

In terms of land, you can't just create it (and in terms of reclaiming land from the sea for example, that land is considered capital).

Sure, there are a nearly unbounded number of .com domain names, but there absolutely is a limited number of memorable, reasonable length domain names in .com, which is what I and everyone else are referring to when we talk about this problem. If there wasn't, you wouldn't have domain squatting.

One's target audience generally has a harder time picking up and remembering completely made up names.

If it were worth $8 per month to someone else, but not to forgottenpass, that person could just pay forgottenpass $8 per month (or some lump sum equivalent). That this does not happen suggests one of the following possibilities:

1. The domain name is not worth $8/month to someone else

2. The domain name is worth $8/month to forgottenpass

3. The market for domain names is inefficient

It would be interesting if there were something similar to real estate tax, where you get charged yearly based on market value.

A simpler deterrent would be to linearly increase cost of the next domain you buy. This is the opposite of bulk discounts, and should make squatting unfeasible (at least not in large numbers).

I think this is exactly how it should work. Perhaps the first 5-10 are normal priced and then increase at some f(x) from there. I think others made good points about having a few for future or personal reasons like email or whatever. But I don't see a great reason why aside from squatting or having tons of companies why someone should own 200x .coms. Though I don't know if it should be per domain per tld or just plain per domain for the counts

> But I don't see a great reason why aside from squatting or having tons of companies why someone should own 200x .coms.

How about because they paid for 200x .coms?

BTW, for you and some others here:

Domain squatting is when someone registers domains in the hope that trademark owners will shell out cash to get the domains.

Domain squatting is not when someone simply registers domains you happen to like and wish you owned instead.

How would you decide that two buyers are the same entity? Unless you are going to force people to upload photo IDs or something, I don't see how this could work reliably.

The reason squatters work is in fact, the dispute resolution cost. A friend of mine passed away, and along with him went control of the domain of a site I was a part of. He had, prior to his death, left a note that me and another friend were supposed to keep the site running. And I own two of the other TLDs of this domain.

Domain squatting is supposedly against ICANN rules, and we have a clear brand we're trying to protect.

So when it was squatted, and listed for somewhere around $1,000, I figured this was an open and shut case with ICANN. Until I found out that filing a dispute would cost me even more. Dispute resolution seems to be offered up to a couple of different organizations, all of which charge exorbitant rates to examine your case.

So squatters have a massive range of profit margin they can make, as long as they keep the prices under the ICANN dispute resolution costs.

When you mean "squatted" - you mean that you (or someone) let the domain expire and forgot the registration fee.

That's not "squatting" or stealing. You just forgot to renew it or have a backup card on file or transfer the domain after his death.

You make it sound like someone came in and stole it when the onus is technically on you to maintain control of this domain. You can register a domain for multiple years if the domain is important in case of an unexpected accident (or death).

Unfortunately, yes, we lost control of the domain because nobody was able to secure a renewal for it. The problem is that after that, it was sold to a third party, not the registrar, who listed it for an exorbitant price. They have no interest in the domain other than selling it to us, and it has little to no value to anyone else.

There should be an easy way through which you can complain against the squatters online. Maybe Google should act as a mediator here as they can blacklist any domain from their searches and thus hold about as much power as the registrars/ICANN.

In any case, what happened afterwards? Did you get back your domain or is it still being squatted?

It's still being squatted. I own other TLDs for it we can operate the site off of. I'm hoping eventually they abandon the domain, since it's unlikely anyone else will want it.

Did you try complaining to ICANN about it via any means? An alternative is to throw a DMCA on them through Google, that's much faster (though its intended use is reporting of copyright violations, not sure how it'll work for reporting of squatting of domain names).

ICANN wants you to file a dispute. Filing a dispute costs more than paying the squatter.

How would you envision that working? If people started asking for a domain you own, you would get billed more because of it? Or would you have to register your desire for a name and give people a chance to say, 'no, i want that, too!'

Either way sounds bad for a legit company just trying to set up their own site, as it gives competitors an opportunity to cost them some money.

If you had something else is mind, please share more...

If I remember correctly, domains used to cost around $78 to register.

When the price dropped dramatically, there was discussion on Slashdot about the effects. Most thought it was great ("hey, I, a poor college student, can get a domain now!"), but there was speculation it would lead to more domain parking.

