Blockchain Blockchain

DigixDAO: Divorce Story

DigixDAO: Divorce Story Blockchain, Blockchain PlatoAiStream PlatoAiStream. Data Intelligence. Vertical Search. Ai.

In the following article, “Digix” refers to the project in general, “Digix Global” refers to the team behind Digix, and “DigixDAO” refers to the DAO structure governed by DGD token holders. “Treasury” refers to the DAO’s spendable Ether that was raised in the ICO.

  • In January, the Digix community (DGD token holders) were given a chance to vote for the project’s dissolution and will walk away with the project’s Treasury proceeds.
  • Parties who took risk on DGD and closely followed the vote technicals were able to attain as much as a ~48% spread between DGD and the underlying ETH of the DAO.
  • The DigixDAO dissolution event is an important milestone that shows the governance space is maturing and stakes are getting higher.

Note: For convenience, the background of the DigixDAO dissolution vote has been moved to the end of the article. For full context, or if you are not familiar with DigixDAO, please start with the “Background” section below.

As a quick refresher, a team-sponsored proposal to return the DigixDAO Treasury —around 386,000 Ether raised in the ICO — to DGD token holders was put forward for a vote in January. DGD token holders could stake their tokens (signal to vote) and vote yes, no, or abstain. The two requirements for the dissolution vote to pass were reaching quorum (total votes denominated in DGD stake must be greater than 40% of total DGD staked) and reaching quota(simple majority greater than 50% must vote yes). Please see the “Background” section at the end of the article for more details.

As a live vote that is registered on the Ethereum blockchain, we were able to parse the DigixDAO contract data to measure voter behavior. Our analysis called data from a private subgraph GraphQL implementation and the DigixAPI. [1] [2]

Because the DigixDAO voting system is 1T1V (one-token-one-vote), we looked at the voters that were committed to vote on the dissolution proposal and their respective stakes. For a more in-depth dive in applying voting theory to on-chain voting systems, see Jake Brukhman’s piece on the Banzhaf voting index below.

Are blockchain voters ‘dummies’?

We can learn a lot about voting systems from the existing literature.

After filtering by the top eligible voters by stake, the voting power concentration became apparent.

Top 25 Voters by Stake (full list here). Orange highlights top four. `

  • cumulative_stake: cumulative sum of stake starting with highest
  • cumulative stake % of quorum: cumulative_stake / quorum (408K)
  • cumulative stake % of quota: stake / cumulative_stake
  • cumulative stake % of quorum (bottom_up): cumulative sum of stake starting with lowest / quorum (408K)

For the quorum to be reached (determined to be 408K DGD based on stake), it would only require the top four voters to participate in the reveal phase. Even more glaringly, the cumulative stake % of quorum (bottom_up) shows that the quorum would not be reached even if every other voter outside the top four had voted while the top four abstained.

For the quota, if we assumed that the top four voters all voted the same way, they would command the majority voting power (greater of 53% of the vote.)

Thus, the top four voters, if treated as a coalition of 440K stake, completely controlled the vote by both (a) establishing or preventing quorum by voting or abstaining, respectively, and (b) controlling the quota and thus single-handedly being able to determine the proposal outcome. (In voting theory, such a critical voter is known as a dictator.)

y-axis: cumulative stake % of quorum

Note that our core assumption was that the set of highlighted addresses were part of the same coalition, which was not a strong assumption at the time. There was no way to determine who the top addresses belonged to specifically or whether the non-Digix Global team were in control of these addresses or not.

Interestingly, a deeper look into the blockchain led to a discovery of a potential coalition, which was comprised of the following addresses.

> 0x2479f13aa8cb50ce134fee0f32e837493a115699
> 0xdb569e2a5af89d492bda0d2bd567c5e5c525c94e
> 0x7006e83f868f1ff893053c100a5cfb39ef586002
> 0x741eb925138c2906ca870fd1100ccf4e8e7893e9
> 0xc203f2019054fce9635b4581c00fef50b91c9851
> 0xd5b4f094d0414bed35cb63820895085db8e6687c

We noted that a set of participating addresses had locked their tokens and staked into the DAO contract in consecutive transactions, each minutes apart. [3] This set of addresses had also sent large batches of DGD tokens between each other since the ICO. Some addresses in the set had also shared the same exchange deposit/withdrawal address. [4][5][6]

Top 25 Voters by Stake (full list here here). Green highlights the addresses in question.

