Designing a Nounish DAO: Analysis of Auction Frequency
0xeB9f
July 21st, 2022

Written by gumfather (@_gumfather). Analytics help from notawizard.eth (@0xdoing)

Introduction

Nouns is an NFT project on the Ethereum blockchain. Launched in August 2021, it has grown into one of the most exciting experiments in community formation and governance in DeFi today. While there are many unique aspects of the Nouns project, perhaps the most iconic is its distribution mechanism: exactly one Noun is auctioned per day in perpetuity, with the proceeds flowing directly to the community treasury to be managed by Nouns holders. This mechanism has generated significant capital inflows with remarkable durability - Nouns routinely sell for 100+ ETH, helping to bootstrap a treasury of more than 25,000 ETH by July 2022 [1].

Given the initial success of the Nouns project, it is no surprise that it has already inspired rich experimentation as other communities iterate on its essential characteristics. One such example is Lil Nouns, a Nouns fork that reduces auction duration from 24 hours to just 15 minutes. As more projects consider a Nounish distribution mechanism, the question of optimal auction frequency is likely to become increasingly debated.

In this post, I will evaluate the impact of auction frequency on a variety of key metrics: revenue, community participation and community growth. Critically, this evaluation attempts to assess the impact of auction frequency on these metrics holding all else equal (e.g., art, provenance, quality of initial team).

For each key metric, I provide an analysis from first principles on the potential impact of auction frequency. To bolster my arguments (or in some cases, identify limitations), I reference data from Nouns and Lil Nouns. Ultimately, I provide a framework for how various types of DAOs and / or projects who are considering a Nounish auction structure might determine their optimal auction frequency.

Revenue

Revenue is a critical consideration for any DAO or project. In a Nounish structure, auctions are the direct source of recurring revenue. In this section, I evaluate the impact of auction frequency on treasury growth and provide a perspective on how durable that growth might be over time.

How are treasury inflows impacted by auction frequency?

Daily treasury inflows can be calculated with a simple equation:

Daily treasury inflows = Average winning bid price * the number of auctions per day

As we adjust auction frequency, we would expect these two factors to move in opposite directions: a lower-frequency mechanism seems likely to result in a higher-average bid price compared to a higher-frequency mechanism. While this may seem obvious, it leaves us with a crucial question: what is the optimal frequency to maximize daily inflows?

The surprising answer: Assuming rational bidders, it is not clear that auction frequency should have any impact on overall treasury growth. Assuming that bidders ascribe the same total value to the project (based on total treasury size, etc.), they should be willing to pay the same daily amount to purchase one day’s worth of votes regardless of auction frequency.

For a concrete example, let’s assume that the current treasury for a Nounish project is 100 ETH after 100 days of auctions. For a project with one auction per day, this implies a ‘book value’ of 1 ETH / vote. We would expect the next auction to be settled with a winning bid of 1 ETH (and therefore a total daily inflow of 1 ETH to the treasury). For a project with ten auctions per day, this implies a ‘book value’ of 0.1 ETH / vote. We would expect the next auction to be settled with a winning bid of 0.1 ETH – since this project has ten auctions per day, this results in a total daily inflow of 1 ETH to the treasury. In both cases, we have a total daily inflow of 1 ETH.

However, there are several reasons why this may not be the case in practice:

