The Web Is Unsustainable | Hacker News Reader

The Web Is Unsustainable

shanehudson.net

34 pointsrm2kdev69 comments

Perhaps we need a way to navigate to websites without requiring the use of a unique identifier such as domain names.

No. I don't care if the identifier is an English word or a SHA512 hash; you need identifier for linking. Linking is what makes the Web what it is, not the fact that you can write a company's name and append .com to go to its website.

On the other hand, ipv6 (and being a number based system, future versions are possible) will allow us to carry on without domain names, so if you don’t mind users needing to use an ip address then there is no problem.

I'm sorry no, that doesn't really work. What happens when you move from EC2 to your own dedicated server, or vice-versa? You lose all your links?

Not defending him but ipv6 supports mobile IP allowing you to move but retain your ip by having it rerouted to your new one.

For how long? Does the old IP just route to the new one forever?

You got to set it up, but there is no reason it can't last forever since there are enough addresses. And when they get rerouted they can advertise that the change is permanent and have people use the new one.

The first OP's assertion is wrong, because you always need some kind of a unique identifier for routing information, there is no way around that. It's possible not having to use it directly, like most of us don't need to remember all the phone numbers anymore, but some identifier is unavoidable.

Domains operate in a manner to real estate. We've been through the "wild west" land grab of domains at this point and now we're working toward the next stage, whatever that ends up being. Remember that domain names are transferable so the reality is that, of all of the millions of registered domains, only those which are profiting their creators are truly taken without much chance of being reclaimed (probably 1% or fewer).

This is a problem that will sort itself out, there's no cause for concern.

I agree with you.. but, then shouldn't we be asking ICANN to raise the registration fees?

Because right now, registration fees are so low that squatters can register thousands of names, and then demand a few thousand dollars if you won't one of those.

Raising it to (say) $25 keeps names affordable for everyone, and would raise the cost to squatters substantially... which should make more names available.

No, the domains are already taken.

yes, they are taken. And when they come up for renewal and the price is $25 instead of $8.. the owner will have to decide if it's worth it or not.

Oh I see what you mean. That's a good point.

Raising the price is what I advocate, too. At the current price, squatting on a domain name can be profitable if it brings in $10/year. Raise the price of a domain name to $50/year, and millions of domains will free up.

In fact, I, too, would probably stop renewing some of the domains that I am not using for any purpose. I'm not even a squatter.

I'm not a squatter either, but if the price were $50/year, I would drop about half of the 30 domains I own.

If you're not using them, isn't that a good thing?

absolutely. But at the current price of $8, I'm going to keep them forever. Never know if I might want to use them later, right?

Really dislike any suggestion like this.

It won't help much, it'll just price the low end out of the market. Given that a good domain can sell for thousands, you'd have to raise the fees a lot to deter people chasing that one hit.

Meanwhile, you damage lots of small people who have domains for personal things that never were intended to make money. The Internet was meant to be an open place with low barriers of participation.

~$25 probably wouldn't stop the issue, but something ~$100 might. "GOOD" domains will always sell for well over that, regardless of a price floor. Owning a good domain name is like owning any other valuable good. Pricing around $100 helps with the long-tail of domain names. Most domains aren't worth sitting on for hundreds of dollars (in the majority of cases, where there's not an obvious quick flip).

I'd even be for a tiered pricing structure. Domains under ~7 characters could be $100+, domains over 15 characters could be significantly cheaper.

> but something ~$100 might

What about the cost to innocent parties? Really dislike this!

> Domains under ~7 characters could be $100+, domains over 15 characters could be significantly cheaper.

Now that could be good - Charge inversely by the character.

Also has the side effect of discouraging URL shortening services :-)

of course really good/valuable domains will always be worth a lot more than the registration fee.

$25 is still basically nothing. Name one other thing important to you that you spend less than $25 PER YEAR on. Geez, right now I spend more on coffee in 1/2 a week than on a domain name. I spend more on toothpaste than a domain name. Who could possibly claim that is too high?