Based on Etherscan alone, we do not know the specific identities behind these addresses. However, from our model above, we know that a coalition of 440K stake is sufficient to control the vote. The set of addresses in question, if treated as one coalition, commands a total stake of ~468K, which is above the 440K threshold. Treating this set as one voting coalition is applicable to the above model because of material impact from a coordinated set of behaviors.

On January 18, 2020 @ 3AM UTC, the commit period ended — thereby locking in the set of voters in the list above. If the set of addresses in question were in fact a coalition, they would understand that the dissolution would be guaranteed since they were certain of their grip on the vote’s outcome. Any price below the value of DGD based on the underlying ETH of the DAO (0.193 DGD/ETH) would be money left on the table. Around the same time, the general DGD/ETH market spiked from 0.14 to 0.17 (a 21.4% return). Whether or not these buyers were the coalition members themselves or other interested third-party participants is unknown.

The existence of the 468K Coalition was revealed on January 20th 2020, as evidenced by the revealed spike in yes votes. The set of addresses in question — which was initially flagged for committing to vote in the same time window — are shown to have also coordinated their invoking the reveal_vote function consecutively in the DigixDAO contract within a 20 minute window. [7]

Timeseries of total `reveal` votes in favor of `yes` “Reveal, vote yes” function of the 468K Coalition

Given the voting weight analysis and the voting results, it’s clear that the before the vote’s end that the dissolution would pass, as there would not be enough stake to overturn the simple majority. From here, we expect that the DGD/ETH smart contract will be published at the end of the quarter, at which point DGD holders can exchange their tokens for the underlying Ether.

Governance of the Long-Term View. The DigixDAO’s successful dissolution reveals that this form of decentralized governance worked as designed but not without quirks. On-chain data will continue to play a major role in understanding the state of these networks — and in some cases may reveal potential activism from arbitrage-focused funds. Unfortunately, in this case the Digix community opted to dissolve the project due to lack of progress. But in the future, projects and stakeholders with longer-term horizons will have to appeal to the long-tail of stakeholders and make cases against such dissolutions.

Investor Spectrums. While Digix may have been saddled by the old ICO era’s baggage, teams today have taken a step back from fully decentralized funding models. Instead, teams are selling equity structures in early stages of projects. By doing so, today’s teams have to position the role of the founding company within a network, and how those value flows play seem to be initially determined by the equity investors and then eventually the token investors of a network. As these networks gradually broaden and mature their set of holders, alignment of incentives and distribution will require the appropriate actors to help bootstrap these networks and companies in the early stages. The correct value exposure along this front may also eventually fall across various risk spectrums and its appropriate investor classes.

Asset Market Infrastructure. Because of the ICO structure and the DGD token distribution, the liquidity on the books was relatively sparse. A healthier order book, along with lending and borrowing markets, may create more robust arbitrage opportunities in the future for such governance systems. In this case, because the arbitrage was in ETH terms, a fund could frame the opportunity as doubling down on an ETH-long position — or, a savvy fund could also hedge against ETH market exposure during the duration of the trade to earn the pure upside in dollar terms. In the future, exchange support for token voting will inform stakeholder awareness and contribute to voter participation and turnout.

In retrospect, the Digix project was one of the first to define the asset-backed token model. Today, we can view it a predecessor to asset-backed tokens we see proliferate in the form of stablecoins. DigixDAO was also the second major DAO project in the world. This case study tells a story of unique cryptonative arbitrage — where cryptoeconomics of the DAO and the speculative markets have worked against each other to cause dissolution. [8] The Digix case-study is one to look back on as we begin to see the proliferation of networks and their asset dimensions.