In late 1990s it was $100 and a 1 month grace period before you need to pay in full.

Domains used to cost $0 to register.

Edit: LOL Which idiot downvoted this? The chronic lack of basic tech/internet knowledge among some people here is ridiculous.

    In 1995, the NSF authorized NSI to begin charging
    registrants an annual fee, for the first-time since 
    the domain's inception.

You wouldn't know if someone is squatting or just building something that might take years before it's launched.

The best way to eliminate domain squatting is to make domains non-transferrable. If you own a domain you shouldn't be able to transfer it to anyone. The only way to "transfer" a domain should be to return it to the domain registry where it may be purchased by any interested party at market price determined by auction.

However this probably isn't going to change.

What if I register a domain with a shell corp and a straw "contact" who is an employee of Mossack Fonseca (or whatever)? If I want to transfer the domain I just sell the shell corp.

Then you have to pay an extra $50 or $100 a year to maintain your corporation per domain, which reduces the number of domains you can own with a fixed amount of expenses.

No, that would just cause a new form of corporation to arise in some jurisdiction that costs a dollar per year to maintain, that's limited to owning a single domain name.

Given the way transfers work, that would also mean you couldn't move a domain from one registrar to another. So if you had a domain that anyone else would want, and your registrar decided to raise their prices or change their policies, you'd be unable to leave without losing your domain.

Domain registrars that engaged in holding domains hostage and increasing prices would probably eventually go out of business.

Even if that ended up not being the case, we'd have to fix how domains are transferred between registrars - same as how telephone numbers can be transferred between telcos but you can't sell your telephone number to someone.

This topic is relevant to my interests. I've googled around a little bit on research on the topic of eliminating squatters, but haven't come up with much. I'd appreciate pointers to papers if anyone has them.

Coase theorem says that shouldn't make a difference.

Coase theorem applies if there are a) zero or sufficiently low transaction costs and b) if parties are able to bargain.

If domains have to be returned back to the registrar b) doesn't apply. Auctions ensure that a) doesn't apply either.

Furthermore, adding a 7 day waiting period before returned domains become available will further ensure that parties cannot trade domains between each other but must purchase from the registrar which makes owning a domain and squatting on it for purposes of selling it pointless.

Right, but both (a) and (b) do apply to the existing system where domain names can be traded. What Coase theorem says is that the resulting allocation will be efficient, regardless of how you do the initial allocation. I.e. it's irrelevant whether people pay squatters to get the domain names they want or ICAAN directly via auctions.

Do you really think transaction costs are low enough?

Buy a domain name, set up a "free this domain name fund" page on the domain, then release it when the funding goal has been met.

People have crowdfunded worse things.

You're in luck. All the new gTLDs cost $30+. Many are a couple hundred dollars or even a couple thousands per year.

The main reason is that most companies don't think it's necessary to pay thousands of dollars per year as a recurring REGISTRATION FEE.

Isn't the problem squatters sitting on domains for free via domain tasting?


I'll give you the names you want for free. But you have to change your DNS settings to point to my cache servers. Or follow my instructions how to set up your own root and cache and/or auth nameservers.

Interested... Can you elaborate on the setup ?

If I'm understanding it correctly you will be faking nameserver entries ?

As someone who's personal domain is my own name, I highly dislike this idea. It hosts my email identity and personal cloud, and would be a pain (and insult) to give up.

And I've had requests for it. Quite a lot of requests, even though it's not a very common name. What would follow, $100 moving to $1000, $1000 moving to $10000? Offer me $100k and I'll not give it up. For companies that own a trademark, this would also be an issue, because trademarks are not general, but product specific, see nissan.com

Too bad, the owner of Nissan Computers had to go through a lot of hardships[1], just because his family name happened to be the same as that of a crony capitalist.

Such stories make me sad, especially since there are extremely few who could empathize with those who suffered and even understand the problem faced by them. I really hope there are people in the US who would support Mr. Uzi Nissan.

[1]: http://www.nissan.com/Lawsuit/The_Story.php

I think the best argument is that I don't often have trouble finding domain names available for normal registration. Raising the price only benefits richer people with less imagination.

> Whenever I want a domain, most of the ones I want are taken

How is this possible with so many TLDs? Or do you mean "whenever I want a .com domain"?

There isn't a domain name shortage.

> An idea I've been tossing around: What if domains cost $100 if more than one person wants it? The idea here is to squeeze the squatters with 000s of domains.

