Summary of Experiment:
In October 2022 I started a thread in the Decentraland forum titled VP for Social Capital! This post proposed that we reward social activity in Decentraland and the DAO with delegated VP. In theory this would incentivise social investment in the platform, help distribute VP to a larger number of users, and give users an avenue for gaining a voice in DCL’s governance without spending money.
To test this idea, and identify flaws, over the past three months I’ve been tracking DAO voting data and parcel explorer data and delegating VP to the top wallet in each category. The results for October, November (which I skipped), and December 2022 are listed below.
.
.
.
October 2022
Top Voters:
- mrfg#1ba1, 0xfa93c9a95727268b88ff77b5396ed3b8b59a1ba1, Delegated 13K VP
- Mattimus, 0x65813baedfd68029af5382672b0e4dd2e6f27952, Delegated 13K VP
Top Explorer:
- Pibcw, 0xf8ef4c0decdf993aec0f0045b7fca55685dbaab6, Delegated 26K VP
November 2022
Top Voters
- N/A, 0x022111936500251ea358a98cbB158d77580EFb42, Delegated 0 VP
- LordLike, 0x247e0896706BB09245549E476257A0A1129db418, Delegated 0 VP
Top Explorer:
- Pibcw, 0xf8ef4c0decdf993aec0f0045b7fca55685dbaab6, Delegated 0 VP
December 2022
Top Voter:
- Guest#60fd, 0x09aD31043b0B939576d5206b064Ca1807C4160Fd, Delegated 26K VP
Top Explorer:
- Guest User, 0x2509dea66db55699381bcd84d6e4158f747e3281, Delegated 26K VP
.
.
.
Summary of Findings:
The two main critiques of the original post (Bot Farming and Data Interpretation) remain major blockers.
Many of the users voting in the Decentraland DAO appear to be bots or best case scenario, they are users who vote blindly without reading or engaging with proposals. Roughly 40% of the top ten voters each month match this profile and these types of accounts usually take the top voter slot and gain the monthly delegated VP reward. I have not dug too deeply into the parcel explorer data, but I imagine bot-like avatars in-world are a problem as well. I want to continue this experiment but I do not want to validate or encourage auto-voting accounts by delegating them VP.
In an attempt to filter out suspicious auto-voting accounts, only wallets that hold Decentraland Names will be eligible to participate and earn VP rewards as I continue this experiment over the next three months. I see this as a short term solution and think we should find a better way to deal with bots. Auto-voting accounts could eventually acquire Names in the future and I’m not a fan of users being able to buy verification. I encourage Decentraland and the DAO to attach some sort of human verification system to DCL passports, perhaps using Decentralized Identifiers, Soulbound Tokens, in world captcha quests, or some sort of friend list vouching system. As Decentraland grows, so will its bot population, we should begin to tackle this problem now.
The other take away from this experiment is that data can be filtered and presented in multiple ways with differing results. The prime example of this being the trouble Decentraland had giving definitive DAU numbers after the infamous 39 user coindesk article.
As the person running this experiment I must rely on and trust that the data provided by DCL Metrics and the DAO Transparency Document is correct and filtered non maliciously. As participants in this experiment, you must trust me (who is not in any way a data scientist) to interpret that data professionally and relay it accurately. I am not calling out DCL Metrics or the DAO, both have been very open with info about their data, but interpretation of this data it seems is somewhat subjective and there is room for unintended bias on their part and human error on my part.
I don’t know what the solution is to this problem, and at the scale that I’m operating I’m not sure it matters all that much. However, if this VP for Social Capital experiment is ever adopted officially by the DAO and scaled up, I’d recommend having multiple independent teams gather and compare data before delegating out any VP.
.
.
.
Next Steps:
As mentioned above, I plan on extending this pilot project for the next three months, this time only allowing holders of DCL Names to participate. I will continue to delegate 26,000 VP to the top Name holding voter and explorer each month however, once this proposal: Should the DAO implement a mechanism to delegate VP to multiple wallets? is enacted, I plan on spreading out the VP rewards. Once able, I will delegate 5,000 VP to the top five monthly voters and top five parcel explorers holding DCL Names.
As with the first proposal, I welcome any criticisms or suggestions.
