In June 2017, Ethereum creator Vitalik Buterin published a blog post called Analyzing Token Sale Models. In it, he examined the different fundraising strategies employed by token issuers. This included offerings that were both capped (those which featured "a fixed number of tokens at a fixed price") and uncapped (those with a limitless supply for sale). According to Buterin, each issuance model has its drawbacks.
As a veteran of Ethereum's own uncapped token sale, Buterin's primary concern with such offerings is that "they give participants high uncertainty about the valuation that they are buying at." That's because if the total available supply is unclear, it's hard to figure out how big a piece of the pie an individual token represents.
By comparison, capped token sales (e.g., BAT) have transparent valuations but make it hard for the average buyer to participate. According to Buterin, capped token sales are likely to be "oversubscribed, and so there is a large incentive to getting in first."
To correct the perceived failings of each of the above models, Buterin proposed five properties of a good token sale:
- "Certainty of valuation"
- "Certainty of participation"
- "Capping the amount raised"
- "No central banking"
The problem, however, was that properties 1 and 2 seemed mutually exclusive, as did properties 3, 4, and 5. In other words, you couldn't have it all in one token sale.
Enter the Interactive Coin Offering
Several months after his post, Buterin, along with TrueBit founder Jason Teutsch as well as developer and Modular CEO Christopher Brown, released a paper called Interactive Coin Offerings.
Teutsch et al proposed a workaround to problems plaguing ERC20 token sales that was primarily focused on the first two properties. They aimed to "establish an equilibrium of purchase amounts whose sum is satisfactory to all buyers at some uniform valuation." In other words, buyers could participate if they were comfortable with the price – and everyone would know what all the coins are worth.
To do this, the authors proposed injecting two elements into an uncapped token sale system: voluntary withdrawals and EDCCs (aka smart contracts).
Throughout the beginning stages of the crowdsale, buyers would be able to bid on tokens or withdraw their bids, affecting the valuation. If the valuation became higher than the bidder was comfortable with, an EDCC would automatically cancel the bid and refund the money. Or a bidder could simply change their mind and voluntarily withdraw, incurring a penalty based on how much time had elapsed since the bid. (This mechanism would be used to incentivize against any denial-of-service attack that prevents potential buyers from bidding.)
After a pre-established number of blocks, buyers would no longer be able to voluntarily withdraw their bids, although automatic withdrawals based on the current valuation would still be processed. As the authors reasoned, this would prevent whales from jumping out at the last minute and affecting the valuation.
Finally, to encourage early participation and create a liquid market, the interactive coin offering utilizes an inflation ramp, by which early buyers receive a discounted price.
Along Comes Kleros
Kleros is a "dispute resolution layer" built onto the Ethereum blockchain. Before agreeing to something (e.g., a freelance contract), two parties can use Kleros to create an EDCC that locks up any funds in case of a potential dispute. If a dispute does arise, Kleros calls on "jurors" from the community to arbitrate, whose participation is rewarded with PNK tokens.
Kleros decided to put the interactive coin offering into practice. It labeled its version an interactive initial coin offering, or IICO.
There are a few differences between what is described in the Interactive Coin Offerings white paper and the IICO as employed by Kleros. Teutsch himself detailed three.
For one, Kleros' sale did not include penalties for a voluntary withdrawal early in the sale (perhaps because buyers would already be effectively penalized in gas costs).
Second, according to Teutsch, "buyers in the Kleros sale were required to include extra data describing where to insert their bids into the on-chain linked list data structure, whereas [in] the original whitepaper incentivized external Observers supplied this data 'automatically' for buyers." Teutsch believes that "a real-time graphical user interface illustrating active bids and their personal caps would help buyers process available information."
Kleros co-founder and chief technology officer Clément Lesaege, who reviewed the results with Teutsch, clarified that it's not actually the buyers including extra data, but their interface. "So, from a buyer perspective, it's completely transparent and almost real time. Contrary to the paper version where they have to wait to have their bids taken into account and need to pay extra to incentivize someone to include their bids."
Yet he concurred that the interface was a challenge. "Most users did not know how to use web3, so we also had, in addition to the IICO interface, a simplified 'direct send' allowing [them] to make uncapped bids."
Last, Teutsch told ETHNews, "The original white paper protocol includes a mechanism for specifying personal minimums in addition to personal caps. The Kleros sale supported only personal caps."