Squatters of course think it is too high. Owning thousands of domains they consider worthless. Well other people don't consider those names to be worth less than $25.. so if you do, maybe you shouldn't have them registered.

I wonder if it would work to price domains based on how many you already have. Charging $(5 * n^2) would keep squatting down while still allowing individuals to own a few domains they're using.

That would be good, but that only works if you have a fool proof system to determine how many someone already has. And what if a company own's a domain name, does it count towards someone's limit? People might just start setting up shell companies to own domains.

Certainly a major issue, although the costs of maintaining those shell companies (taxes and fees and whatnot) may help check it somewhat.

> of all of the millions of registered domains, only those which are profiting their creators are truly taken without much chance of being reclaimed (probably 1% or fewer).

I don't think this is true. Domains are so cheap to keep around, squatters only need to profit on a small percentage of them to keep a large number of them on hand. If a squatter lets go of one, another squatter will simply pick it up.

I'll give you an example: When I was in high school in the late 90's, I registered my last name as a domain name (uresk.com). At some point (1999, I think) I couldn't afford the registration fee (~$35 per year back then, I think) and had to drop it, thinking I'd pick it back up later. Ever since then, the domain has bounced around various squatters and is to this day unavailable to me. There is no way they are making any money off this domain name:

- I have an incredibly uncommon last name that is shared by maybe only a thousand or fewer people in the world.

- The .net version of the domain name (which I own) gets only a few thousand visitors per year.

- The sites run on the .com version have never had anything but parking pages up.

The only reason the domain name keeps getting registered is because of the criteria that someone owned it before. I'd be shocked if the domain name became available in my lifetime. I see this same situation played out whenever I go to find a new domain for a business idea I have - all of the decent domains are just be squatted on, and the only way to get them is to pay insane fees to squatters.

So, no - the problem isn't going to resolve itself.

The diagnostic is correct, domain names are not sustainable. The drug prescribed is not.

The solution I thing will come up is again hashes. Now we have hashes for file handling (in git), hashes for users in Google+, hashes for money (bitcoin). I think we are going towards a world full of hashes (pun half-intended).

If we had a single hash used for identifiying a piece of content (eg a given blog post, the home for this blog, a comment to this blog post), then a web site would just be a hash table, you would ask for hash 3ed44a2d38 and get the content. That may include other hashes. Then search engines would just be weighted keyword->hash tables. Content would be passed along servers and proxies in a p2p way. Because hash is unique to a content, you won't care if the bits are served by X or Y, as long as the hash checks.

Then no need for domain names anymore: Wanna go to BBC website? Search for BBC and pray your search engine has it as #1. Wanna bookmark the home of BBC website or an article? Store the hash.

Issues:

- Over reliance on search engine and links. Answer: we do have this issue already.

- Authentication of content: is this really BBC website? This can be fixed similarly to current certification system, and you would certify your main home's hash.

- BS! That's no real hash! What do I mean? Yes, right, the main issue is that content is changing, so a real hash would be changing too. I said google user id is a hash, that's wrong (or is it?), it looks like one though. I have been fiddling too much with git recently. For me the best of the world is when files, dirs, actions, everything is unique, unmodifiable, and identified per its hash. Any modification of the state of the world generates new content, relations, and their hashes. Infinite history for free. But how to apply this to the Web? Can we? Nearing one o'clock in the morning here: I may clarify tomorrow.

I suggested two ideas I had, not saying by any means they are correct... just wanted to start a discussion. And it worked! Some very great points in this post and all other comments.

http://en.wikipedia.org/wiki/Content-centric_networking

Useful and readable domain names are running out.

This is false as of this moment. Startups in the current batch have been able to find decent names that weren't taken. Maybe there will start to be a problem in the future, but 26^n is big.

