Meeting - 5th July 2022
Meeting 2 - Fund 8 - dRep White Paper Working Group Meeting
Last updated
Meeting 2 - Fund 8 - dRep White Paper Working Group Meeting
Last updated
Aharon Porath
Consenz
Kenric Nelson
Steph Macurdy
Stephen Whitenstall
Thorsten Pottebaum
George Ramayya
Frank Albanese
Philip Lazos
https://quality-assurance-dao.gitbook.io/community-governance-oversight/
Documentation funded by Fund 7 - QA-DAO Transcription Service
https://github.com/Catalyst-Auditing/Community-Governance-Oversight-Coordination/issues/86
Brief Updates/Notices from each participant - All
Kenric Nelson - 00:37
Steph Macurdy- 01:14
Aharon Porath - 01:35
Thorsten Pottebaum - 02:10
Stephen Whitenstall - 02:37
Any Budget updates or issues ? - Kenric Nelson
Project Start - 1st June 2022. End - 30th September 2022
Small amount reserved for community members
Thorsten to recruit reviewers - August 2022
Any updates on Project Plan timetable and deliverables - Stephen Whitenstall
No updates
Please prepare Detailed Outline of your section for next meeting - All
Philip Lazos
Email follow-up - apologies sent
Aharon Porath, Kenric Nelson
How different building blocks will work together. How fits into Consenz platform. Liquid democracy is the main component. But also references direct democracy. Will be discussed further in Consenz meetings. - Aharon
If you prefer to work in your own document can you provide a link in the main White paper document ? In August 2022 we will open up to reviewers and then the public at large. - Kenric
Will we use an academic peer review process ? - Aharon
The White paper approach is editor based and community reviewed. It will be formatted as an academic paper. At the end of this process we will decide on whether to submit to a journal. A Follow-on proposal in Fund 10 is possible for a peer reviewed process. In general I see the academic research process spanning 3 years. - Kenric
What do you mean by original results ? - Aharon
Our goal for this project is to define the measurements and to start building the tools to monitor the process of decentralized decision making and to monitor the outcomes of that. How good are the decisions and do they produce a strong ecosystem? - Kenric
Those are the things we really should be focused on. You are doing that Aaron by documenting the processes that go into producing high quality documents as a community. - Kenric
Steph Macurdy
An original result would be to calculate the Banzhaf Power Index for the Catalyst community.
That is a big goal, because defining the processes of how Catalyst works into a formula for calculating the power index has subjectivity built into it.
How do we define the rules and parameters of what constitutes a voter and how do we get that data?
I met with a Wolfram expert on the function that outputs coalitions or entities. - Steph.
But what we could do is take some simple examples.
Kenric brought up the [Genesis] key ownership between IOG, Emurgo and the Cardano Foundation as a simple example. Because there's three entities and seven total votes. How do we define what the power index is within that group? - Steph
Kenric and Steph talk through the Banzhof process of simplification - treat each wallet cohort as a single voting block.
I've given you access to the data we have from Funds 3 through 6. It is probably good to just talk through it. - Kenric
I think the first step is some simplifications.
Where we treat each wealth cohort, wallets with between 500 and 5,000 and then between 5000 and 50,000, etc, going out as a single voting bloc. That way, there's not as many players and the computational calculations are dramatically reduced. - Kenric
By breaking a coalition into two we can play with the percentages. It gives us different outcomes. Synthetic data to see if we have any dramatic swings in power. - Steph
Does the data we have tell us which wallet voted for which proposal ? - Thorsten
We do have that breakdown on those on Funds 3 to 6. But that's quite a lot of data. The combinatorics of that are pretty immense. - Kenric
I'm thinking the principle of Banzhaf is to express whether my vote had an impact. And you could approach it now from two different directions. If you take the biggest wallets, look at outcome of the vote is, and look at the voting of this wallet. [That might provide an insight into] into the level that [wallet] effectively mirrored the outcome already? - Thorsten
That would be an indirect measurement of how far a specific wallet is determining the outcome? If it were always one to one, then it could be a dictator situation where the voting power of that wallet flips [the result] every single time. - Thorsten
If you take the top 10 or 20 wallets and do analysis on that set to see in how far the voting of these wallets mirrored the actual outcome. And that might be an indirect indication [of the power index] ? - Thorsten
My colleague and I did some initial work which we weren't able to complete that is similar to what you're describing. Even if you don't have the mapping, in addition to doing the combinatorial theoretical calculation that Steph is describing, you can also do a kind of empirical analysis of an actual vote. - Kenric
So you can look at the results and say how many times could you look at the whole outcome of all the votes? And examine each wallet, [considering] each wallet like one additional vote. Would that had been enough to win, or the other way around? - Kenric
If you took the winning projects and subtracted one wallet. You could look at the percentage of times that a particular wallet had the ability to change [the outcome] from a winning to a losing proposal. And get a kind of empirical Banzhaf index for that particular vote voting round. - Kenric
There is probably a second element to that as sometimes close votes can be swayed by small wallets. - Thorsten
Sometimes exactly right. If the vote was very close, then even small wallets would influence it. But what you'll see is a kind of cumulative distribution function where the large wallet always can influence a larger percentage of outcomes. - Kenric
It will interesting to see which dReps reference a certain percentage or vote. So one can determine the vote in terms of dictatorship or being the deciding vote to effectively push a certain proposal through. To understand in which areas there is a concentration of voting power. Or where are borderline cases and which mode might have a difference. - Thorsten
Kenric Nelson
What I did here is I inputted three players to represent IOHK, the Cardano Foundation and Emurgo.