When asked why Kleros didn't implement minimum valuations to begin with, Lesaege said, "Kleros took the decision of using the interactive coin offering after reading the original paper." (ETHNews reported on the interactive white paper's release in September 2017; it was later updated and republished on December 11, 2017, with Brown listed as a coauthor.)
Regardless, Lesaege and Kleros apparently did consider incorporating minimum valuations but opted against it. "Personal minimums would make [things] more complex. Buyers [had] never [set] parameters when buying in a token sale before and [leaving this out] seemed to be the reasonable way not to frighten buyers with excessive complexity." Indeed, Kleros appears to have kept an eye on a key property of buyers mentioned in the white paper: "preference for simplicity."
But it may have equally been for reasons of security, which, Lesaege told ETHNews, was a challenge:
"In my opinion, the IICO is the most complex token sale method which has ever been implemented and code complexity leads to risks of bugs and vulnerabilities. We made some modifications to simplify the code and ended up with only 349 lines of code for the IICO contract."
Did Kleros' Model Live Up to Expectations?
Kleros held its token sale from May 15 to July 15, with 160 million tokens (16 percent of the total) sold. Users contributed 5,794 Ether to the sale and some took advantage of the voluntary withdrawal mechanism; 267 Ether was pulled back. EDCCs also kicked in to refund 11 Ether after the valuation rose above individual contributors' limits.
While the model mostly worked as intended, there was also a bit of weirdness.
Although the white paper envisioned heavy involvement at the beginning of the sale due to the inflation ramp, that wasn't the case for Kleros. Instead, said Teutsch, "few buyers participated during the 'Free Withdrawal' period. However, there was a surge in participation at the end of this period as well as the very end of the sale."
Teutsch speculated that "buyers wanted to wait to see what other buyers did before participating." That seems normal, especially when gas costs are taken into account. As Teutsch put it: "The gas cost of submitting a bid and then withdrawing it again later exceeds the cost of simply deciding not to participate. Thus, rational buyers were incentivized to wait until the last minute to make their decisions."
Lesaege, however, said it wasn't odd at all, given Kleros' token sale design: "This was quite expected, the first month just acted as a 'pre-bid' period which is equivalent to someone setting up a pre-bid smart contract allowing bids to be including in the first block of the sale (as we've already seen for other ICOs). So, the real IICO started at the beginning of the second month and that's where we saw the highest amount of bids as expected."
The surge at the end of the overall sale, on the other hand, could be tied to last-minute FOMO, says Teutsch. Equally likely, perhaps, is that "the requirement to use a Web3 browser may have added friction."
A preliminary analysis conducted by Kleros and shared with ETHNews goes into greater depth. It notes that, after excluding manually withdrawn bids, 35 bids stipulated a personal cap. Five of these were automatically withdrawn when the valuation went over their personal caps.
While the team notes that most of the Ether was contributed through capped bids, overall, just 16 percent of all 307 bids used a personal cap. "Small bidders," the report says, "seem particularly willing to accept dilution," suggesting that most buyers weren't as concerned with valuation as Buterin, Teutsch, and Brown are.
Although the Kleros team is still analyzing the results, the report concludes, "Estimating the amount of arbitration cases available in advance and how much PNK will be required to be drawn as a juror with a sufficient frequency to be worth a given contribution is difficult, and it may not have been worth it to small contributors to expend the effort to make this estimate."
Those results shouldn't call into question the fundamental assumptions of the white paper and the Kleros IICO. Users who wanted to participate and didn't wait until the final day were able to participate on their terms. Whether or not they cared about the valuation was up to them. The point was that they had the information.
The Future of Interactive Coin Offerings
Kleros' token sale won't be the last interactive coin offering. It brought in $2.5 million and, for the most part, met its goals. Both Lesaege and Teutsch stated that they expected other projects to toy with the concept. Additionally, key challenges that came to light during this token sale are already being addressed, specifically the user interface problem. According to Teutsch, Christopher Brown's team at Modular, which worked on interactive coin offering EDCCs, "is in the process of developing another user interface."
Lesaege expressed gratitude to Teutsch, Buterin, and Brown for putting the concept out there, and obviously finds promise in it. But he suggests we need more use cases of an IICO to work out the kinks because Kleros' sale was "not overhyped, nor underhyped" but in the Goldilocks zone. "To really see IICO in action," he said, "we may need to see [it] used by an overhyped sale."