The Future of ENS Incentives
This is part of a series of articles I’m writing on the Ethereum Name Service (ENS) in retrospective of the ENS Workshop in London, August 2017. I’m going to encapsulate some of the ideas that were produced from this event, to help the larger Ethereum community prepare for changes and contribute to ongoing conversations. If you’d like to get more involved with ENS, you can always visit the Gitter channel, or start a discussion on the Github repository.
There are a lot of variables that we considered during the 2017 ENS Workshop, but one mechanic in particular stole the show for seeming to hone in on one of our main goals:
Now in case that’s a bit terse, I’m going to space it out a bit more, to give you a sense of what we’re thinking.
ENS Goals
The first version of the ENS was created with a simple Vickrey auction that converted into a deposit that could be released in exchange for the name after one year. This achieved the basic goal of distributing some names, ensuring disputed names went to the person who valued them the most, and that no money was burned before we thought through any fee systems more carefully.
After discussing goals for a while, we came up with a few items, which defined some amount of our focus over the workshop.
- Rent for Utility / Incentive to Release.
- Simplified auctions for a single bidder.
- Certainties for Owners
- Guarantees for Subdomains
Let’s address those in reverse order:
Guarantees for Subdomains was covered in my previous article, and you can see how we’re preserving those guarantees there.
Certainties for Owners seems fairly straight forward to me. As long as we make the smart contracts highly public, the owners know what they’re agreeing to. Currently those guarantees include the ability to redeem their deposit after one year, and the ability to transfer ownership of that name at any time.
Simplified auctions for a single bidder is a topic we explored to some degree, but I think we somewhat discarded it as a priority. While simplifying the auction process would be a nice user experience, the auction period is valuable for resolving disputes, and so I believe the current consensus is that users seeking total convenience should pursue purchasing domains from a third party, which can come with subdomain guarantees.
Lastly, and most interestingly, is the Rent for Utility / Incentive to Release. I think the important thing here is the incentive to release. The issue is that for all uncontested names, the initial ENS registrar gave away ownership of names for a minimal price of 0.01 ether. For some names this may be excessive, but in the case of names that may represent a brand or person, (ie unrealized disputes), this name squatting could simply lead to a class of name-barons who set the prices, which everyone else is forced to pay over, creating a profit for those squatters.
Vlad Zamfir described this goal in an interesting term, as “economic efficiency”, which (as I understand it) meant that the person who values it the most should pay as little as possible to people who don’t value it as much. (I’m asking him on Twitter to word it again, he’s very good with these types of phrases.)
Available Mechanics
We considered a variety of parameters that might help us achieve those goals.
- Auction Model
- Entry Fees (sunk cost)
- Deposits (refundable)
- Rent fee (ongoing sunk costs)
Because we want to encourage people to deregister names that aren’t actually valuable to them, we mostly agreed some kind of ongoing fee was desirable. We did not totally settle on whether this fee should be burned, or whether we would cede it to a trust or foundation. One thought was that these fees could be used to fund dispute resolution, which is a topic for another article.
Deregistration Target
The most novel mechanic suggested over the workshop was by Vlad Zamfir, and it goes like this:
- The ENS DAO sets a target deregistration rate (let’s say, we target 3% of domains to be released every quarter).
- If this rate isn’t achieved, the name rent rate increases.
- If this rate is exceeded, perhaps the name rate rent decreases.
Or as he would put it (one more time):
Now, this is just a mechanic, and leaves a lot of parameters open for filling out. For example, are rent fees proportional to the winning auction price, or do all names have a flat rental fee?
Other Options
Some of the other options that were considered:
- An immediate fixed fee on purchase prevents griefing by winning auctions then releasing names.
- Rental fees could be automatically deducted from a deposit.
- In the case of an owner defaulting on a name payment, there would be a period of suspension before the name is released, to allow clients to warn users of a potential name transfer, and to give the owner time to renew the name.
- There was general resistance to the notion of someone being able to bid a name away from a current owner, for obvious security reasons.
These are a lot of loose tools and mechanics that we began assembling, and there is a need for deeper analysis of the various incentives and tradeoffs when these parameters are set to different values.
Hopefully by writing this in public, more people can see the issues at hand, and chime in with mechanics that might be useful for the ENS.