26^n is big (for n > 7) but also irrelevant. "Useful and readable" is a tiny fraction of that space.

Edited to add:

While there's obviously some correlation between small n and the usefulness and reasonableness of the name, my point was not primarily related to length - the parenthetical was added as an afterthought. I was simply pointing out that the overwhelming majority of sequences of letters are not going to be easy to use or remember.

Phone numbers aren't useful or readable, so the machines that talk to each other use those and I just click "Mary" in my contacts list.

I don't think such a system would be difficult to put in place on the web right now. Linking is done with whatever gobbledygook URLs you don't feel like reading, and if you want to revisit a page, you bookmark it with an alias of your choosing.

EDIT: I should say that I think this is the workflow for >90% of web users already. If they don't just use Google every time.

Dropbox was getdropbox.com for a long time, which is 10 letters.

Prepending "get" (or sticking "app" on the end) always makes me kind of queasy, but it worked well enough for Dropbox. I kind of like Exec's approach of using iamexec.com.

Square, SquareUp.com, owns Square.com but still uses the former domain.

And yet thousands of users every day manage to navigate to news.ycombinator.com (n=11, at least), or wordpress.com (n=9 + the blog's name). There are massively popular blogs like overcomingbias.com, or marginalrevolution.com. I still can't spell disqus properly on first attempt, that didn't stop them.

It's important to have a good name but not crucial and it doesn't have to be short.

Sorry, I didn't mean to imply that "useful and readable" implies an n <= 7. Small n makes otherwise unintuitive or hard to remember things easier, but there are obviously longer "useful and readable" domains. Note, however, that while overcomingbias.com is useful, most domains in its immediate neighborhood are not. qvercomingbias.com, wvercomingbias.com, etc.

There is a growing a problem with startups finding good names, which is one of the reasons I've tried to offer personal .com naming assistance with NameMage.com. Although this isn't a sustainable long-term solution to the growing domain problem, I'm hoping this service can provide a solution for those in need right now.

I think this is a case where someone is trying to come up with technical solutions for a perception problem. Jumping back to the Harry's Chips example, why do consumers have such negative views of URLs like this:

harrys-chips-of-york.com

harrys.yorkbusinesses.com

harrys.chipshops.com

harryschips.info

DNS (as it stands) is a perfect system, but {short name/no dashes or numbers}.com is unsustainable. As soon as people start looking at subdomains/atypical tlds as being just as valuable, the problem will right itself.

Short name no dashes dot com is not unsustainable; it's completely over with for those without funding. There's nothing wrong with .info and the like (I use a .info for my personal domain) except perception. My hope is that the new GTLD rules will make the domain extension less important, and if so it will make squatting values go down.

Kind of off-topic, I'm slightly reminded of this: http://xkcd.com/1007/

People have been making claims of this type ("the Web/Internet is unsustainable") for at least the past twenty years, and I've been reading articles to this effect for the past sixteen years. Every single one of them was wrong.

With such a bold claim, and such an awful track record, why should I believe that this time will be different? You can only cry "wolf!" so many times before the only rational thing I can do is ignore you.

Also, why is it a requirement that I be able to get a short domain for cheap? The fact that some names are cheap and others are not is no more surprising than the fact that some real estate locations are worth more than others, or that some people have higher incomes than others. All of those are the inevitable consequence of free choices made to make every individual chooser's life better (to the best of their knowledge and ability), which is a great thing!

I don't like any solution that means the user personally has to deal with IPs. The abstraction away from IP's to names is a brilliant feature with all kinds of benefits, and one other areas should be looking at.

I even recently proposed a quick hack for the phone system: http://jarofgreen.co.uk/2012/07/phone-numbers-in-dns/ :-)

Also just wanted to point out that your solution 1 needs to incorporate the use case that people want to look at local services away from their local areas at times: to take just one example, thousands of theatre people are about to descend on Edinburgh, UK, for the Festival from all over the world, and I'm sure they have checked out the local area online first.

