It's clear from the very article of this submission that this is the case. For instance, the I2P fields are combined in a structure that best represents the resource on the I2P network. A different structure is necessary to represent a Tor or freenet resource identifier.

If Internet naming is going to be fragmented, I'd rather it not be wasted on Alternative DNS roots, which fall privy to the same fundamental problems as our current domain name system. A Namecoin system cannot be so easily censored.

But that's my point: why? I'm not saying they should implement a DNS root, I'm saying I don't see the benefit in using a new format, when:

1. The DNS packet format is compatible with existing resolvers, which eases the transition (for example, someone could use the new system without installing anything, by trusting a Namecoin resolving server) 2. It uses less space for the common use case - JSON is smaller than e.g. XML, but it's still hardly a small format. 3. It's extensible for other use cases (Tor, I2P) without bloating the common - see EDNS

Frankly, I think they're reinventing the wheel, poorly, by designing their own format.

Names registered with namecoin have to depend on some blobs which can be updated by the name owner. The blobs need a format of some kind, and since they're already using a different protocol than DNS to achieve this P2P system, they might as well use JSON.

Also, as I stated in my comment, the structure of JSON fits better for this type of system than DNS's formatting. DNS requires redundancy to describe round-robin, and it's not easy (or elegant) to express complex structures in DNS, compared to JSON.

One example of this in practice is the I2P field I mentioned before.

It is already possible - http://dot-bit.org/HowToBrowseBitDomains#Replace_your_DNS_se... . There are two traditional DNS servers that resolve .bit addresses.