Pay your fee once, and the domain stays active forever.
The cost of running the servers for a top level domain are really small, and probably going down with time, so it should be possible for the domain registrar to keep operations going with just a trickle of new registrations.
As a compromise, perhaps you could charge a nominal fee for domain name ownership transfers.
as such, a forever registration would have to be a quasi-pyramid scheme where the initial users keep their domains only so long as there are more new users coming in each year than existed in all the past years combined (or something similar depending on charges).
also, what good is a forever domain when it doesn't also have forever hosting? You die, the domain continues on forever, your hosting provider stops hosting and now no-one else can ever use that domain. Kind-of a jerk move in the end.
So not that different than today. You stop paying for your domain and squatters make sure no one can ever register that domain, for a reasonable fee, ever again.
It doesn't have to be a pyramid scheme. The registrar can also invest the money, and use the interest as revenue. So instead of paying 40 USD annually you pay 1000 USD for the initial registration. 4% interest of that would be 40 USD.
Beyond domain squatters, who would benefit from that?
It means I could maybe publish my memoirs and family tree and have it still available in 500 years.
It also makes web domains like other property - you don't purchase a nice watch and then expect it to be reassigned to someone else if you don't pay your annual watch renewals.
Your property is both relatively scarce, limited, and requires significant service compared to random domains on some random tld.
Everyone would love to have a three- or four-letter domain but, like land, they're not making new characters for the ASCII alphabet.
This usecase makes no sense at all. If you plan to have a server going after you're no longer able to maintain a server, your only option is to pay someone to manage your infrastructure and/or services. Once you're paying some company to manage your infrastructures, managing domains becomes a subitem in your invoice.
To get around the problem of a numeric TLD, I suggest using Roman numerals (and telling Myanmar it has to go back to using .bu).
While I'm sure this would be great for your ego, it makes no actual sense. Very few documents from so long ago are available, not because it was somehow miraculously impossible to preserve documents but because overwhelmingly people don't give a shit.
You are demanding that future people should give a shit, and they aren't going to oblige.
I for one love to read historical documents and primary sources. I'm glad some people in ancient times had the foresight to record a bit of their lives in the hope that future generations might find it interesting.
I’ve always found personal accounts and memoirs to be the best kind of writing. You get to read beneath the surface between the person they were, the person they wanted to be and the worldly circumstances they found themselves in.
You always have the option of committing your memoirs to stone, absorbing the long-term cost up front yourself.
Of course, go far back enough (I'm not sure if 500 years is enough) and mathematically speaking, the number of your ancestors will exceed the population of the world, which shows some other issues that you run into.
That said, this is not something I expect to happen in the regular Web, this seems to be a job for IPFS, blockchain or whatever. With plain HTTP, even if you have your forever domain, you'd still have the issue that the webserver it is pointing to can change, break or just disappear.
What's the benefit of a UUID over a raw IP address?
Not a technical requirement. Subject alternative name (SAN) certificates support IP addresses.
Unlike the Common Name, which is arbitrary human text (so e.g. " oops.example" is a valid name despite that space at the start, as are "c-programming.example\0\0\0" and "Ł0Ł.example") the SANs are defined so that it's harder to screw up and a machine can reliably just bit-for-bit compare the data. There are two main kinds: IpAddress SANs are defined for the IPv4 or IPv6 address just as bits, you can't express 200.300.400.500 as an IpAddress, whereas that's a (stupid but) valid Common Name; DnsName SANs are defined using a restricted character set and as specifically the ASCII LDH DNS hostname subset, which makes it much less tempting to try to incorrectly write a Unicode name here, what's actually required is the ASCII LDH name, ie Punycode.
Registering a domain is not an issue to anyone.
Some domain name registrars even offer multi-year subscription plans for those who don't want to bother with the 5minute work it takes to extend a subscription, and send out warnings months in advance to renew subscriptions.