  • Rarity: This has been a key driver of NFT interest and value – less frequent auctions may increase perceived rarity (even absent explicit rarity built into the actual tokens) and therefore increase both interest and average bid price relative to higher frequency auctions
  • Forks: Nouns DAO has already been forked several times (e.g., Lil Nouns). There may be some premium ascribed to Nouns over its forks, as each forked community increases the value of Nouns itself (either directly or indirectly), that would result in a higher average bid price and faster treasury growth for the original community. Though not a perfect analogy, we can see that derivative NFT collections generally command lower prices than the original collections (e.g., Mutant Ape Yacht Club has a significantly lower floor price, at ~20 ETH, than Bored Ape Yacht Club at ~100 ETH [2])
  • Auction settlement mechanism: After auctions are completed in Nounish structures, a settlement transaction is required to mint a Noun and begin a new auction. This does not happen automatically – someone must trigger the settlement to launch the next auction. Therefore, auctions do not always start immediately after a prior auction has completed. While this may not have a large impact on lower-frequency auctions, it can dramatically decrease the number of auctions compared to the theoretical daily maximum for higher-frequency auctions. Lil Nouns is a good example: it has averaged ~41 auctions per day over its lifespan, whereas the theoretical maximum would be 96 auctions per day. This may lead to decreased inflows relative to expectations, as bidders cannot predict precise auction frequency with certainty and may therefore deviate from perfectly rational bidding behavior. If Lil Nouns bidders initially bid with the expectation of ~100 auctions per day, this may have established an average winning bid price below where it would have been if they had expected ~50 auctions per day. Price expectations are sticky and difficult to displace, potentially making this self-fulfilling and reducing treasury growth
  • Extreme auction frequencies: In practice, there is likely to be a ‘sweet spot’ of acceptable auction frequencies for this type of model. Projects that choose a frequency that is extremely high (e.g., one auction every fifteen seconds) may find that there are limits to rational price discovery – even if the average bid price should, in theory, find a level where total treasury growth finds equilibrium with lower-frequency models, it may not occur in practice due to the level of consistent attention and participation demanded by such rapid auctions. This could result in more volatile winning bids, and therefore less efficient auctions. Gas fees are also likely to become a more significant factor at higher frequencies (as they make up a higher percentage of total bids and purchases), which may depress demand
  • Post-launch change in auction frequency: While we might expect that the initial choice of auction frequency would not impact treasury growth, the impact of a shift after launch would likely be different. As Kiran suggests here, if an existing DAO like Nouns were to increase its rate of emissions it would likely result in an uptick in treasury growth. This is because the expected value of an incremental holder has already been established (by the lower frequency). To make this clearer - a Nounish project might start at one auction per day and establish an average winning bid of 100 ETH. At this point, each additional Noun auctioned would have an expected value of 100 ETH (based on the number of Nouns and treasury size). If the number of auctions per day increased to four, each incremental Noun would still have an expected value of 100 ETH. As long as there were enough bidders with a reserve price above 100 ETH, we would expect the price to remain steady even with the faster emissions rate. On the other hand, there is nothing to necessarily suggest that starting the project with a faster emissions rate would lead to this outcome. In fact, we might expect the project to establish an average winning bid of 25 ETH, finding equilibrium with the initial 100 ETH / day value

We can look to early data from Nouns and Lil Nouns to begin to hypothesize about the impact of auction frequency on revenue growth. Since inception in August 2021, Nouns DAO has grown its treasury significantly at an average rate of ~100 ETH per day. After 339 auctions (as of 7/17/22), this has resulted in the inflow of 33,917 ETH since inception.

Nouns has averaged ~100 ETH / auction since inception
Nouns has averaged ~100 ETH / auction since inception

The advent of Lil Nouns, in May 2022, provides an opportunity to compare the relative impact of more frequent auctions with the original Nouns DAO mechanism.

While Lil Nouns has conducted 2500+ auctions, it has a more limited number of total days for treasury growth analysis. We have compared its first 68 days in operation to both the first 68 days of Nouns (to hold constant the impact of starting a community) and the same 68 days of Nouns (to hold constant the impact of market conditions).

Comparison of Nouns and Lil Nouns treasury growth over 68 day periods
Comparison of Nouns and Lil Nouns treasury growth over 68 day periods

We can clearly see that Nouns has been able to generate more treasury growth historically (at ~100+ ETH per day compared to ~24 ETH per day for Lil Nouns). This contradicts my hypothesis, but could potentially be explained by a combination of the ‘rarity’ phenomenon and the nature of Lil Nouns as a fork (and explicit ‘lesser’ derivative) of Nouns, which may depress its overall value compared to Nouns. As both projects mature, it will be interesting to see if there is any convergence in total daily revenue.

How durable will treasury growth be over time?

Sustaining treasury growth over time is another key consideration for Nounish DAOs. An analysis of the number of unique bidders and winning bid variability over time can provide insight into the impact auction frequency might have on project sustainability.

While not strictly an input for treasury growth, the number of unique bidders over time on a Nounish project can serve as a proxy for general interest in the project and provide insight into its long-term treasury outlook. We would expect, all else equal, that auctions with a higher number of unique bidders tend to result in higher average winning bids and therefore that projects with a higher number of unique bidders over time will tend to be more successful and last longer.

