Highlight: Robin Hanson's more owner-forgiving modified Harberger tax

The fee for a bid is the possibility that if you bid dishonestly you get called on the bid and you end up having paid more for something than the value you get from it. So people casually putting up bids for things just in case is imo intended behavior. If this design leads to people paying fees that are too high, then that means that the tax rate was set too high and should be lower.

I fully agree with Nick here.

ENS will eventually be a critical piece of the Ethereum infrastructure and if we want to see mainstream adoption, people need to be sure that they can rely on their ENS address as a part of their digital identity. Especially if you know how decision processes in many conventional companies work, you realize that no one will ever give ENS a chance if an address cannot be guaranteed or if the company cannot at least plan with a certain duration during which they are the sole owner.

Imho, the proposed fix for the Harberger tax is not a fix. While I like Vitalik’s idea of not privileging people of today over people of the future in general, the proposed Harberger tax seems just like a sophisticated way of again privileging people with money instead, or in other terms: people who have early ethers or a pile of ‘old’ fiat money. And that is because @jefflau 's comment is plain true when he says:

If someone is long on ETH, attacking a domain by adding a bunch of ETH to that name is effectively free, especially to a name that cannot be realistically given up (i.e. mew.eth due to subdomains). I could see this happening just to ‘grief’ high profile projects rather than any intention to buy the name to scam, possibly by a competitor with a lot of ETH.

And thus the sales in the proposed system are only in theory owner-initiated. In reality, the sales are only owner-initiated until the company or the individual runs out of cash - even if this company or individual is the entity that puts the address to its best use. Also, (and I appreciate this discussion and don’t want to sound rude but) the proposed fees I read here (although they are only first ideas: 10 $ per year per day, doubling even ‘only’ every 3 months, …) are completely over the top and would most likely ruin ENS as a system - you certainly read how people reacted when we discussed the fixed annual fee and whether it should be 5 or 10 $ per year? No one would understand why a common DNS domain costs a few dollars per year while an ENS domain costs a fortune to maintain.

By the way: Harberger tax is just too complicated for mainstream adoption - obviously for none of the tech-savvy users in this forum and not for the smart blockchain startups. But it’s paradoxical that we try to make the use of ENS easier and, at the same time, propose a at least non-intuitive billing system.

So from my point of view, the most recent solution, where we have an annual fee that can be slightly adjusted by the ENS team if necessary is still the most feasible approach.

There are compromises to consider. For example, there are plenty of domains that cost $5 per year that really really don’t need to cost $5 per year. So I’ll propose a “capped modified Harberger tax”: a domain by default costs nothing to rent, but people can submit bids, and if you submit a bid you have the ability to raise the rent of a domain to a maximum of $5 per year (that might require a $500 bid). This would be a strict net improvement to users, because users would not pay fees for domains where the fee really isn’t necessary and is a pure economic deadweight loss (eg. long 12 letter domains), but the fees would still exist for domains that really are rivalrous. From there you can talk about raising the maximum to eg. $50 per year if someone bids $5000 for a domain.

Is there any argument why such a capped system would not be an improvement over the status quo?

Your suggestion solves names where usd 5 is too high but dont solve names where usd 5 are too low. It may add a slight improvement but not worth the complications for the sake of trying out new system where the fee is affordable. Again, may be a good application to 1-2 chars where we expect the annual fee are expected to become expensive.

Got it, so sounds like “try it out in a more limited context and see how people respond to the mechanism”.

I think the idea is to set the limit to higher than $5. e.g., set the limit to $50 / year, which is still less than many TLDs. I would tentatively argue that we should set the limit closer to $100 or $150, which is just about the most expensive class of TLD available. The vast majority of people will end up paying far less than this (likely close to $0), but we would have a system that allows us to figure out who should be charged for their domain (because they are competing for a good name) and who should get it for free (because they took a name that no one cares about).

Wouldn’t that even encourage blanket purchase 1000 of names for free now as it becomes easier to sell to others?

You would need to lockup enough capital in each name to protect against someone draining your paid up capital and then purchasing the name for cheap. Though, 1% tax rate may be too low to defend against that since you would have 100 to 1 leverage on each name. One can even imagine 50% or 100% tax rate so the person holding the name really has no advantage over the person trying to buy the name. With 100% tax rate you can buy a name by just being willing to pay more annually than the current holder, up to whatever cap is set.

This works better, but I’m still concerned about names where the main value accrues to people other than the nameholder. Do we really want to put a (potentially very large) price tag in front of every operator of a common-good domain that issues subdomains? We can build a contract that either raises rent from, or pays out to, the owners of those subdomains, but then we’re back to them having to pay constant attention to the system.

With a hard cap on rent, we can mitigate this somewhat. One thing that we need to bear in mind, though, is the design of the permanent registrar: the registrar itself allows its controllers to extend the rental period of a name, and to register new names, but not to shorten that period or transfer the name. Any innovation needs to be implementable as a controller under that regime.

If you are issuing desirable subdomains, I feel like you should be charging for them in order to cover your costs. If you have wall.eth and are issuing subdomains to people, you really should be designing your system to be sustainable so you can afford to cover your costs.

Generally speaking, I’m not a fan of trying to design a system that is optimized for “the public good” because it will almost certainly be exploited by someone. In the case of names, this exploitation occurs in the form of squatting. While I can appreciate the desire to try to hand out names to those most deserving of them, I think that is an unrealistic goal without massive centralization (where the centralized authority gets to define “good” and is capturable).

We could leave the caps as they are now but have Harberger tax up to that cap so that most people pay less (since ENS is not a profit seeking venture, this seems reasonable). This still gives people a maximum that is no different than today (thus only a reduction in cost compared to today) but reduces cost for many people.