Don't know what the solution is, but then I'm not really convinced there is a problem with this ...

That is one option, allowing people to own the domain names but also allow many websites to somehow have the same name (with the unique identifier being IP or similar).

I've always been frustrated that the WWW URI doesn't support name-based virtual hosts in the authority section. While a name-based virtual host is usually specified as a domain name, it doesn't need to be. Many web sites share a single IP, so it should be possible to access them without requiring DNS resolution (examples showing name-based virtual host in brackets as one possible approach):

    http://[www.example.com]192.0.2.1/
    http://[www.example.org]192.0.2.1/
    http://[www.example.net]192.0.2.1/
    http://[foo]192.0.2.1/

Or requiring DNS resolution:

    http://[www.example.org]www.example.com/
    http://[foo]www.example.com/

Note that this doesn't serve the same purpose as a subdomain, because it allows one to explicitly specify the name-based virtual host to serve from a domain or IP address, instead of having a single part serve two very different purposes.

Of course there are issues. For example, you could create URIs like:

    http://[www.google.com]192.0.2.1/

And SSL poses a challenge:

    https://[www.example.org]www.example.com/
    https://[foo]www.example.com/
    https://[www.example.com]192.0.2.1/

But from a technical perspective, this opens up a lot of possibilities. Currently, the only way to achieve this is to write your own headers when making a request, so it's already possible, just not supported in mainstream clients.

I think you're looking for a problem where there is none. The only case for a "problem in the distribution of domains" would be your "local shops that can't grab their names".

The DNS is a very well thought system, I think it's one of my favorite protocols. ICANN (or any sub-agancy of it) can just decide to take some domains to make the names more "granular". A first approach I thought of in just a few seconds is, with the US as an example, taking over "state code".(com|org|net|...) and only delegate those to small business operating within one state. You get a lot of "domains" for free.

ICANN are selling TLDs like .pepsi and .sony - if we really run out can't we just open that up to the general public at a low cost, the same way anyone can get a .com? Then all the chip shops in the world can use harrys.chips, johns.chips, etc. Other companies could register their own TLDs as well, with the added advantage the URL is a single word which is the name of the company. The only reason this is not already possible for all but the biggest companies is ICANN charges such a high fee for a TLD - IIRC it's about $180k, with no guarantee you actually get it.

There are probably more than one Harry’s Chips in the world.

This ambiguity can be solved, actually. Use hash-based addresses for linking content and to resolve IPs, and use human readable dns names as bookmarks.

For example:

    Harry’s Chips 1 has hashlink http://F00BAR1.hash
    Harry’s Chips 2 has hashlink http://F00BAR2.hash

User goes to harrys.chips and is presented with a list of options:

    Harry’s Chips [5 km from you] [visited 5 times] [recommended by Your Friend]
    Harry’s Chips [...]

That way squatters go away. Of course this is just variation on what search engines do right now, so it will never take. Too much hassle for not enough gain. But this is probably how you will end up designing any system that tries to tackle the Wild West of how DNS works right now. The added bonus is that it works very nicely with all kinds of distributed systems. On the other hand, it enables all kind of new phising attacks. Probably not more than current DNS system does, though.

Well, that just shifts the problem of name resolution into the TLD space. Trying to arbitrate .apple between Apple Computer and Apple Records is no different than trying to arbitrate apple.com between them.

More generally, the problem is that any single namespace containing approximately the entirety of the world's electronic communication will result in name collisions. This already occurs in natural language absent of electronic communications. Monster Park in San Francisco is sponsored by Monster Cable, except that it's commonly thought to be the careers site Monster, or even Monster Energy Drink.

This collision problem will happen no matter what the medium is. .com names will collide, TLDs will collide, top search results will collide, keywords (like AOL's) will collide, all because natural language collides. The best we can do is include an arbitration system to resolve obvious offenses when possible while keeping out the speculators and spammers, which approximately is what ICANN is doing with the new TLDs.