So they dump all their crap, no resale value names that don't bring in more than $100 per year, as opposed to ones that don't make more than $10 per year.

As for non-"domain squatters", it seems like a great way to fuck with someone; just claim you want their domain and force them to cough up $100 or lose it.

This idea - which seems to be based on nothing more than bitterness that everyone in the world didn't sit back and wait for you to have first pick of names - is so transparently nonsensical that I find it hard to believe you are serious.

> Whenever I want a domain, most of the ones I want are taken.

You should have bought them first then, shouldn't you.

> Not by people who are making use of it, but by people who are squatting in hopes of extorting anyone who would actually use the domain to produce value.

Who are you to define value? Where people own 1000s of domains, they are always parked and bringing in money for the owner. Their profitability is the reason they are registered.

Who are you to decide that's not a valid way to make a living from domains? At least they're not spamming.

> I'd actually be happy if Verisign or any other private company or any government extracted such an unfair price

I'm sure you would, because you imagine you'd be a winner in such a situation, unfairly appropriating someone else's property.

> free up so many more useful domains for use.

There are literally millions of possible domain names that are unregistered. If your first preference is taken, do what everyone else does without complaint and think of another name, or pick a different extension, of which there are hundreds now.

>I'd love to hear good arguments against this

I'd love to hear a good argument for it.

Laurent Chemla, the founder of gandi.net, a French registrar wrote a book [1] on this topic, describing himself as a «thief» for selling stuff that have no cost in the first place (domain names).

[1] confession d'un voleur (confession of a thief) : http://www.confessions-voleur.net/confessions/ (in French)

I don't claim to know what the fair price for the .com registry. I will say this though the hosting isn't free. There are millions of requests per second across all networks from all across the world. They must respond with low latency. The major registries are nearly always under DDoS attack. There is a reason verisign has some of the best DDoS protection.

All the data centers across the world, the network connectivity, the DDoS hardware, servers, custom software and staff to run it all must be expensive.

> All the data centers across the world, the network connectivity, the DDoS hardware, servers, custom software and staff to run it all must be expensive.

We’re just talking about the DNS part here; not the hosting. Maintaining a bunch of DNS servers doesn’t cost $9 per domain.

Since Verisign is public, we have a lot of visibility into their profitability.


From page 28, no more than 38% of revenues could possibly be used to support their registry systems and obligations (this number includes costs of revenues, administrative expenses, and interest expense). The remainder, 62%, is used for expansion and profit: 9% sales and marketing ($90m), 6% R&D ($63m), the remainder to income tax and shareholders. And obviously sales and R&D take up some of administrative expenses and usage of debt raised as well.

That's a pretty sizeable margin for a non-differentiated product. If they didn't have the contract for .com and .net, there are many other players who could jump in: per page 6, "there are over 840 other operational gTLD registries" that do not use Verisign's services.

Don't bet on them losing the contract, though, unless you want to bet against Berkshire Hathaway - which recently upped its investment in Verisign despite uncertainty about ICANN's future. For better or for worse, a lot of smart people think Verisign's current business model is here to stay.



CIRA, who run the .ca (Canada) registry operates as a non-profit, with open books. You can read their annual report to see what they spend to run the .ca registry. Their registration fees are roughly comparable to what Verisign charges registrars. https://cira.ca/sites/default/files/public/2015-03-31-financ...

Like I said, I don't know where the line of fair pricing should lay. Clearly, it could be cheaper if they are making $1 Billion. I just don't think it could be a few cents per domain.

I met someone involved in the .com registry. They have multiple data centers per continent with 200 racks of equipment. Those are the major sites. Each major city/peering point may also have a smaller presence of something like 10-20 racks.

It is more than just a few dns servers. I know some of the people at DNA Made Easy. They spend a lot of time and money building and maintaining infrastructure and peering/bandwidth agreements at major peering points around the globe.

They are obviously much smaller than a major registry but they are very performant and compete at the lower end of pricing. Their prices are probably a good indication of the cost per query($125/50 Million queries). I guess the question is the number of queries per domain and how different the level of DDoS is from a DNS provider.

While I am sure the load is high, everytime you look up a .com you are not connecting to Verisign Server.

DNS does not work that way, Major DNS companies like RackSpace and AWS's Route 53 I suspect have higher load DNS than the Root Registries, and they offer the service for free...

