Increase premium fee or have a new auction mechanism

bbc.eth is bought at 100K

$100K might not be enough. Especially a bunch of premium names are going to expire.

ENS should take the profit of these auctions. Otherwise people try to front-run others by paying high gas fee, which only benefits the miners.

With the current linear decrease, there’s a tradeoff we have to make; at a $100k starting price, the price decreases by about $150/hour, meaning that most domains’ purchases are compressed into the last day or two of the auction. If we increase it further, that gets worse.

We’re working on a new version of the pricing oracle that uses an exponential decay; the premium can start at something ridiculously high, like $100M or even $1B, and decrease by 50% each day. This ensures that high value names are adequately covered, but low value names get plenty of time as well.


yes, I think decreasing by percentage, instead of a fixed amount would make more sense.

To me Dutch auction seems to be the best option, given mechanics and various constraints around it, such as complexity of code required.

@garypalmerjr insists that English auction will be better, because there is better visibility for all participants. In practice that means that once you bought something of Dutch auction then item is gone and other agents, who also might’ve participated, didn’t get a change to do so.

I think pure linear English auction is problematic, because in the final phase of auction there will be gas bidding war.

The solution here is “Candle modified English auction”, it was successfully implemented by Polkadot like so Parachain Slots Auction · Polkadot Wiki

In medieval ages, to prevent “last second bids” the auction was conducted as long as the candle is burning, last valid bid is the winning one. Although this approach is also not without its limitation and difficulties.

Frankly in my opinion, Dutch auction is the most efficient approach given ENS

→ simple to understand
→ price discovery is efficient
→ easy to build
:slightly_smiling_face: whats not to like :upside_down_face:

If you really need that name, and price meets your “reservation” price, then just buy it and use. Pressure resulting from anticipation that someone may get it will result in the highest possible price, in turn if you paid “pretty penny” for that name, you might as well use it properly.

The reason I mentioned “Candle modified English auction” is to gauge community’s opinions on that option, its possible that it might be viable option, but it still requires a lot of research. There has to be a lot strong reasoning to support implementation of Candle auction vs exponential Dutch auction.

I mostly agree, but while a Dutch auction is “easy to understand”, it’s far easier to misunderstand. There are countless examples of people on here, the Discord server, and twitter, all smart people I think, who completely misunderstand how the Dutch auction works or what the point of it even is. It can take a long time and lengthy explanations/discussions for people to finally “get it”.

Myself included. I didn’t get it at first either. The technical mechanism of the Dutch auction is simple, but the game theory is just not what people are used to, so it takes time to think through it and realize that it’s functionally the same outcome as an English auction. And most people do not take that extra time to think through the nuances. They take a cursory glance, assume it’s moneygrubbing devs helping the whales get names first, shout at the ENS team, and then leave in a huff.

1 Like

I think 100k is enough. Maybe this purchase was just an old bot that didn’t get the memo, instantly sniping the domain without caring how much it cost?

Checking OpenSea and comparing it against historic sales, the highest sales higher in nominal value range from 50 to a 100ETH. Some of these sales happened 2-3 years ago though, making the nominal USD value lower than today’s.

I think a 100k$ is just fine. Especially because the few historic sales that are higher than this in value, were speculative sales between peers. I wouldn’t remotely expect 100k$ sales to keep consistently happening and the money rolling in the treasury’s bank.

1 Like

I’ve noticed the same with users. In fact, we had one person in the Discord earlier today who did write a comment about predatory pricing/moneygrubbing devs who later removed the comment and apologized after he had read through the EP5 discussion page we linked, saying he understood it.

A lot of people don’t read through it though, and are stuck with their first impression.

Even if we’re happy with $100k, an exponential curve will ensure more time is available for lower bids. With a linear model that starts at $100k, the period from $2000 to 0 takes about 13 hours. With an exponential model starting at $1B, the period from $2000 to 0 takes about 9 days, giving people much more time to select lower-value names.


I am expecting the revenue of ENS is going to be so high this year…

Oh yeah sure. No argument there about making the curve more user friendly. Was commenting only about the initial value.