The problem is, of course, that it's NOT human-to-human communication. It's machine-to-machine communication, and human-to-machine communication, and DNS was designed to create a mnemonic representation of a way to reach a machine. "Hey, tiger's down!" "Hey, one nine two dot one six eight dot one oh three dot two five three's down!" (Even if you take an example with 192.168 being where ALL the addresses in the network come from: "Hey, one oh three dot two five three's down!") In order for the machine to be able to determine what to do, it must be programmed by the humans that created it. In order for the machine to be able to determine what to do in a situation that has more than one possibly-correct resolution, it must be given a set of rules in order to determine what needs to be done. I will admit that it is possible (not plausible, but possible) for even an email system to be programmed with enough intelligence to be able to deal with the conflict. However, the key word is 'system' -- defined as "all pieces in a computing environment that contribute to a given piece of data being processed the way the humans that are using it desire it to be processed." Which includes all mail servers (speaking SMTP), as well as all mail clients (speaking SMTP, POP3, IMAP, whatever else), and name resolution (DNS, /etc/hosts, NIS, whatever else), and even the underlying virtual circuit technology (TCP/IPv[46]). The entire system must be programmed in a way that is consistent with how the user wants it to work... and getting even two sites to upgrade to a newer version of sendmail at the same time is difficult at best. Much less to get two sites to change their DNS configuration at the same time. The point is: In order to do our jobs, we have to simplify these complex systems we're responsible for as much as possible. At least with a globally-shared root zone, we're removing THAT piece of complexity from the equation so we can determine what needs to be done at a higher level. (Am I lazy? Perhaps. But I'd rather be lazy than crazy.) -Mat Butler Speaking for himself, not for his company. -----Original Message----- From: Vadim Antonov [mailto:avg@kotovnik.com] Sent: Tuesday, March 13, 2001 9:06 PM To: Geoff Huston Cc: nanog@merit.edu Subject: Re: Statements against new.net? Of course, one may choose to treat RFC as a gospel, but to me (and i hope to anyone interested in how cognition works to the point of actually getting acquainted with the relevant research) the attached passage sounds quite like a bunch of random noise :) Mostly because it assumes that human-to-human communication is a reasoned process, concerned with consistent intepretation. In fact, most of what makes, for example, art interesting is that it does not have a singular, well defined interpretation. --vadim PS This one, i guess, is brought to you by the Society Against De-Humanization Of Internet Users <tongue firmly in cheek> PPS Yes, I think any form which _restricts_ potential models of communication is bad. Such as forcing communications to be moderated by a singular hierarchical structure. This whole thread won't be there in the first place if the scheme actually worked well in the real world. Hierarchies do not scale and cannot adequately tolerate internalized adversity. On Wed, 14 Mar 2001, Geoff Huston wrote:
At 3/14/01 07:56 AM, Vadim Antonov wrote:
That is based on the assumption that consistency is necessary or desireable :) Of course, it is dear to an engineer's mind, but the case from the sociological point of view is far from clear-cut. In fact, way too many woes of human societies can be (at least indirectly) attributed to the misguided attempts to enforce consistency.
This assumption is explicitly addressed in the RFC - I quote:
------ 1.1. Maintenance of a Common Symbol Set
Effective communications between two parties requires two essential preconditions:
- The existence of a common symbol set, and
- The existence of a common semantic interpretation of these symbols. Failure to meet the first condition implies a failure to communicate at all, while failure to meet the second implies that the meaning of the communication is lost.
In the case of a public communications system this condition of a common symbol set with a common semantic interpretation must be further strengthened to that of a unique symbol set with a unique semantic interpretation. This condition of uniqueness allows any party to initiate a communication that can be received and understood by any other party. Such
a condition rules out the ability to define a symbol within some bounded context. In such a case, once the communication moves out of the context of interpretation in which it was defined, the meaning of the symbol becomes lost.
On Wed, 14 Mar 2001, Mathew Butler wrote:
The problem is, of course, that it's NOT human-to-human communication. It's machine-to-machine communication, and human-to-machine communication, and DNS was designed to create a mnemonic representation of a way to reach a machine.
If it is machines communicating there's no need to do any mnemonics. In fact, it is still humans communicating, with the aid of the machines. So... we have two design constraints: 1) people need to be able to locate and revisit somethings in the network 2) any meaningful hierarchial labeling of the real world is quite impossible, and runs into problems of scaling, adversity, and entrenched notions of ownership. Propping up DNS as-is only guarantees that the whole thing is going to be pushed off the cliff on the second side. Neither it is very good on the first count. So, instead of trying to fix the broken concept, and raising ridiculous protests and indignation when someone tries to rock the boat, isn't it easier just to go for a real permanent solution? Which is to replace "mnemonic" DNS with something deliberately mnemonic-free (like numeric strings :) and leave the human-interaction part to the better and wildly successful concept of navigation in a general graph. The exising deployed software is nearly sufficient (and in many cases quite adequate) to make this mode of communication easy to use. My proposal is to create a special hierarchy (similar to tpc.int) which can _only_ be used to register numeric "names" on first-come first-served basis. The "current" DNS then can go down in flames, for all i care. Actually, I think this is inevitable, since some day someone will find a way to win a lawsuit against the whatever central naming authority is. Anyone who thinks numeric IDs do not work when "better" alphanumeric IDs are possible needs to take a look at the ICQ. It is _very_ successful in case you didn't notice. And so is telephony. And in case you didn't notice that most people in the world do not speak english, and do not use latin script, let me tell you the simple (and quite obvious to someone who is not spoiled by American isolationism) fact: for the majority of world population ASCII strings are only marginally better than numbers in being "mnemonic" - and it is much easier to pronounce numbers in a native language. --vadim
participants (2)
-
Mathew Butler
-
Vadim Antonov