> While I am sure the load is high, everytime you look up a .com you are not connecting to Verisign Server.

It depends on what's in your resolver cache. If coming from a cold cache, you would start from the root hints file[1] -- if you ask them for an A record for news.ycombinator.com, they give you the NS records for com. ([a-m].gtld-servers.net, and some A and AAAA records for those); these are the Verisign servers, then you ask one of those, and they tell you to go to amazon dns, then you ask amazon dns and they tell you it's a CNAME to cloudflare and you have to chase that down.

Next time you ask, hopefully, you'll have the delegations for com., ycombinator.com, and cloudflare.net. still in the cache; but if you have a small cache, or a large amount of diversity in domain names, you're still going to make a large number of requests to the Verisign servers; anyway the delegations are served with a 2 day TTL, so you'll need to come back periodically.

[1] ftp://rs.internic.net/domain/named.root

> if you ask them for an A record for news.ycombinator.com, they give you the NS records for com.

Why does it have to go to the .com registry? Why can't the NS servers of the registrar (GoDaddy/Namecheap/etc.) themselves provide that resolution (ycombinator.com == xx.xx.xx.xx IP) ? After all the registrars are assigned for that specific purpose, aren't they?

Looks like a centralized bureaucracy to me if each request has to go to the root com/org/net DNS servers.

> Why can't the NS servers of the registrar (GoDaddy/Namecheap/etc.) themselves provide that resolution (ycombinator.com == xx.xx.xx.xx IP) ?

They could. Any name server along the chain could just answer the question if it knew the answer. It's a matter of configuration whether a certain domain is delegated to a different nameserver. And that doesn't end at ycombinator.com. The nameserver for ycombinator.com could delegate subdomains to further nameservers, and so on.

> After all the registrars are assigned for that specific purpose, aren't they?

No, their job is to enter their customers' data into the registry's database, that's it. That is, who the owner of the domain is, and which nameservers it should be delegated to. Some registrars also provide DNS hosting, in which case their customer can choose to have them provide the DNS server that's put into the registry's database for the domain to be delegated to.

Also, the DNS doesn't just map domain names to IP addresses. You can have as many subdomains and hostnames below some domain, each pointing to different IP addresses, and you have other information in the DNS, such as which server is responsible for receiving emails for some domain name.

And in particular, none of this information is necessarily static, or even global. Large websites might have multiple webservers, and when one of them fails, they automatically remove its IP address from their DNS server so the clients only try to use the ones that actually work. Or others run DNS servers that return different IP addresses, depending on which part of the planet the request came from, to direct clients to a server that's geographically close. All of that is only possible because the responsibility is delegated to their own nameserver that they can program to behave however they want.

Also, no, not every request goes to the root servers or the com/net/org servers. Your computer, for example, asks your internet provider's DNS resolver server for any names it needs to look up. Now, your provider's server caches responses. So, if you were to go to ycombinator.com, for example, it wouldn't start at the root again, as is already knows which nameservr is responsible for ycombinator.com, as it has already looked up news.ycombinator.com in order to load this page. So it would ask the nameserver that's responsible for ycombinator.com directly. And that's also the main reason why it's built the way it is: The delegation combined with the caching distributes the load to the many individual name servers of each domain instead of having all the requests hit one central server.

Somebody put in a proposal to run the .org registry at cost for $2/domain/year a while back, but it was rejected because their infrastructure was only something like triple-redundant Postgres instead of quintuple-redundant Oracle. This is known as a "beauty contest"; ICANN picks whoever has the most over-engineered solution and when you cross that with "cost plus" thinking leftover from the military-industrial complex you end up paying $9.

.org runs on postgres.

What do you think of SRS?

Yeah, but is that anywhere near the 1 billion in renevue (According to the top answer) that verisign brings in?

And all other registries with prices under ICANN who can't even use the benefits of scale that much are subsidizing the domain names?

I thought that DNS hosting was paid for by ISP fees?

wouldn't it make more sense to pay based on how much your site gets used then? Or would that be too expensive to track?

Ignoring the fact that the revenue goes to a private company there's one good reason why 18¢ would be a bad idea - already too many domains are hoarded by a small group just for resell; I imagine it would be even worse then.

And thus lets charge 5x more and gave this money to the government appointed but not government owned company. Hardly a justification.