The topic of unique bidders can be considered over two time frames: long-term (over weeks and months) and short-term (within a single day or week):

  • Over the long term, projects that are able to sustain a high number of unique bidders are more likely to maintain their treasury growth, whereas projects with a dwindling number of unique bidders are likely in trouble
  • In the short term, high volatility in the number of unique bidders may indicate a suboptimal auction frequency. Auctions with relatively fewer unique bidders may be more likely to be resolved with lower winning bids, which reduces the efficiency of the mechanism and slows treasury growth (see previous section for more details on this point)

While the impact of auction frequency on the number of unique bidders (and ultimately project longevity) is not yet entirely clear, I’ve laid out several hypotheses on the potential impact:

  • Short-term: Lower-frequency auctions are likely to have somewhat lower volatility in the number of unique bidders, as participants have more time for price discovery and can monitor the auction relatively infrequently over a longer time period. Higher winning bid prices also make the relative cost of gas for bids cheaper, encouraging more bids in each auction. Conversely, higher-frequency auctions demand a more consistent level of interest and attention throughout a given time period to sustain a consistent number of unique bidders. While there is likely minimal impact within some reasonable band of auction frequencies, extreme frequencies may result in high volatility
  • Long-term: The impact of auction frequency on long-term volatility of unique bidders is not entirely clear. Lower frequency auctions likely limit the universe of potential buyers given the elevated average winning bid, which may result in an earlier tapering of interest. On the other hand, higher frequency auctions require a sustained level of interest over many more auctions

As we compare Nouns and Lil Nouns, we can see that Nouns, with a longer auction duration, attracts more unique bidders per auction than Lil Nouns (~8 for Nouns compared to ~4 for Lil Nouns). It seems that, despite the lower barrier of entry for Lil Nouns given the price point, the longer duration auctions for Nouns is actually a more important determinant for unique bidders (at least so far). Interestingly, Nouns has largely sustained the number of unique bidders over time at ~6-8 per auction, suggesting no significant tapering in interest to date.

Comparison of unique bidders across Nouns and Lil Nouns
Comparison of unique bidders across Nouns and Lil Nouns
Durability of unique bidders over time (Nouns)
Durability of unique bidders over time (Nouns)

Despite the relative stability in the number of unique bidders, there has been relatively high variability in winning auction prices for Nouns and Lil Nouns. This brings up a natural question:

“Why the focus on unique bidders? After all, a steady supply of unique bidders over time did little to insulate Nouns from wild price swings over time (dropping from 150+ ETH in the first several months to as low as ~50 ETH, before converging back to the overall average winning bid price of 100 ETH). There must be more important factors at play, right?”

This is true –  there are undoubtedly other factors influencing the average bid price for Nouns over time. However, the number of unique bidders provides a useful proxy for interest in the project over time, while allowing for the abstraction of confounding variables (e.g., broader market conditions, performance of ETH). It also seems reasonable that Nounish projects may experience more dramatic price swings early in their evolution, as the market attempts to determine a fair price. This trend is clearly demonstrated with Nouns, which has experienced far less variability in recent months as the project has matured.

Comparison of winning bid characteristics across Nouns and Lil Nouns
Comparison of winning bid characteristics across Nouns and Lil Nouns

Community participation

Community participation is another critical consideration for new projects and DAOs. For the purposes of this analysis, we will consider the impact of auction frequency on two measures of community participation: number of proposals and voter participation.

What is the impact of auction frequency on the number and diversity of proposals generated within a community?

A larger community, holding all other factors equal, will tend to generate a greater number of ideas and proposals. On an absolute basis, therefore, a higher auction frequency seems likely to result in a larger number of proposals.

Additionally, this larger community is likely to be more diverse, and therefore generate a more diverse set of proposals, due to the structural elements of a higher frequency auction. A lower bid price, and more opportunities for entry, expands the set of potential community members relative to a lower frequency auction that may select for a more homogeneous population (e.g., significant economic means, deep interest in crypto projects).

We can already start to observe this trend when comparing Nouns and Lil Nouns: Nouns has averaged one proposal every ~3 days since inception (114 proposals over 343 days) whereas Lil Nouns has averaged one proposal every ~2.4 days (28 proposals over 68 days) [3].

The impact on the number of proposals per holder is less clear. It may be that in a lower-frequency community the average individual holder is more engaged due to the higher entry price and greater accountability fostered by the smaller community. On the other hand, lower-frequency communities may suffer from a lack of enthusiasm or decline in interest over time that leads to a decline in proposals per holder. These communities may find themselves with a relatively disinterested legacy holder base that is minimally bolstered by a slow flow of new entrants. Higher-frequency communities, meanwhile, have a more consistent flow of new holders that may serve to inject new ideas into the community and maintain more consistent interest.

So far, Nouns is outpacing Lil Nouns on this measure. Given that there are ~ 1,400 unique holders of Lil Nouns, compared to ~230 for Nouns [4], we can calculate an average number of each in existence over the lifespan of the project (~700 and ~115, respectively [5]). Though Lil Nouns has produced proposals at an overall faster rate (with ~25% more proposals / day than Nouns ), the difference in proposals generated clearly falls short of the difference in unique holders.

What is the impact of auction frequency on voter participation?

