There do exist systems of roots that are loaded with zone files that are either replicas or supersets of the NTIA root zone. Despite nearly biblical degrees of techno-wailing, gnashing of teeth, and rending of garments, these have not brought the internet to a halt or even caused any clearly demonstrable problems to their users or, more importantly, to those who chose to use the legacy root servers.
If the presence of competing roots - something that anyone can set up and anyone can chose to use - were, in fact, able to cause harm then that would reflect an excessive fragility of DNS that ought to be repaired rather than left as a point of vulnerability and potential attack.
The ICANN assertion that there be exactly one set of root servers - a catholic root - is both contrary to the end-to-end principle and the natural desire of people to form communities, often communities based on a common language and thus common space of names.
Just as it takes two to tango it takes two to cause a naming inconsistency. And ICANN has on at least one occassions knowingly intruduced inconsistencies - when ICANN introduced .biz on top of a prior established use of that name in other DNS root zones.
How are such inconsistencies to be resolved?
There are several methods.
There is the method of having a mini-government or god of names, such as ICANN has become.
There is the method of lawsuits over trademarks - a method that works only to the extent that name spaces can be considered as trademark-able.
Then there is the method of allowing users (or their agent's, i.e. ISPs) to chose.
I prefer that third method.
Has the internet become a Procrustean world in which users are denied the right and ability to shape their own internet landscape of names?
The human desire to communicate creates a force that will cause undesired inconsistencies in names to wither and fade. But that fading of inconsistencies does not necessarily mean that the sets of root servers that act as the portals into those name spaces need to become one.
One can go further and suggest that it is possible for end users to bypass the whole issue of which system of roots to use. A while back there was a system called "Grass Roots" - it was a web based package that let users pick and chose the TLDs they wanted to include in their view of the internet name space. In cases of conflicts the user got to chose which version he/she wanted. The result was a body of information that let the user act as his/her own root. (The system also allowed the underlying server information to be periodically refreshed.) This system worked quite nicely.
Re: The InternetsJoi Ito – Jul 22, 2005 1:44 AM PST
There are several methods.
There is the method of having a mini-government or god of names, such as ICANN has become.
There is the method of lawsuits over trademarks - a method that works only to the extent that name spaces can be considered as trademark-able.
Then there is the method of allowing users (or their agent's, i.e. ISPs) to chose.
I prefer that third method.
In many countries, the government controls the ISPs. Also, ISPs compete with each other. I would hate to have to ask questions like, "What ISP do I have to join to access your service?" "What is your address on my ISP and how do I access your service?" "How do send email to you from my ISP?" "In country X, you can't access our service because they have blocked it." Doesn't scenario 3 cause this kind of addressing fragmentation?
Also, doesn't this give an enormous amount of power to the big ISPs? Won't Microsoft take over an ISP and only allow access to "trusted" domains?
competition policy is one of the main concerns with the way this is moving.
lots of countries have government owned majority telcos with an [un]healthy interest in preserving their monopoly, and in the position of competing with other ISPs in the country who are also forced to be their customers.
I believe that you are jumping to the conclusion that consistency of naming requires that there be one and only one set of root servers.
What is desired is consistency not singularity.
The desired characteristic, naming consistency, can be achieved even with separate systems of roots.
See my paper My presentation on Internet Naming to the US National Research Council (Powerpoint) on the matter of name consistency.
There do exist systems of roots that are loaded with zone files that are either replicas or supersets of the NTIA root zone. Despite nearly biblical degrees of techno-wailing, gnashing of teeth, and rending of garments, these have not brought the internet to a halt or even caused any clearly demonstrable problems to their users or, more importantly, to those who chose to use the legacy root servers.
If the presence of competing roots - something that anyone can set up and anyone can chose to use - were, in fact, able to cause harm then that would reflect an excessive fragility of DNS that ought to be repaired rather than left as a point of vulnerability and potential attack.
The ICANN assertion that there be exactly one set of root servers - a catholic root - is both contrary to the end-to-end principle and the natural desire of people to form communities, often communities based on a common language and thus common space of names.
Just as it takes two to tango it takes two to cause a naming inconsistency. And ICANN has on at least one occassions knowingly intruduced inconsistencies - when ICANN introduced .biz on top of a prior established use of that name in other DNS root zones.
How are such inconsistencies to be resolved?
There are several methods.
There is the method of having a mini-government or god of names, such as ICANN has become.
There is the method of lawsuits over trademarks - a method that works only to the extent that name spaces can be considered as trademark-able.
Then there is the method of allowing users (or their agent's, i.e. ISPs) to chose.
I prefer that third method.
Has the internet become a Procrustean world in which users are denied the right and ability to shape their own internet landscape of names?
The human desire to communicate creates a force that will cause undesired inconsistencies in names to wither and fade. But that fading of inconsistencies does not necessarily mean that the sets of root servers that act as the portals into those name spaces need to become one.
One can go further and suggest that it is possible for end users to bypass the whole issue of which system of roots to use. A while back there was a system called "Grass Roots" - it was a web based package that let users pick and chose the TLDs they wanted to include in their view of the internet name space. In cases of conflicts the user got to chose which version he/she wanted. The result was a body of information that let the user act as his/her own root. (The system also allowed the underlying server information to be periodically refreshed.) This system worked quite nicely.
In many countries, the government controls the ISPs. Also, ISPs compete with each other. I would hate to have to ask questions like, "What ISP do I have to join to access your service?" "What is your address on my ISP and how do I access your service?" "How do send email to you from my ISP?" "In country X, you can't access our service because they have blocked it." Doesn't scenario 3 cause this kind of addressing fragmentation?
Also, doesn't this give an enormous amount of power to the big ISPs? Won't Microsoft take over an ISP and only allow access to "trusted" domains?
competition policy is one of the main concerns with the way this is moving.
lots of countries have government owned majority telcos with an [un]healthy interest in preserving their monopoly, and in the position of competing with other ISPs in the country who are also forced to be their customers.
http://igov.apdip.net/ORDIG_Policy_Brief.pdf
http://igov.apdip.net/ORDIG_Paper.pdf
recommended reading I'd say