You're completely right, but most of today's problems are caused by artificial scarcity leading to speculation.

I feel like domain names are difficult to find not because they are really being used up, but rather because there is no system governing them. What if we treated domains like trademarks, so if you don't use it, or stop using it, someone else who is using it can take the domain from you?

The thing is, this is also a problem for company names, and product names, and... well just about everything involving names. Almost everything is running out of names. We need to do a better job of recycling old names if any of this is going to be sustainable across any market.

I don't like the idea of taking names from people who are otherwise maintaining the requirements for the name simply because someone else has a trademark for it. Also, multiple companies can have the same trademark in different industries (not to mention different countries).

Why not just raise the price for domains? If the fee were $25 or even $50/year, a lot of people (including myself) would drop domains we weren't using. And that's still a pretty low bar for any name that is even slightly valuable to the domain name owner.

Define "use".

This is one of those articles I wish I could downvote.

Soon, companies will be able to compete with gTLD, some have registered hundreds and will push them as the part of the internet for different topics. Google's ".lol" is probably the most obvious example, finally a place for all the funny sites.

After 27 years, people will start to get used to the fact that the internet is more than just ".com" (and maybe one's country's ccTLD).

I had to scroll down awfully far in the comments before I found a sane comment.

There are reasonable complaints about the process of producing these new tlds right now, but anyone who thinks there is any solution to this problem other than new tlds has not actually thought about the problem or doesn't understand how the domain name system works.

> Perhaps we need a way to navigate to websites without requiring the use of a unique identifier such as domain names

Or we add a .suffix to make it unique, like how the domain name system already works FFS.

Content-centric networking aims to solve some of these problems: http://en.wikipedia.org/wiki/Content-centric_networking

"Content-centric networking (also content-based networking, data-oriented networking or named data networking) is an alternative approach to the architecture of computer networks. Its founding principle is that a communication network should allow a user to focus on the data he or she needs, rather than having to reference a specific, physical location where that data is to be retrieved from. This stems from the fact that the vast majority of current Internet usage (a "high 90% level of traffic") consists of data being disseminated from a source to a number of users.

The contemporary Internet architecture revolves around a host-based conversation model, created in the 1970s to allow geographically distributed users to use a few big, immobile computers. The content-centric networking seeks to adapt the network architecture to current network usage patterns."

It is funny that http://www.shanehudson.com/ is the AdvoCare site, same as http://www.advocare.com/ registered on: 22-Jul-95(1), and shanehudson.com was registered on 10-Mar-2001(2). His site, shanehudson.net, has been registered since 26 Oct 2007 17:12:02(3), yet the following domain names are still available shanehudson.org, shanehudson.info, shanehudson.biz, shanehudson.us, etc.

Maybe the Web will become unsustainable when those domain names are registered.

(1) http://www.networksolutions.com/whois-search/advocare.com

(2) http://www.networksolutions.com/whois-search/shanehudson.com

(3) http://www.networksolutions.com/whois-search/shanehudson.net

This sounds like the argument google etc. are using for making everything "social". You don't type in a domain name for "ham's fish and chips", you just search for it, and google knows which community you're part of and directs you to the correct one.

Obviously everyone here will hate that idea, but "normal" users may welcome it.

How would this scheme support hyperlinks from one page to another? You wouldn't want the page that your hyperlink points to to change based on the changing whims of some search engine.

Hyperlinks could work as they currently do, pointing to the (hidden, non-human-readable) real URL - but the UI would never show the URL. Or we could just accept that they change - in normal conversation you don't give out unambiguous pointers to specific things, you say words and trust they'll mean the same thing to listeners (at least within your community). If someone's telling me about a concept or a news story, I probably don't want a link to a specific web page.