Eventually I’d like to see this experiment evolve into a larger VP rewards system using the DAO treasury or something similar as a VP pool, but for now it’s just one small step at a time.
.
.
.
More on Bots and Auto-voting Accounts
After staring at DAO voter data for the past few months it’s become obvious that a large number of participating wallets are bots or auto-voting accounts. Of the top ten November 2022 voters, four wallets appear to be auto-voters. Of the top ten December 2022 voters, four wallets appear to be auto-voters. For both November and December, the number one most frequent voter and winner of the delegated VP reward, appears to be an auto-voting wallet. Because of this, I skipped VP distribution for the month of December so I could take some time to look at the voter data more carefully.
There is no way to be sure these wallets are bots, but they all share commonalities that paint a pretty clear picture. All suspected auto-voters on my list have less than 10 VP, all vote regularly and frequently in at least 5 snapshot DAOs (usually more), the majority of them vote “yes” or the first vote option on proposals, 7/8 of them have never logged in to Decentraland, and none of them hold a DCL Name.
If you dig a little into these account’s DAO profiles and voting history, a lot of their votes don’t make sense. Guest#60fd for instance, the top voter for December 2022, seems to have conflicting ideas on the controversial topic of the Tudamoon ban, voting to unban him from the Decentraland discord but to keep him banned in the DCL DAO discord.
If you examine the DAO profiles for the top voters in October (mrfg#1ba1) and November (N/A), both accounts generally vote “Yes” across the board. If you dig into these wallet’s snapshot profiles, you’ll find they have similar patterns of activity across many other DAOs, and indicator of spam voting in hopes of receiving airdrops or rewards.
Another comical example of auto-voters not reading or engaging with proposal’s content occurred a few weeks ago in the proposal: “Should the DAO restrict the ability for members with less than a certain amount of VP to vote on proposals?” This proposal was an attempt to filter out bot-likely accounts in the DAO by ignoring votes with low VP. Ironically, the majority of the wallets that voted in favor of this proposal (59 of the 66 yes votes), had less than 10 VP themselves. It’s easy to assume from there that these accounts did not read the proposal or understand its implications and likely fall under the category of auto-voters. It’s also unsettling to see such a high number (90%) of auto-voters in the “yes” category. It makes me wonder what percentage of the “no” and “invalid question” voters were also auto-voters. I’d be interested in seeing a study that estimates what percentage of voters (and DCL DAU) appear to be bot or autovoting accounts.
To many, the information that there are bots voting in the DAO and inside Decentraland is not news. The governance squads have brought it to our attention multiple times, and bot-farming was one of the main criticisms of the original VP for Social Capital post.
It is assumed that these auto-voting accounts are voting across multiple snapshot DAOs in the hopes of receiving a reward or airdrop. I worry that by offering VP rewards to top voters, I am just validating the existence of auto-voter accounts and encouraging them to multiply and scale up their efforts across other DAOs. I hope that by limiting this experiment to Name holding voters, it can continue the original spirit of the VP for Social Capital program while filtering out bots and auto-voters.
.
.
.
More on Subjective Interpretation of Data
When I started this experiment in October I used the DAO Community Engagement Dashboard as a source for voter data:
After digging around in the DAO website however I found four different places that showed voter information, each reporting slightly different numbers. I’m not 100% sure, but I believe one display reports total votes cast by an account, one reports the amount of times an account has voted in the past month, one reports the amount of times an account has voted on proposals that were created in the past 30 days, and one… I’m not really sure what it is displaying.
Thankfully after asking some questions, I was directed to the Decentraland DAO Transparency Spreadsheet which is the source of all the web displays and which updates directly from snapshot (with a slight delay). Since then, I have been sorting the data myself from the master excel spreadsheet to create my monthly voter lists. Based on these lists I have been delegating VP to top voters.
To create my voter lists, I filter all votes cast down to those in the timeframe of the past month, sort columns by wallet address, then total the amount of times that wallet has voted.
I’ve verified that the Dentraland DAO Transparency Spreadsheet does not list votes on deleted proposals and only records the most recent vote if a user changes their vote on a proposal.
Overall I believe my method for voter list creation is straight forward, recreatable, and accurate, but if anyone has issue with how I’m doing things, I’m open to suggestions.
.
.
.
Thank you for reading. I plan on posting another report with another update in three months on May 1st 2023.