The internet has changed and evolved ever since it’s ancestors first came to life in the late 1960’s. Some technology fades away and is forgotten; other aspects continue but are overlaid, like geological sediments, so that they are now longer visible but are still present under the surface.

The Domain Name System—both the technology of DNS and the deployed naming hierarchy we all use—are among those aspects of the internet that, although they feel solid and immutable, are slowly changing underneath our feet.

Act I: In Which DNS Fades to Translucent Grey

Internet Domain Names had a good twenty-year run from the early days of the World Wide Web (1995) through 2015.

Some people made a lot of money through domain name speculation. Others made money by wallpapering Google Ad Sense advertisements over vacuous websites. And a busload of attorneys made a good living chasing down shysters trying to make a buck off of the trademarks of others.

And through its perceived control of domain name policy, ICANN grew into a ever-bloating, money absorbing bureaucracy worthy of Jonathan Swift.

But things are changing. The days of domain names as the center of internet policy and internet governance are ending. Domain name speculation will slowly become a quaint shadow of its former self.

What is driving these changes?

It is not that the Domain Name System (DNS) is becoming less important as a technical way of mapping structured names into various forms of records, most often records containing IP addresses.

Nor is the Domain Name System used less then heretofore.

Nor are the knights of intellectual property becoming any less enthusiastic about challenging every domain name that they feel does not pay adequate homage to the trademarks they are protecting.

And national governments continue to believe that domain names are the holy grail of levers they can use to impose their views of right and proper behavior onto the internet.

All of that remains. And it will remain.

What is happening to DNS is more subtle: Domain names are slowly becoming invisible.

For many years internet users could not avoid domain names. DNS names were highly visible. And domain names were everywhere. DNS names were part of e-mail addresses, DNS names were prominent parts of World Wide Web URLs, and DNS names that were based on words formed a rough, but useful, taxonomy of web content.

But the sea-level of internet technology is slowly rising. We now live in a world of web search engines. We now have personalized lists of “contacts”. We now use a profusion of “apps”. And we now spend much of our online lives inside walled gardens and social networks (such as Facebook, Twitter, or various games.)

Even in places where users formerly uttered or typed email-addresses (containing domain names) or web addresses, we now enter keystrokes or words that are used by user interface code to search for the thing we want and make suggestions.

For example, when I send an e-mail, I usually don’t need to type more than two or three characters of the name of the desired recipient; for every keystroke the software goes to my contact list, does a search, and shows me the possible outcomes. Similarly, on web browsers the old “address bar” has become a place for the user to send search targets to a web search company. In both of these examples the user no longer really deals with domain names (even though in both of these examples there are domain names—sometimes visible, sometimes hidden—underneath the search results.)

In the world of Apps, games, and walled gardens there may not even be a way for a user to utter a domain name.

And if a user does mention a domain name it is frequently in the form of a shortened URL that has no resemblance to the actual domain name of the target resource.

You can confirm this by asking yourself: “When was last time I used a domain name while using Facebook or Twitter, or when playing my favorite game?” Few of us have ever used a domain name when giving an order to an Amazon Echo (“Alexa”) or a Google Home (“OK Google”).

Act II: DNS Remains, But Quietly Hovers In The Background

DNS is not being abandoned; the domain name system is as robust, powerful, and important today as it ever was.

However, DNS is being veiled. So that rather than being a central figure, visible to all, it does its job behind the scenes where few but internet operators and repair techs see it.

In days past, you or I may have gone to a web browser and entered a URL that looked like, http://upstairs-thermostat.myhouse.tld/. But today I use an Internet of Things device and say “Alexa, set the upstairs thermostat to 68 degrees.”

Same activity, same request, same devices, but the domain name has gone away and been replaced with a more convenient handle.

I use the word “handle” quite intentionally. One of the aspects of the post-DNS internet is that names are becoming contextual. These new names often exist within the context of a particular person (as in a personal contact list) or a particular walled garden (such as Twitter).

Contextual names let us escape the rules and disputes—and costs—that came from the “globally unique identifier” view of the domain name system. You and I can each use the name “upstairs thermostat”; the context prevents collisions; the context differentiates between your “upstairs thermostat” and mine.

These new names will often be used on software that internally uses domain names to tie things together. There is no doubt that Twitter, for example, has lots of internal domain names. But those domain names have become merely internal gears and wheels, they have become as invisible as the pistons in the motor of a gasoline powered automobile.

The DNS system will remain as a means of using structured names—words connected by dots—to obtain various forms of records that can contain things as varied as IP addresses, geographic locations, e-mail exchange server lists, VoIP PBX locations, etc. But it will be software rather than humans that originates those structured names and uses the lookup results. That software may, and frequently will not, make those underlying structured DNS names, or the lookup results, visible to the human user.

The fading of domain names brings benefits.

Some troublesome things will begin to end.

Domain names will no longer be perceived as being particularly valuable ways to express semantically meaningful labels.

• This will remove much of the energy that powered the DNS trademark wars that we have seen over the past twenty years. (But don’t expect the trademark protection industry to give up their relentless effort to own even private, local uses of some names—that company in Atlanta will probably want to try to prevent people in their own homes from using the word “coke” to refer to any brown carbonated sugar drink other than their own.)
• And it will also tend to de-energize marginal internet activities such as typosquatting in order to pick up advertising impressions or click dollars from people who accidentally mis-typed a domain name into a browser.
• And it will obviate the need for most of the functions of ICANN.

