Do you want an open standards-based internet or a closed proprietary one? All of your examples of systems that don’t require DNS: skype, AOL, MSN are closed proprietary systems. They work, but they’re a pain to inter-operate (if at all).

How will those “hard-wired IP’s in Google links” get the new location of a document when it moves?

Yeah I know, I have an obvious bias.

To Martin, are you suggesting that on my business card (suppose I work for VISA), I should print “Please using Google search ‘VISA’ for my webpage” or a string of numbers? I really don’t get your point.

I agree that other nameplaces on the Internet should be ‘governanced’, but I have to disagree that the importance of DNS is degrading. In fact, it gets more important when all the new technology develop.

I’m not arguing DNS isn’t very very important, just that DNS is not an integral part of the Internet, and that it is seeing a relative decline in importance as other namespaces gather momentum.

Bob Frankston has written extensively about how DNS confused naming and addressing and left out a middle layer of persistent but neutral identifiers for Net resources.  (Widespread 404 errors were the outcome until every web server and CMS had built its own different way of solving this issue of relocatable resources.)  The cost and limitations of DNS are part of what drive people to create new private naming and routing mechanisms.

The hypothetical VISA example actually illustrates a weakness of DNS, because it confuses the domain record with a resource.  We kludged the convention “www.” to reflect the fact there isn’t a “WWW” record type in DNS to say what the home/root web page is for a domain.  Really I should be able to derive a lot more about your oganisation just from the “visa.com” bit.  You business card should just say “Organisation domain: visa.com”.  That would then bootstrap me into the various web services VISA offers.

DNS also doesn’t extend to anything finer-grained than domains (as you would expect), and ENUM carries much of the problematic baggage of DNS too.  Cue more private directories.

If we were doing it again, we would probably have an extra layer of indirection, RDF-like record constructs, and a fine-grained role model, with more metadata about where the information came from and who is stumping up for its authenticity.

In the absence of DNS evolving fast enough, private alternatives will continue to proliferate.

I’m not saying this is a good thing, just observing what’s happening.

I tend to agree that the rise in the use of search engines lessens our dependence on DNS.  (Of course, DNS provides other capabilities to the web, including aliasing and load sharing, that aren’t easily reproduced using raw IP addresses.)

However, I think that the general trend is toward an increasing reliance on DNS.  DNS is critical in the support of email, still arguably the most important Internet application.  Most anti-spam mechanisms, including SPF and DKIM, use DNS (mostly because it’s the only ubiquitous Internet service that’s available to them that’s also capable of storing the necessary information).  And inside companies, DNS is required to support Active Directory and the dynamic registration of DHCP clients.

I’m sure that someday, DNS’s rightful successor will emerge and will provide a more capable platform for supporting these functions.  But what is that platform, and who’s working on it?

Perhaps DNS’s weakness is more of a philosophical one.  By bundling so much into one service, maybe it doesn’t enable the right level of competition between competing ideas of how things progress?

On the other hand, if DNS is good at one thing it is pointers and delegation, so perhaps that isn’t so; innovation at the edge clearly can and does occur (viz the many anti-spam ideas being deployed simultaneously).

Why hasn’t DNS [the system, not technology] naturally evolved to encompass ENUM-like functionality?  Did we just take a wrong turn early on?  Perhaps it’s that damned ‘@’ symbol to blame!  Everything to the left in an email address was excluded from DNS, presumably because there were scalability and management concerns.  Why didn’t we all get personal subdomains rather than usernames?  Why didn’t we ever evolve a ‘USER’ record type?  Why did we stall at ‘MX’, and not evolve a long list of ‘well-known user services’ just like we did with port numbers?  Or is the opposite true: was adding ‘MX’ a mistake, and precluded the early emergence of better service discovery mechanisms (a kind of proto-UDDI)?

Whatever the truth, it is clear that private ‘typed’ directories (people, places, things, ideas) and namespaces are booming.

Actually, in the early days, DNS *did* contain information about individual mailboxes—see the now-defunct MB (MailBox) resource record.  A mailbox was identified by a domain name with no “@” or local part.  The format of the RNAME field in the SOA record is an artifact of this.  I’m not familiar with the history of the failure of the MB record, though.

> Whatever the truth, it is clear that private ‘typed’
> directories (people, places, things, ideas) and namespaces
> are booming.

DNS is an ubiquitous service. Like electricity, or water.

Those private typed directory things are highly visible, heavily marketed, products.

Apple, meet Orange. You’re now going to be compared.

Given that is so, how come so many identities are gravitating to private oranges rather than public apples?

Why haven’t open, public directory services scaled across more data types and services?

In the case of DNS, I don’t think scale is the issue.  DNS has proven its ability to scale.  I’d wager the issue is functional:  DNS is really a naming service, and many of these applications need a directory (for example, with rich searching capabilities).  DNS can’t give you that.

So DNS, whilst clearly a great success, hasn’t managed to take on naming functions from broader directory services and communications systems.  Your business card sprouts ever more identifiers!  What lessons can we draw from how things are playing out on the battlefield?