I can only think of one theory that could defend the US gov decision. Perhaps they didn't want to directly charge for the com domain to make it appear more open and "commercial" to foreign actors (states, companies).

For .dk domains they simply disallowed reselling domains. I think this has mostly solved this problem. I think I remember having heard cases where companies would contact domain owners with an outrageous offer and use their acceptance to 'steal' the domain from the owner.

I'm having trouble fact-checking this, do you have a source?

In Danish: https://www.version2.dk/artikel/advokat-saadan-lirker-du-dom...

It was passed in 2005, and in 2010 it was in effect for all domains, new and old. The article suggest there's a bigger burden of proof that a domain was registered for resale than what I suggested.

The solution to this should be simple. Every n years, domains are put up for auction. To keep the domain, the original owner must pay x% of the top bid, otherwise the top bidder acquires the domain for his bid price, maybe add a year or two of grace period so the original owner can migrate.

That is the stupidest idea ever. If i have a domain name and if somebody rich wants it, it can take over next year. It hijacking.

How about if that additional revenue from the sale goes to you?

Then I will bid $100 quadrillion to keep my domain and pay myself for it out of the proceeds, minus 18 cents for ICANN.

Edit: Actually I missed the part about paying a percentage of the top bid.

What if we did the same with names in general? If someone wanted to be brador on HN, and you didn't out-bid them, they were to get your name. Oh, and your real name, too, of course. If someone pays enough, you have to get a new name. And a new telephone number. And a new house. And anything else you own, really. If someone offers more money than you have, you are forced to sell at that price whatever you have. What could possibly go wrong?

So long as i get say 90% of what the final buy price is and I get to bid on my own things too? Yeah that would actually be awesome.

You do realize that anyone who had more money than you could take away everything you have with no recourse? One simple strategy would be to force-buy from you your bed and all your food at a price a little above the market price. Now, you can either accept the offer and thus get paid only about 90% of the market price, or you can reject and pay 10% of the market price. In the former case, you buy a new bed and new food, and the cycle repeats. In the latter case, the cycle repeats directly. In any case, after a few cycles, you won't have anything left.

Also, anything that you own that is more valuable than you have money, will just be taken from you, as people will offer to pay a little more than you can pay, thus forcing you to sell, and then will resell at market value, thus earning money from the difference. So, people without money couldn't own anything anymore.

1. Food and bed are commodities, a domain name isn't.

2. > Also, anything that you own that is more valuable than you have money, will just be taken from you, as people will offer to pay a little more than you can pay, thus forcing you to sell, and then will resell at market value, thus earning money from the difference.

This makes no sense. The proposal is an open auction. Anyone can bid. The seller cannot turn around and sell to someone else, because that someone else would have bid too if they wanted it, driving up the price. Hence, the price the auction ends at is the true market value of the domain, not $10.

> Food and bed are commodities, a domain name isn't.

Which is relevant how? That you cannot just buy an equivalent replacement, therefore, it's even worse if you are forced to sell?

> Hence, the price the auction ends at is the true market value of the domain, not $10.

1. Per your suggestion, the seller still is ten percent short on the market value, so someone with money can still bankcrupt you simply by repeatedly forcing you to sell (or prevent you from ever obtaining whatever goods/property this rule is supposed to apply to).

2. No, it wouldn't actually discover the market price for an illiquid good, as the seller is effectively prevented from bidding what they value it at. If the thing that they are being forced to sell is worth a million dollars to them, they cannot actually bid that much if they don't happen to have 100000 dollars to spend on top, even though they supposedly own this thing that's worth a million dollars, which should normally be enough to buy that same million dollar thing. The only way to make that work would be if the seller got all of the money. In which case it would be completely pointless, as the seller could simply always bid 10 quadrillion dollars, win the auction, and pay themselves.

Some people own domains/websites for reasons other than Money....

Wait.. so I am Ford the car company and I own ford.com. A competitor (Kia?) could bid 100 million dollars for ford.com, and I would have to pay 10 million dollars to keep my own domain?

Not sure this is great.. I could see non-profits losing their domains to porn sites.

Obviously there would be UDRP.

Interesting idea, but I think a slight modification is in order.

Like you said, if someone buys and domain for $5 a year and holds it for a few years, and then someone offers $1000 for it during a renewal auction, the owner could pay $100 to keep it. If they decided that it wasn't worth $100, the new buyer would pay the $1000 to get ownership.

