The Metamorphoses of Token Distributions Since 2009: Lessons for an Optimal Mechanism
In this post, we examine various token distribution models, roughly in chronological order. We then try to abstract the properties of an optimal token sale mechanism.
Part I. A Brief History of the Future
In what follows, we give a brief history of token distribution models, largely intended for non-crypto native readers.
In Part II, we ask ourselves the question what properties a token sale mechanism would have if designed by God.
The original: Bitcoin
Token distribution models have gone through a number of metamorphoses since the “fair” Bitcoin mining in 2009.
In 2009, Bitcoin’s inventor(s) engineered a mining system that distributes new token supply in the form of bitcoins to those who participated in supporting its network’s consensus.
Although its Proof of Work resulted in a hardware arms race, the goal was to bring as many people into the network as possible to ensure a sufficiently large distribution of stakeholders.
From roughly 2013 on, Initial Coin Offerings (ICOs) became en vogue as a novel way of launching a new blockchain project.
To support the launch of a new protocol, participants contributed bitcoins in exchange for tokens at network launch.
In 2014, Ethereum itself raised US$ 16 million in BTC in this way.
The launch of the Ethereum mainnet and the promise of computational power on blockchain, together with the introduction of the ERC-20 token standard, opened the floodgates for a wave of ICOs in 2017 and early 2018.
In the midst of this craze, Vitalik published a blog analyzing token sale models, expanding on (then) new token sale models which evolved from the earlier, cruder capped or uncapped sales.
In his post, he also addressed one set of criticism of the Ethereum sale and other uncapped sales: that they give participants high uncertainty about the valuation they are buying at if they do not know the total supply of tokens on offer.
He notes that
[…] the model certainly has some rather weird economic properties that we would really like to avoid if there is a convenient way to do so.
This contrasts with capped sales, most popular throughout 2016 and early 2017, which give buyers a large incentive to get in first and were typically oversubscribed.
The Netscape moment may have been Brave’s mid-2017 offering of the BAT token which raised the equivalent of US$35 million in… 30 seconds!
The Gnosis metamorphosis
The BAT sale lead to people trying to outbid each other’s transaction fees and clogged the Ethereum network for 3 hours after its sale started.
Gnosis tried to address this by introducing a reverse action mechanism.
Without going into detail here about the economics of the reverse Dutch auction, what we can learn from the Gnosis sale is that it combined a capped sale with a variable number of tokens distributed, depending on how long it took for the auction to finish: the longer it took, the more tokens as a % of the total became available to its token buyers.
This auction design was intended to allow token buyers, if they cared about valuation, to wait until the valuation of the Gnosis project fell to a level they were comfortable with before buying.
However, what happened instead is that FOMO kicked in and everybody bought in on the first day. The cap of US$ 12.5 million was reached with only 5% of the tokens sold, giving Gnosis a valuation of US$ 300 million. (Note: at the time of writing, messari.io calculates Gnosis’ “Liquid market cap” at US$ 1.841 billion).
Valuation heuristic vs. “free float”
The Gnosis sale indicated that the crypto community does not anchor its decision to purchase a token (solely) in valuation.
What they seem to care about more is how long tokens may be locked up and subject to vesting, or to use a Wall Street term: how much of token supply has “free float”.
Anybody who trades TradFin stocks will have their eye on free float (% of overall stock listed) when putting on a trade. It is this free float that affects the market because only that portion of the overall tokens is freely tradable.
In the cases of Gnosis, shortly before its sale started, the team responded to criticisms that they’d risk ending up as the “central bank” of their own GNO token, holding the majority of the total issued (which is what happened).
In response, they offered to not sell 90% of the GNO held in treasury. Effectively, 85.5% (90% of 95%) of all GNO tokens were taken out of the market, leaving a free float of only 14.5% of all tokens issued.
On this basis, the valuation of the Gnosis project, calculated as the amount of tokens in free float multiplied by the then GNO token price, was much lower. It would foreshadow later debates about token restrictions and vesting.
IEO’s 15 minutes of fame and the rise of the machines
In 2019, projects increasingly sought to list their free float on centralized exchanges. resulting in 15 minutes of fame for Initial Exchange Offerings (IEOs).