I expect all of them, and I'd be very surprised if any domain name registrar didn't offered this feature.
For example, both Namecheap and GoDaddy allow customers to extend domain registrations up to 10 years in one go.
Network Solutions does offer a 100 year plan (and used to offer a 1000 year plan, but I just checked and that one is gone) but the way the implement is by registering for 10 years and then every year for the next 90 years adding a year.
You only really have the domain registered for 10 years, and you are betting the Network Solutions will still be around for 90 more years, and Network Solutions is betting that the price they have to pay for domains in that TLD will not have gone up so much that the total cost of all those extensions is more than what they earned by having your initial large [1] upfront payment invested for 100 years.
My recommendation: register your domain for 10 years. Don't turn on auto-renewal. Add an annual "Check my domain registration" item to your calendar. When that items come up each year either (1) add another year to your domain, or (2) check to see if it is near expiration and add years.
[1] $1600 for 100 years of a random .com.
Do you remember how much the 1000 year plan costed?
We had that for the .ro TLD. What happened was that a lot of useful domain names got acquired by random entities and sat there unused. And i don't mean "you can have this domain for $5K" unused, properly unused.
They switched to a yearly fee model (about 1/5 of what the forever fee was, plus credit if you paid the forever fee in the past 5 years) and that seems to have cleaned up the space a bit.
Maybe make a new TLD specifically for infinitely lasting names, say .permanent, and make all top level names under that be 128 bit numbers in some text format randomly generated when you register the domain such as hex or base64. E.g., f560618f355b0ddf560456d41d635e3d.permanent or 9WBhjzVbDd9WBFbUHWNePQ==.permanent.
With those names it doesn't matter that they get locked up forever with the initial registrant.
> As a compromise, perhaps you could charge a nominal fee for domain name ownership transfers.
When the name is generated also generate a public/private key pair and associate the public key with the name at the top level servers. Someone is considered to own the domain if they can prove that they have the corresponding private key.
The top level would allow someone who has the private key to ask for a new key pair to be generated to replace the current one.
Want to transfer ownership? Give the new owner the private key and they can use it to do a key change.
This should allow making the top level service entirely automated and cheap to run.
Vitalik Buterin wrote an interesting analysis about how to allocate scarce domains, contrasting unstoppable’s approach with recurring demand based fees:
https://progressandpoverty.substack.com/p/should-there-be-de...
Requiring periodic renewal ensures that the entity that is responsible for the domain is still alive, and still wants it. Allowing an entity to register a domain forever creates the possibility of zombie domains, registered forever through a registrar who has long since gone bust, by a registrant who long ago lost their password... but the registry is obliged to still maintain the last state of that domain's NS records forever.
Nobody would be able to initiate a transfer; nobody would be able to buy the domain off the owner... the resource would be permanently dead.
The one-time cost for a domain should be substantial, with the capital put into long-term investments like index funds, which are then used to pay domain renewal (and maybe static webhosting) costs year after year.
If people can set up trust funds for their families and estates, I don't see why it's not possible to do a niche "trust fund" for websites.
Happy to have someone steal my idea. :)
Forever is good. Especially if you know it is forever than you say thing that you want to say forever.
We can't expect the dead to continue to exert pressure over the living.
I dont care if it eventually gets lost amongst the huge amount of content but I still want my stuff to be out there.
IPNS "records" have a TTL so you need someone with the key actively re-publishing them. I guess in theory you could publish for a very long time but you risk issues of someone maliciously re-sharing an older record that hasn't expired.
I guess once you die you can hope that someone has recorded the last version so you can "move" from IPNS to a regular CID. But IDK how much I would want to rely on that for long-term legacy especially if the TTL on your IPNS record is shirt. Although I guess IPNS records are small enough that it would be fairly easy to set up a service that recorded all published IPNS records. Maybe publish it as a merkle tree so it is infeasible to tamper with old records.
But neither of these things will ever happen, for approximately the same reason.