In this situation, I think it would be fair that the old owner get $100 of the buyers $1000. The same portion would be returned to the owner as it would have cost to keep the domain.

Although, this could lead to attacks if the price didn't drop back down to $5 a year. If someone owns whatever.com, and someone offers $1000 for it, the owner wouldn't just have to decide if its worth $100, but if it's worth $100 a year in perpetuity.

Go home gruez - you're drunk.

This page is stuck in a redirect loop for me. This one works: http://webmasters.stackexchange.com/questions/61467/if-icann...

Previous thread where I explain that the stackexchange accepted answer (now with 78+ votes instead of 41) outlines more of the history rather than answer the question in a technical way for a technical audience:


(small correction: my previous answer had a typo of "17 million" instead of "11.7 million")

update to avoid confusion: when I wrote the previous post and relinked it this morning, the accepted answer was the "government history" by ShitalShah.

After I wrote today's post, the accepted answer was changed to the "costs dissection" by Stephen Ostermiller. That answer addresses the question better.

I assume either the original poster (Indra) or an admin changed the accepted answer.

this is a classic example of how monopoly position is monetised. nowdays monopolies are established and maintained by the government. there is a lot of money to be made, being a monopoly, but government cannot show this on its books (joe the plumber may not like that). so they create a commercial proxy entity, that accumulates most of the profits. the part left in the shadow is how moneys make it back to individuals in the government who arranged this deal, but judging by how vigorously government protects the proxy, there is little doubt that this is happening. for starters i would run a cross check between names of verisign subcontractors and names of family and friends of top icaan officials.

There are a few different places where the money goes in a domain registration transaction.

ICANN gets $0.18 per gTLD registration.

The registry (Verisign, Neustar, Donuts, Radix, etc.) sets a wholesale price that the registrar pays the registry for each domain year. That can be anything from a few bucks to hundreds or thousands of dollars (see .cars, for example).

The registrar adds a markup to what they charge the registrant on top of that wholesale fee.

In general, ICANN gets very little, the registry gets the most and the registrar somewhere in between. The registrar supplements this small yearly amount with add-on services like WHOIS privacy, or by selling email or hosting alongside the domain.

The registrar also pays a yearly fee to be accredited by each registrar which also adds to the cost to consumers.

After Verisign's SiteFinder stunt, I was shocked to hear their contract was renewed. This is the first I've read about Verisign suing ICANN over the matter and getting a settlement giving them the contract renewal.

Sadly, if Verisign is not earning $7.85 per domain, I guess every possible combinations of meaningful dot com domains were already exhausted.

I operate a registrar, I'll outline a couple pieces of the puzzle here and the justification for the price. ( I couldn't read the stack exchange article - redirect issue)

First its recognized these are digital goods where the incremental price is very nearly approaching 0. Technical aspects of domain registration have very little to do with the price.

That said ICANN has various requirements for registrars and registries that do require more than just keeping an entry in a database. For registries, you have to not only have a hot failover data center location for your registry but you are contractually obligated to test it (I believe every 6 mos. but its not my part of the industry). For registrars you have to escrow your entire domain settings both incrementally and fully (daily and weekly) for all domains, this way if you disappeared as a business tomorrow they could recover everything. This is audited regularly and swift action is taken for those who are delinquent.

Those are the technical issues, the rest of the cost is highly related to the administrative burden of domain management.

Some of these are:

  DMCA takedown requests

  Generic legal requests

  UDRP Claims

  Governmental abuse claims

  NGO abuse claims (i.e. other hosting firms etc)

  WHOIS verification claims
We get these kinds of requests on a very regular basis (ie hourly/daily), it takes a team of people to manage them and its a never ending torrent. Some of them can be disposed of quickly others take a lot of time.

There is also a huge amount of what I'd term "misguided requests". For example:

A company who makes a ICANN complaint because after the UDRP ruling saying they won a domain and access to the domain was provided, they failed to renew it, let it expire and another company got the domain. This happened and turned out as you would think. This still took 1 person a week of investigation and back and forth to dispose of.

Complaint that a domain registrant did not receive expiration notices when in fact there is no history that they'd ever been a registrant of the domain.

People who file false WHOIS complaints with ICANN because they don't like the domains owner, whois complaints MUST be verified or the domain is taken offline. These complaints create a huge burden.