This lasted until founders realized IEOs risked putting their project at the mercy of centralized exchanges, in addition to the often outrageous listing fees and commissions.
Binance for one used its IEO listings as a way to bribe its customers with pre-launch tokens.
IEOs got gradually replaced when founders started to just throw their token on decentralized exchanges, where they could gained almost instant tradability via Automated Market Making (AMM).
2020 and continuing into 2021 saw further metamorphoses in token distributions.
DeFi projects now typically distribute (a portion of) their tokens directly into the wallets of their users to incentivize usage of their protocol and to give token holders a say in their governance.
Some DeFi token distribution models may still have pre-mines and private pre-sales, others may no longer include a fundraising component at all and limit distribution to users that are providing liquidity to the network.
In the latter case, we may have gone full circle and are back at Satoshi’s “immaculate conception” distribution of BTC to its users.
Be rich or be right?
In all this, there’s a fundamental schizophrenia between token distributions that do not have pre-sales and those that do.
Such pre-sales are typically secured via a Simple Agreement for Future Tokens (SAFTs) even before the token itself is issued (hence the SAFT’s name), or under bilateral Token Purchase Agreements or variants such as call options at a low exercise price.
Irrespective of the way the pre-sale happens, we have seen SAFT discounts or call option strike prices vary wildly, even if purchasers contribute at or around the same time.
By their nature, SAFTs and call options are private agreements so there is no mechanism, beyond the integrity of the project leads, to maintain price consistency across contemporaneous investors.
As a result, pre-sales lack transparency. Worse, they risk puffing up projects like a soufflé, filled with hot air from moon memes on Telegram and the marketing hype resulting from the public announcement of a VC participating in a project’s pre-sale.
Many of these crypto soufflés collapse when tokens become tradable and early backers lock in gains from tokens they purchased at steep discounts.
This behavior is in itself not irrational, but it has lead to a counter-movement against any form of pre-sale, especially if such private sale is dominated by a VC (see our previous post).
At the extreme, some projects leads declared they’d rather be right than rich and distributed their entire token reserve to the community.
Some came to regret it.
Most famously, the prolific Andre Cronje, who developed much-used and widely-admired yearn.finance basically as a hobby, in his viral second rant called himself an idiot for giving away his $YFI tokens:
When I decided to distribute YFI 100% it was because I believed it would allow me to exit to the community. However, I am still blamed if the price goes down, I am still constantly plagued by “when next release”, “when update”, etc messages. I still have all the responsibility and expectation, except I have 0 of the reward or upside. Don’t do this, I was an idiot.
Luckily, Andre didn’t just sulk and worked on a solution in the form of an unsecured funding DAO on yearn.finance.
This Delegated Funding DAO Vault lets founders borrow in an unsecured way to cover the costs of launching their project, more specifically the typically high cost of an audit of their smart contract code.
Righteous or naïve, the development towards “fair launches” is proof of ongoing experimentation with new token distribution models and alternative capital formation mechanisms more broadly.
2. God’s Token Distribution Design
What can be abstracted from the above overview? Can we extrapolate a token distribution model for the future?
If God were to design a token distribution model, how would it look like?
In what follows, we try to list the properties of an ideal token sale model i.e. a distribution mechanism that still has an element of capital raise to it.
The Urtext
Vitalik’s same blog of June 2017 serves as a good starting point. In it, he lists the following desired properties for a good token sale mechanism:
Certainty of valuation
This is defined as a floor on the percentage of all tokens a token buyer is getting, essentially framing this as an anti-dilution issue.
In a more traditional fundraising context, such anti-dilution comes from there being a ceiling on the valuation at the time of the fundraise. i.e. at the moment you invest.
Certainty of participation
This second requirement speaks for itself: Participants who try to buy your token should have a reasonable chance of succeeding.
However, what happens if some people want to value your project higher than your cap? Practically, this means they’re trying to buy more tokens than are available in a capped sale, so they’ll fall out of the boat.
For purposes of our optimal token distribution design, this means there will be a trade-off between valuation and participation, known as the “first token dilemma”.
There are further trade-offs at play:
An optimal token sale model probably caps the amount raised.