Would readers draw the conclusion that this is a result of too much, or too little ‘governance’? Do we suffer from the idea that what’s good for DNS is good for the Internet?  This strikes me as particularly dangerous, and is what, at heart, I’m objecting to.

Governance appears to entrench the technology, and make it harder to introduce alternative object naming and management systems.  There are plenty of alternative ideas, such as those devised by Robert Kahn, or the many federated ID structures out there.  They are not all functional analogues or supersets, I grant you.  DNS will long have its place.  But its institutionalisation deserves to be questioned.  Sometimes I even wonder if we’d have been better selling the whole thing to Network Solutions in the first place…  (Heresy!  Burn him alive!)

In a sense, DNS is the “Intelligent Design” of the Internet; a belief in a benevolent, bright creator of public good.  But that’s ITU/NGN/IMS-think, not the way of the Stupid Network.  An open, public network can live with or without open, public network services.

DNS was set up with a specific purpose in mind. Not as a telephone directory or yellow pages

Sure you have various people overloading dns and using it for what they want to do, at the edge - and the fact that dns gets used for this is a pointer to two things, its ubiquity and its astonishing capacity to get overloaded with all sorts of stuff for which it was never intended in the first place.

However confusing technological issues with governance issues is not a very useful way to go - nor is projecting it as a cabal. Intelligent design - well, do you think there was this big bang and packets started flying in all directions?

Seems like we’re actually in agreement, to some extent.  It keeps being stretched into things it wasn’t meant to do, which is a tribute to its designers.  The “Intelligent Design” bit wasn’t really a reference to Paul Mockapetris and the other founders of DNS, but rather to various public-interest policy wonks and campaigners who get wound up so much by DNS issues like it was life and death.  Sorry if that wasn’t clear.

The original point was that the world doesn’t end immediately when you don’t have access to DNS, and that governance effort is often misdirected because it confuses the interest of the network with the services that ride upon it.  I don’t (yet) see any argument to the contrary.

>> The original point was that the world doesn’t end immediately when you
>> don’t have access to DNS, and that governance effort is often misdirected

Nor does traffic completely stop because there’s been a pileup on the freeway.

>>  because it confuses the interest of the network with the services that
>> ride upon it. I don’t (yet) see any argument to the contrary.

We’re in agreement here.

>> “Firstly, we’re spending more and more time finding things via search. I bookmark things much less than I used to. I don’t type domain names in very often. The standard approach is to Google the approximately right term.”

- -

1. Crazy, if you don’t bookmark or type domains anymore how do you get to google.com? What if google.com was still: http://www.stanford.edu/services/websearch/Google/...

2. You are looking at the trees and missing the forest.

The internet is just a fraction of one’s daily media consumption, how do you use google or any search engine in Print, Television, Radio, Phone(IVR), Outdoor Advertising and most importantly Person to Person communication? Hey Bob to get to my site just google, it’s listing 13,433.

3.  The tagging sites are a new form of “Bookmarks” just with peer review, so if you use the tagging sites you are still bookmarking sites.

4. Goggle’s search results are taking a hit lately with all the SEO and SEM firms making pages and pages of links that go nowhere. With the tagging sites this will only get worse.

“Some sites use tricks to manipulate the search engine to display them as the first result returned for some keywords. This can lead to some search results being polluted, with more relevant links being pushed down in the result list.”

Domain names were a consequence of memory capacity limitations which no longer exist.
Uniqueness is a problem only as long as there is no restriction on the search. Registration is one way of limiting the search. Imagine you could register any sequence of words that uniquely identify you against your IP address.  You could even register multiple sequences against the same address - as long as you were doing unspoofably. For example your name, occupation and place of work would be unique for most people, as would your name and home address, your name and a list of hobbies, etc. With imagination a short string can suffice: you could still use “http://www.microsoft.com” or “combatant linguaphiles”.
Google has shown that a simple look-up table is no longer necessary. 

To Colin - Keywords, you mean?  AOL’s been doing that for quite some time

>> Domain names were a consequence of memory capacity limitations which no longer exist.

Not Really True.
Step away from the computer, humans like to name things and Domain Names are just another example.

1. Users needed an absolute way to get from point A to point B. (IP did this)
2. Users needed an easy way to remember the point B identifier (domains names did this)

Just as people in your office don’t say I need to talk to “blue eyes, brown hair, 6 foot, male, Christian, 25 year old. They say I need to talk to Colin. The internet is no different.

To Illustrate the point:

In 1969 while speaking at a small scientific conference, Sir Roger Penrose, a Cambridge physicist announced his discovery of what he called a “gravitationally totally collapsed object.” The world yawned.

Months later, he changed his description to a “Black Hole” and the news of his discovery raced around the world. Today, the term Black Hole is a part of the world’s working vocabulary.

http://en.wikipedia.org/wiki/Names
http://www.stanford.edu/group/stanfordbirds/text/essays/Taxonomy.html

My point is not that a list of attributes is required, but that hieroglyphs such as dot, com, http,  au, etc. are not.
Hence the googlewhack example “combatant linguaphiles”.

It would be nice, I agree, if “Colin” was my unique identifier :-)