How to stop domain squatters:

Years ago I remember cases where people who were deemed to just be "squatting" on a domain could have it taken away by a arbitration-like service at the registry. There were tests to help them decide if they were truly squatting or were using it themselves. Even so, opponents argued that there were many mistakes, that the arbitration services favored big companies and the process was abused to shut down legitimate comment (like "companyxsucks.com" sites) and the only people who really liked this system were the lawyers.

Is this a route we want to go down? Have some kind of arbitration service who can decide if someone is a squatter and reclaim their domains?

Google already solved this problem. Domain names are just an implementation detail now.

Are you saying that having an appropriate domain name is no longer relevant because the simple fact that Google is aware of a site, people (read: potential customers) have a likely chance of finding it?

Domain names containing relevant keywords are given a boost in Google SERPs for said keywords. For instance, google "marketing click" (no quotes) -- three of the top four results have the keywords in the domain name (click-marketing.net, clickmarketing.com, click-marketing.co.uk).

While not always true (google "juice fast"; you'll notice that none of the top 10 results have the keyword in the domain name), there is plenty of evidence to support a clear benefit for a domain name to contain relevant keywords.

Domain names are just nameplate, people keep forgetting that. Utility lies with search engines. What would really solve end user usablity problem is improving search - and companies improving their standing not through interrupt marketing but via inbound marketing. This name-issue is moot point, because if service is large, useful and popular people will remember it or will easily find it via google. Use bookmarks and what not. Sure it is somewhat important in the startup market if you company depends of VC funds. Business card based advertising of your brand is tenuous at best... my 2c

As a couple of others have alluded to, I think the solution here is pretty simple: search. Businesses have had to distinguish themselves by name and brand already for a couple of centuries and I've seen advertisements on a number of occasions use instructions to "Google us" before. Just because search is currently a bit wild west doesn't mean it will always be that way. I can imagine a more standardised search algorithm being provided as an alternative to DNS which would be used both for locating sites and linking to them.

Isn't this what the new TLD domains rollout addressing?

gTLDs were invented to create jobs for lawyers, marketers and others

There already are a large number of mostly ignored TLDs (.info anyone?). Everyone wants .com's.

I would argue that for most users domain names simply aren't important anymore. When I want to go to the website for, say a local bicycle shop, I just Google them, and it doesn't matter if they have companyname.com or companyname-of-burlington.com, the GOOG finds them (not just by my terms but also proximal searching, etc).

To some degree I do think this "problem" will be self-resolving -- the domain landgrab is largely over, and many are stuck with domains that they are costing them yearly for absolutely nothing (when searching for domains I tend to find that the overwhelming majority that are taken are dead-ends or placeholders. At some point people will just abandon them).

I think the problem is that our domain name hierarchy is way too shallow and way too contentless. Domains like .info are worth less than .com because I have to remember that it's a .info, and there's no good pattern that tells me I should be looking for foo.info instead of the foo.com for the foo I want.

The English language is unsustainable. We'll never find word for all these new things we are inventing. All the simple ones are taken.

And really, how many of these businesses will be around in 50 years? I think there might be some churn.

Many of the good domain names are still "available".. for the right price. It just means that companies need to resort to domains like getdropbox.com before trying to acquire the good ones like dropbox.com.

Isn't ICANN introducing .anything domain names? Once these become more prevalent, the price will go down and this will be a non-issue.

Flagged for the stupidest headline and "problem" I've read in a week.

The Earth Is Unsustainable...

tl;dr: All the good land is taken and there is no more free or cheap property available that is appealing to me. PANIC!

What this fellow is missing is actual numbers of professional squatters to regular squatters (people like you and me).

Are most unavailable, unused domains being held by a few hundred/thousand pros, or does the long tail of people like me who maintain about 20-30 domains that remain unused out number the pros?

There's a market opportunity here somewhere. We have lots of people wanting unused domains, and lots of domains locked away not making much money.

Reply on news.ycombinator.com