Also nobody, in particular the token issuer, should end up with a controlling % of tokens post-sale, i.e. become a “central bank”.
Finally, the sale should be efficient so no money is left on the table.
The above trade-offs have become known as the “second token trilemma” since, again, all three cannot be satisfied at the same time.
Subsequent models have essentially tried to somehow reconcile the above trade-offs, and we list them below to the extent that we see some of these solutions as integral to our optional token sale design:
Smart-contractified reserve distribution
The idea here is to address the central banking problem by putting conditions on how a project’s token is distributed by way of smart contract.
As mention above, this could be achieved by enforcing vesting through smart contract code, automating token rewards by linking them to how long token holders have held on to their initial allocation, pledging the token reserve to an automated market maker tasked with maintaining price stability, linking token awards to the actual spend on a project’s product if the use case lends itself to this, and further programmatic rules that seek to remove arbitrariness in how a token reserve is managed.
All of the above goes beyond onchain voting which many projects use to boost their decentralized creds but which comes with its own set of inequities, since in its crudest implementation (“one token, one vote”), such mechanism can lead to a very large concentration of power in the hands of the largest (read: richest) token holders.
Buyer limits
In a further effort to distribute token ownership more widely, projects could cap how much each token buyer can purchase in a pre-sale or main sale.
However this is easy to circumvent by using several wallets, unless participants are all asked to whitelist, which in turn would limit who can participate: under current securities laws, a whitelisted sale would most likely be restricted to accredited investors, which again favors the haves over the have-less.
So what is worse: a sale in which anybody can contribute as much as they want or a sale in which only a few can contribute up to an individual cap, knowing that this cap can easily be gamed?
The evidence is mixed, with some projects that have capped per-wallet purchases showing equally high levels of ownership concentrations as projects without individual buyer limits:
If the objective is to achieve a more equitable token distribution, perhaps a crude individual buyer limit is the wrong tool and there is a subset of mechanisms that can be used post-sale, for instance:
Token rewards for longer-term holders: As described above, the idea is to rewards longer-term participants and their participation in proposals, voting, or other forms of work that help grow and sustain a project.
Non-linear rewards: This can be achieved by way of weightings, tiers, and/or logarithmic curves, so that rewards do not scale linearly with the amount of tokens a user has and thus disproportionately benefit whales.
Phased rewards: The idea is to allocate rewards to different token holders in function of the stage of the project’s development.
Rewards based on user archetype: Reward could be based on the users’ historical onchain activity to attract the archetype the project wants for their network, e.g. a token holder who typically invests, stays with, and votes on a network over longer periods of time.
Continuous offering
From the above, it is clear that God will have to make some hard decisions when designing His ideal token sale model.
One property however we believe would mitigate if not address a lot of the shortcomings of current mechanisms is the idea of a continuous sale.
We blogged about this earlier in the context of how tokens linked to a project’s revenue could continually be offered under a bonding curve.
Some interesting thinking came out of a project called Fairmint who are running a continuous offering of their own (security) token.
More generally, a bonding curve would allow for the continuous pricing of a token in function of user demand.
Such time dimension has the potential to at the very least narrow the mismatch between the rewards of the project team and the interest of the token buyers.
Essentially, it rewards patience and the Universe is patient: only when the project grows and its users increase is this reflected in the price of new tokens issued.
Issuance under a bonding curve hence puts the cookie jar out of reach of the founder team, who can no longer help themselves to big portions of tokens out of the reserve: all issuance is programmatic.
Conclusion: Out of Chaos, Order
An overview of the various ways in which blockchain projects have raised funding since Bitcoin’s “immaculate conception” token distribution leaves even crypto-natives bewildered and guessing what’s next.
The only right answer is that it remains to be seen: We will have to wait until the laws of self-organization have worked their magic and an orderly design emerges.
The directional contours of such design are already visible and the direction is clearly towards decentralization, openness, and spontaneity.
Founders who court concentrations of centralized money are working against this arrow of time.
Instead, they should have faith that if they continue to experiment with token models that reward generosity, contribution, networks, capability, community, values, intellectual fierceness and integrity, funding will follow.
Join our official OtoCo Telegram group to hear about the design of our own token distribution model.