Voter participation rate, defined as the percentage of total possible votes that are cast for each governance proposal, can be a useful metric when evaluating the overall level of community engagement and interest in a project.

A lower auction frequency, while restricting the total number of community members and ultimate votes, is likely to result in a community of more engaged voters who tend to vote more often. Since holders in these communities tend to pay more on average for each vote, and each vote has a larger impact on the ultimate result, there is a larger incentive for individual holders to cast votes for each governance proposal.

On the other hand, a higher auction frequency seems likely to result in lower voter participation. As previously noted, we’d expect the average holder in these communities to have less significant economic means. Voting occurs on-chain and requires participants to pay gas – this may make the average holder in a community with more frequent auctions less likely to vote as they face the double whammy of 1) gas fees and 2) a lower relative value of their vote compared to a holder in a lower-frequency community. Ultimately, this might result in a community where only the largest holders regularly participate in governance.

This hypothesis has been supported so far by the average voter participation rate for Nouns and Lil Nouns, with Nouns averaging a ~30% voter participation rate (roughly double the rate for Lil Nouns). These rates have been durable to date, with voter participation rate generally fluctuating by ~10% around these averages and no noticeable drop-off for Nouns since inception.

Voter participation rate compared across Nouns and Lil Nouns
Voter participation rate compared across Nouns and Lil Nouns

*Excludes ‘Canceled’, ‘Active’, ‘Passing’ and ‘'Pending’ proposals

^Voter participation rate defined as number of votes divided by total eligible supply at time of each proposal

Community growth

Defined simply as the total number of holders in the community, community growth is controlled directly by the auction frequency: more frequent auctions will result in a larger, more rapidly growing community.

On an absolute basis, it seems obvious that more frequent auctions will result in both 1) more total votes and 2) more unique holders. Indeed, Nouns has conducted just shy of 350 individual auctions since its formation in August 2021. Lil Nouns, since its formation in May 2022, has already conducted nearly 3000 auctions.

Larger communities, all else equal, will likely make it easier to market and drive new interest to the project – a community of thousands (often) has more reach than a community of hundreds. Additionally, providing more frequent opportunities to join the project may make it easier to sustain and grow interest from the broader public.

However, there are ways that lower-frequency projects can compensate for these potential drawbacks (and that we’ve already seen from Nouns). Fractionalization provides an opportunity for more members to join the community (and at a lower price point). In the future, we may also increasingly see entire DAOs purchase individual NFTs to join lower-frequency communities – this provides a similar benefit to fractionalization in growing the community.

Implications for auction frequency

Based on this analysis, we can begin to lay out a framework for how Nounish DAOs might select an auction frequency:

DAOs favoring a more active community, with a greater number of proposals and diversity of membership, should likely select a more frequent auction structure. These DAOs would prioritize overall community growth, and the network effects associated with a larger holder base, over maximizing community engagement.

On the other hand, DAOs prioritizing a more engaged community may favor a less frequent auction structure. This structure may help to foster a more tightly-knit community, but would likely result in fewer total proposals and (potentially) a less diverse holder base. The cap on absolute community growth should be considered carefully, as the DAO will have fewer holders to propagate its mission and ensure continued interest. At the same time, this slower growth can lead to increased ‘prestige’ from membership and a stronger brand. The Nouns one-auction-per-day mechanism, as an example, has proven to be an exceptionally strong meme that has driven significant interest in the project, helping to propagate the Nouns brand and mission widely.

There is not an obvious right answer today on the question of optimal auction frequency for Nounish DAOs. As more experiments are conducted, and existing projects continue to provide real-world data, we will hopefully get a clearer picture. Nouns, and subsequent similar communities, should encourage these experiments to determine the best way to build the community and grow the treasury.


[1]  Treasury balance as of July 19, 2022 was 25,530 ETH

[2] Mutant Ape Yacht Club is a 20,000 NFT collection, which is double Bored Ape Yacht Club. Nonetheless, the difference in price is ~5x and therefore cannot be purely explained by the increased supply

[3],[4] As of 7/17/2022

[5] Simplified calculation assuming linear growth in number of unique holders over time


Stay Up to Date on Llama 🦙

Follow us on Twitter, check out our website, and subscribe to our Substack.

Artwork credit: 0xEFRA.

Arweave TX
HxQe_lMxdaU0LZVEpd-I6GxJLOCTeeM2uAz-qynlUr0
Ethereum Address
0xeB9f74528aA4F9cA15612c68dC3eE944dF808611
Content Digest
z0b3URy7KEV9ZuKpjNrOmIn8StXzL7Bvj4IC55aKEpI