Opportunities will arise for application-specific or community specific naming systems:

• New names can be more descriptive of classes of possible targets rather than being tightly bound: You could, for example, say “ATM” and not be locked into ATMs operated by wellsfargo.com.
• New names do not need to fit into the confined strictures of domain names.
• New names need to be less like “names” and more like “descriptions”—more on that below.

Act III: There’s Still Gold In Internet Naming

The loss of domain names as baubles for speculation does not mean that entrepreneurs and Procrustean government bureaucrats must fold their tends and skulk away in to the night.

Even after DNS becomes merely an internal organ of the internet, there will be plenty of opportunities for fun and profit.

Companies that are presently operating as domain name registries and registrars, are well poised to capitalize on the new systems: they already have much of the customer and user facing “front office” infrastructure that will be required to service whatever naming systems may arise.

There are two broad areas in which internet naming will probably evolve: entity naming and describing.

Entity Naming

The need to attach names to specific things will be with us forever. And there will always be a need to turn names into some sort of concrete handle to those things. This will be, as it always has been, tied to the problems of figuring out where that thing is (i.e. its address) and how to get there (i.e. the route.)

One of the prime values of DNS as it exists today is that almost everybody voluntary chooses to use a single base root. So we have a global shared system that assures that all names attached to that root are unique.

That uniqueness is important, but it is not always necessary—sometimes people want a solid distributed name-to-record lookup system that is not dependent on a global root outside of their control. Sometimes people just want a private name space for some private purpose. DNS technology, as opposed to “the” domain name system, provides a useful tool for these purposes.

The name model of DNS is extremely useful, but it is simplistic: It is a hierarchy, represented by names separated by dots, that leads to sets of records that can contain various types of data. That simplicity has allowed DNS to be robust and reliable. But that same simplicity creates limits.

The world is evolving so that that simple model of names-to-records will become increasingly inadequate. I’ve written a couple of papers on this topic:

• On Entity Associations In A Cloud Network – The argument made here is that as we move towards cloud based resources the simple mapping of DNS names to a relatively fixed set of answers is not sufficient to accommodate the motion, partitioning, and coalescing of cloud based computing and data resources.
• Thoughts On Internet Naming Systems ‐ This presentation addresses certain presumed characteristics of domain names that are not necessarily true in practice and likely to become even less true in the future. For example, many of us tend to presume that DNS names will generate the same answers no matter who requests a DNS lookup. And all of us are increasingly aware that DNS names are not permanent, the underlying records, and sometimes even the DNS names themselves, sometimes change or even disappear.

And even though name-to-record look machinery such as DNS will remain valuable, it must evolve so that it can have greater security and consistency.

The larger area of future change lies in the area described by the first of the papers above—in the realm of lookups based on descriptions and attributions.

Attribute and Description Based Systems

Whether in real life or on the internet, often you want something that is a member of a class rather than a specific member of that class. You often just want “a Pepsi” rather than a specific bottle of that drink; you usually don’t care which bottle for your needs, the various bottles are equivalent and interchangeable. A word for this is “fungible”.

As is described in my paper On Entity Associations In A Cloud Network the internet is evolving so that there may be many resources that would satisfy any one of our (or our application’s) needs. DNS is often not the best solution for this kind of resource search. Attribute and description based systems would be better, particularly if they had some leeway to find things that are “near” or “similar to” the description or attributes.

We are familiar with this kind of search. For example, web search engines, such as Google, try to show us web search results that locate the best or nearest solutions, not necessarily the perfect solution. And many apps on mobile devices aspire to discover resources based on their distance to your current location.

We can anticipate that use of this kind of thing will increase.

Descriptions and attributes can be self-published by devices and services as they are deployed (or as cloud entities split or coalesce) or they can be published by those who manage such devices or services. This publication could be in the form of simple ad hoc text, as is done for much that is on the web, or be formalized into machine-readable data structures in JSON or XML.

There is lots of room for innovation in this realm; and possibly lots of room to glue-on revenue producing machinery, much as Google did when it attached advertising to web searching.

Epilogue: The Internet Twenty Years Hence (2037)

Relatively few of us remember the internet as it was twenty years ago when the World Wide Web was just getting started. What will it look like twenty years in the future?

We can be sure that whatever it looks like to users, that there will be a lot of ancient machinery, such as DNS, lurking inside.

It is likely that human users will increasingly interact with computer and networks resources much as they interact with other humans—in ad hoc and informal ways. Humans are notoriously vague and ambiguous; that will not change in the future. This means that our computerized systems will have to become more human in the ways that they resolve that ambiguity into concrete results and actions. This, in turn, means that computerized systems will have to become more aware of context and use fewer “names” and more “descriptions” when trying to satisfy human requests.

The introduction of context into network naming will mean more opportunities for damage to human privacy. The tension between convenience and privacy will increase.

As the network world becomes more contextual, it will become harder to diagnose and isolate problems and failures.

Footnote: What Do We Do With An ICANN That Has Lost Most Of Its Purpose?

The vast bulk of ICANN’s machinery and staff is present to support the domain name selling industry. As this paper indicates, we can anticipate that that industry will shrink and consolidate. And fights over domain names will fade as domain names lose their semantic weight or become hidden artifacts rarely seen by anyone except internet technicians.

The ICANN traveling circus of international meetings will become as interesting as a meeting about the future of Lotus 123.

ICANN’s income stream will shrink; ICANN will no longer be able to support its grandiose office suites, staff, and hyperbolic procedures.

ICANN will have to retreat back to what it should have been in the first place—a technical coordinator, a source of operational service levels for DNS roots and TLD servers, and secretariat for protocol parameters such as DNSSEC keys and IP protocol numbers.