On April 27, 2019 Aragon Network Vote #2 (ANV-2) concluded. Within hours of the polls closing, commentary and feedback started pouring in. This post is going to review what different commentators and stakeholders have said about the vote, and also share some of my own thoughts about what we can learn from the process, the results, and how to continue improving Aragon’s governance processes going forward.

If you are unfamiliar with how Aragon governance works, we have a page on the main Aragon project website that can help get you up to speed.

Before continuing, on behalf of the whole community, I want to thank everyone who participated in ANV-2 and shared their feedback via the many different channels through which we received it: the Aragon Forum, Aragon Chat, Twitter, Reddit, personal blogs, and others, including the “official” survey we shared for input with the final results of the vote. We have published the anonymized responses to that survey publicly here if you are interested in digging into the responses yourself.

I think the format of the Aragon Network Vote #1 review worked well so I’ll use it again for this and future reviews. First up, what I think went well with the vote.

What went well

We received a lot of feedback about people’s experiences participating (or not participating) in ANV-2. This is already an improvement over the last vote. For example, there were more than twice as many responses to the survey this time around. We also received plenty of feedback through other aforementioned channels that I’ve spent time reading and thinking through. This is great news because it shows people care. So thanks again to everyone who shared their thoughts.

Digging into the feedback, there was plenty about what could be improved, which I will get to later in this post. Regarding what people liked or thought went well, two categories of comments stood out:

  • User experience. People like the design of the voting app and think it is intuitive and easy to use. This is a win for the product team for sure.
  • Community. The fact that ANT holders can participate in the governance of the project using a clearly defined process is exciting to many and is seen as a marked improvement over the opaque, non-participatory governance models of other projects and institutions. Our process can be improved, which we’ll get to, but it’s a good start.

From my own perspective, there are a few additional aspects of this vote that I think went well.


There were a few awkward mistakes that happened during various stages of ANV-1 that didn’t happen this time. In ANV-2, the process for submitting proposals was properly followed by all of the authors, there were no proposals submitted late accidentally, and the Aragon Association finished their review well ahead of time.

While I think this is partly owed to the improved communication around this vote that made the whole process clearer and more understandable for all participants (which itself was assisted by feedback from ANV-1) this was really a team effort on the part of the whole community to raise awareness about how Aragon governance works and the importance of professionalism in doing it right. Go team!


The communications around this vote were much improved, again thanks to specific feedback we received after ANV-1. After ANV-1 we created a mailing list specifically for people who want to sign up for Aragon Network Vote alerts straight to their inbox, which can help cut through the noise of all the other activity we have going on in the community. Even with these improvements, we will continue to explore ways that we can reach ANT holders and get them the information they need to effectively participate in Aragon governance.


Once again, we had a ballot packed with high quality proposals. Nearly all of them were submitted by new authors who did not participate in ANV-1, another sign of increased interest in participating in Aragon governance. Two proposal authors also now bear the distinction of being the first to not only have proposals included on two ballots but also to have those proposals approved by ANT holders: congratulations to @anteater0x, author of the approved AGP-9 and AGP-10 proposals, and @ganeJackS, author of the approved AGP-18 and AGP-43 proposals.

I’m personally really happy with these results and am excited to continue working to improve the AGP process so we can do even better next time! Which brings me to…

What could be improved

Being a young project working with new technology that unlocks new possibilities for governance, Aragon is on the bleeding edge of the bleeding edge. It’s expected that there will be much to improve as we quickly outgrow processes that worked well enough early on.

By putting theory into practice, we are learning where these weak points are and bringing them to light so that we can more closely examine them and come up with solutions that could work to make our product and our governance system better.

Based on the feedback we received, a few areas of potential improvement stand out.

Proposal reviews

As is often the case when a deadline is set, several of the proposals submitted for consideration in ANV-2 were submitted relatively last minute, with only an hour or two to spare before they had to be finalized for the Aragon Association review. Community members (rightly) complained that they would have preferred more time to give the proposal authors feedback in hopes of influencing the wording of the proposals to make them more favorable to ANT holders.

There have been some ideas floated about how to fix this and I myself have an AGP in the works that I plan to present to the community in the next couple of weeks to address this issue.

Voting app loading

As one of the oldest Aragon organizations on the Ethereum mainnet, the governance.aragonproject.eth organization takes quite a bit of time to load its history when a user first visits the organization. This is something the product team is actively working on improving.

The good news is that, as of Aragon 0.7.2, which was released a few weeks after ANV-2, the Aragon client now has improved caching so once a user finishes the initial load the organization will load almost instantly each time the user returns to the organization. In addition, that initial load should now only take a few minutes at most instead of up to half an hour as it did in the past. We will of course continue to push the boundaries of what is possible for a fully decentralized, blockchain-based application user experience.

Voting dynamics

After poring over the data from the vote, Evan Van Ness pointed out that one voter (who had 792,191 ANT at the time they voted, or roughly 2% of the supply) was able to swing several AGPs on the ballot in their favored direction last minute.

This sparked discussion in the Aragon community and the wider Ethereum community about the positives and negatives of token-weighted voting. While removing ANT holders as the “final approver” of AGPs is presently blacklisted from the AGP process (and we haven’t seen much demand for that anyways) there are some ideas that have been presented to mitigate this kind of last-minute influence over the results by larger ANT holders.

Techniques are being discussed such as lock-weighted voting to discourage moneyed interests from buying or borrowing ANT to maliciously influence vote results, and commit-reveal voting to prevent voters from using vote progress to decide when or whether to cast their votes, among other options. There is not yet consensus on a preferred method at the time of this writing, but more likely than not we will see an AGP in a future vote cycle to try and address this issue.

Vote analysis

ANV-2 finished with an average of 45 addresses and 1,512,026 tokens (or 3.8% of the supply) voting for each proposal. AGP-42 author Ameen Soleimani may be pleased to know that even though his proposal was ultimately rejected, it had the most participation of any proposal, with 69 unique addresses and 1,959,858 tokens participating.*

Here is the complete breakdown:

AGP Addresses Tokens
AGP-9 48 1,743,746
AGP-28 38 1,311,927
AGP-34 46 1,590,694
AGP-35 46 1,572,931
AGP-37 32 1,280,019
AGP-40 44 1,444,196
AGP-41 62 1,665,175
AGP-42 69 1,959,858
AGP-43 23 1,039,690

You can find the data used to pull these figures in Aragon One team member Brett Sun’s Tableau profile here, and of course if you want to examine and analyze the raw data yourself it will be stored forever on the blockchain.

* Note that the number of unique addresses can be easily inflated by splitting one’s tokens across multiple addresses, and so this metric should not be taken as a reliable indicator of unique engagement. Tokens participating, on the other hand, has a cost associated that scales linearly with the number of tokens and so is the most accurate and reliable metric we have for voter support. This is one reason why votes are weighted by tokens and not addresses in the first place.

Next steps

After going through the experience of another Aragon Network Vote and reviewing the feedback other folks shared about the process, what they thought went well and what could be improved, I am working on several AGP drafts that aim to continue improving the AGP process. I also invite anyone else who has an interest in Aragon governance, and has ideas about how the process could evolve to improve decision-making by ANT holders, to get in touch by creating an issue in the AGPs repo, starting a thread in the Aragon Forum, or dropping by the #governance channel in Aragon Chat to discuss.

This post was a collaboration between

John Light, Aragon One

  • John Light
  • Aragon One