I don't think blockchain is a suitable for DNS (or for anything actually).

To "run a blockchain" requires a lot of infrastructure. At a minimum I think you need communication between all the participants (otherwise how would one tell the others it has successfully produced the next block?) and you need some kind of pool of waiting registrations that they can all access (otherwise what would they build a block from?).

The block chain is just a ledger, and a ledger is a terrible format for DNS data because it requires scanning every ledger record to find a match (so it scales linearly with the number of times anyone modifies a DNS entry). To solve this any real DNS will need to covert the ledger into an internal database. I would think all this complexity would raise costs, not lower them.

The existent DNS (a simple distributed hierarchical database) is replaced by a voluminous distributed ledger system. This change by itself doesn't resolve any of the problems you mentioned.

You've said a few other things that don't make technical sense:

• "hosting fees will decrease": A .com or .org DNS domain name is about $10 a year and is independent of the amount of traffic you have. Different DNS subdomains also charge different amounts, so it's mostly a nominal fee.
• "A domain must keep a minimum amount of traffic": There is no accurate tracking of traffic to a domain name because it's often cached (this is why you're advised to wait multiple hours after making DNS changes to see the effects).

I think you are conflating hosting with DNS. The DNS is just the resolution of a human readable string to a bunch of keywords (e.g. www server addresses, mail addresses, metadata tags, etc). Hosting is providing the necessary servers and bandwidth to deliver the services (like email, websites, torrents, etc).

I want to point out that systems don’t have to have a blockchain to be decentralized. Aside from that distinction, you’re right that theoretically, there’s a decentralized version of most computing systems. They’re really two sides of the same coin.

My knowledge is pretty limited, but a fundamental aspect of the self-hosting problem is the scarcity of ipv4 addresses and that ipv6 adoption is incomplete. It’s non-trivial to set up a public server when most home networks are behind a NAT

One of the few use cases that really makes a lot of sense and has promise imo.

There's a lot of confusion, other comments addressed it pretty well. Meanwhile you might be interested in looking into IPFS. It's kinda like the internet over torrent.

How could laws be enforced to public block chains, though?

Isn't the point of block chain to have full integrity, "no censorship" and "no external control", or something like that?

I like how .b32.i2p and .onion does it. It's all randomly generated and free to register/deregister, but also allows for some convenience with addressbooks (i2p) or headers to automatically switch from the clearnet address to .onion.