All of these requests have rules about procedure, deadlines that must be complied to and a form of investigation that must be adhered to.

While many domains are nice and quiet and don't need much attention. The legal, abuse and governmental drivers that run the modern internet create a ton of overhead for registrars. That overhead is manifested in your fees.

This is only about normal fees. Premium domains are strictly market driven and the "scarcity" of a domain/tld is highly subjective. Costs are less about overhead and more about what the market bears.

Hope this has been insightful for anyone who read this far!

Distributed DNS like service would solve most of the problems. Something similar to DHT for DNS.

Who resolves disputes in a distributed DNS system? Is it just a race for land, first who plants a flag gets it?

The internet's domain registry does absolutely need management. It just needs management which is unbiased and competent.

In practice, Google is already a mediator of sorts. They decide that "car parts" points you to carparts.com while "computer parts" points you to newegg.com. (I hear there are other search engines that provide a similar service.)

All we really need is a canonical identity that doesn't change when IP addresses change. It could just as well be a random string.

Perhaps we don't need manage everything. Perhaps not every action needs arbitration. On a bright side such system can't be shut down or censored.

The retail price of goods and services is determined by supply and demand, not cost plus markup. Registrars will sell you both 99¢ domains that nobody wants and premium domains at thousands of dollars. Supply just increased thanks to new TLDs, yet demand seems to be moving slow to these new options. When more of the new TLDs are registered and adopted by the market, overall prices will go down. Eventually, if a blockchain-driven domain registry takes hold, ICANN domains may become free or even disappear.

I think if Google and Firefox join forces and create a Dns override in their browsers that resolves domains without querying the root servers it could be a real game changer. I remember this used to be an option in very old versions of IE where some company it using some plugin. This can be an end to all such monopolistic tactics and domain squatting.

I don't think you understand how DNS work. And it wouldn't change a damn thing anyway, it is exchanging a monopoly for another one. DNS aren't only used in browsers.

You mean that breaking DNS would be a game changer? Yes it would be a game changer in the most horrid and absolute sense.

Too bad lots of other applications also use DNS so that solution will only work for websites.

You don't get from words in the address bar to pictures in the window without the roots.

It still blows my mind they Verisign to retain .com despite all the obvious corruption.

Why do we even need domain names? I mean unless it's a brand, we rarely type in a domain name, it's usually bookmarked or linked.

> Why do we even need domain names?

Let’s assume that we retain the level of indirection by assigning UUIDs instead of domain names. If you used Tor, you may have caught a glimpse of how randomly looking domain names feel.

How nice would it be if we could make the domain a technical detail? Let’s say the browser would prominently show whatever entity the certificate is registered to instead (and HTTPS is everywhere).

To an operations engineer, readable DNS offers some advantages: in some environments we still refer to machines, and being able to distinguish between “eu-sigma” and “us-gamma”, for example, is much easier than between “7c07c564” and “74c47513”. However, some may disagree[0], and anyway this doesn’t justify the need for a global DNS.

As to the end user, the primary concern is that in 2016 we still type in domains[1], and at those times better memorability and shorter length matter a lot. You can also see domain names referenced in anti-phishing warnings—but with proper security practices that wouldn’t be a thing.

So I see that plus inertia—an upgrade costs time and money, and the existing system works (and makes some people money).

[0] https://news.ycombinator.com/item?id=8010838

[1] Anecdata. Not sure if any research attempted to quantify how often do domains get typed these days, perhaps this isn’t even an issue anymore.

Because if you used bare IP addresses

1) a site or email service couldn't change hosting providers (and hence IP) without breaking all links and bookmarks

2) you couldn't have more than one site or email namespace hosted on the same machine, since the server wouldn't be able to distinguish them

Having a layer of indirection gives you much more flexibility.

And if you don't really care about the name, you can a domain for less than $2/y on one of the new TLDs.

What does the bookmark or link point to, if not a domain name? If it points to an IP address instead, then changing servers would break all links and bookmarks.

not to mention that you can always bookmark a bookmark.

Domains used to cost $100 a year. I don't see any reason to complain (although before they cost $100 they were free).

In short the domain name system is a racket.

Why do registrars charge more per year if I try to register a domain for more than one year?

Registration is open for Startup School 2019. Classes start July 22nd.

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