Then I inputted the initial weights of these seven keys.
So [originally] IOHK had three, the Cardano Foundation and Emurgo each had one.
A vote of five total was required to make a decision to approve a decision.
One of the things you see right away is that in order to reach five, you have to have approval from IOHK, Cardano Foundation and Emurgo, we're not able to make decisions as a pair.
And that has a significant influence on the power that these players have.
So if the way you you look at all the potential coalition's if someone wanted to try to act on their own, for instance, but in this case, those would all lose, or if you have pairs. While two pairs that include IOHK are winners, pairs that exclude IOHK lose. And then there's the coalition with all three approving that's also a winner.
In bold is shown the five cases where you have a critical vote. The critical vote means that if you took that vote out, the win would turn to a lose.
As you look at the ratio of the at the bottom you see that IOHK ends up with a power index of 60. Even though three sevens is more like 43%, their actual power is 60%.
For Cardano Foundation and Emurgo they start with two sevens, which is about 28%. But that gets reduced down to 20%, in terms of their actual influence on outcomes.
What strikes me is, most people would not be able to see that subtle increase in power that the person with three keys gets. There is this kind of subtlety in the power distribution that's not very visible to most people.
Furthermore, I think this plays out in the dynamics between the three players over time. For instance, [...] maybe I won't even refer to the names anymore as I go forward. In the abstract, if the two weaker parties feel strongly about a particular issue, but they can't get agreement from the stronger party they may disenfranchised. Over time, if they can't influence things they disagree on with the more powerful player they may feel that continuing to be involved in this game isn't to their benefit.
To conclude, the recent conversation within the community has been that at some point, the Cardano Foundation and Emurgo delegated their keys to IOHK.
The second player who still holds two keys has no power left. [...] Just one person delegating gives IOHK dictator control over the decisions and the other player becomes a dummy vote. Even if they choose to vote, it's never going to influence anything.
This is a nice, very simple example. It illustrates why this work is so important because even in a more complicated game, with many more players, these kinds of power issues are going to play out.
If you are going to set up a system where decentralization is an important value to maintain over a long period of time you have two very difficult things to accomplish.
Number one, you have to set up initial conditions that favor decentralized decision making. But that is difficult if the initial conditions favor a few people. Those players will tend to have the ability to further increase their power in the system. You get a kind of natural convergence to more centralized decision making.
I don't feel that this is well understood in this community that the process towards centralization is very natural, it is the default. So you have to have controlling elements to guard against that, not just in the initial conditions, but in the way you monitor and take corrective action when you see it. When you see the decentralization process, degrading.
There was a famous result called preferential growth. What this showed is why networks tend towards a power law distribution in how many connections a particular node has. The larger nodes tend to grow preferentially, because new members to the community have a preference to link up to popular nodes.
For instance, if you go back to the early days of the Internet there were 10 different, very competitive search engines. But as Google became [the preferred option] people got the most use out of its position and it became extremely dominant. To the point that it's now centralized a lot of the functionality of the internet. And [similar effects] are playing out already in the cryptocurrency space. - Kenric
There's a kind of folk folk philosophy going around Catalyst with the phrase "You can have decentralization by centralizing first". And it's a perception that's just wrong. - Stephen W
It's not to say that decentralization doesn't have strength and can't work. The whole history of Linux is a very powerful case study. A lot of people were dismissive of the Linux community in comparison to IBM, Microsoft, and Apple who were quite dominant in computer architectures and and software. And yet, a few later, a very central player, IBM kind of accepted that Linux had become dominant and they built Linux into their own architectures. And then eventually bought Red Hat.
So a kind of decentralization did win in this instance. But in a sense, it still gets captured by the central powers within the computing industry. - Kenric
A brief update over Charles's talk [Roadmap and Governance], there was a mention of something called an Edinburgh decentralization index. There is documentation of that I can provide links to. And it's obviously something we need to be aware of going forward in terms of how this work will relate to decentralization index. - Stephen
by Kenric Nelson
When Cardano was initially set up, there were 7 genesis keys established that are still used for CIP changes and Hard Fork Combinators. Initially, they were distributed 3 IOHK, 2 Cardano Foundation, and 2 Emurgo. A vote of 5 is needed to approve a change. Most people’s intuition is that IOHK’s 3/7 votes is only a slight advantage. In fact, this is enough to establish veto power.
Using the Banzhaf Power Index website, one can compute the percentage of decisive votes for each party. As shown in Figure 1, IOHK is critical in 3 of 5 instances. A critical vote refers to a winning coalition in which dropping that vote would cause the coalition to lose. CF and Emurgo each have 1 of 5 critical votes. Thus IOG has a 60% power index compared with its 43% number of votes. CF and Emurgo only have 20% power despite 28.5% of the votes. The subtlety of this nonlinear power relationship is an excellent example of how savvy players can set up systems that appear equable but actually have quite distorted power relationships.
Currently, at least one of the weaker parties (CF or Emurgo) is reported to have delegated their genesis keys to IOHK. This gives IOHK the 5 keys necessary to be a dictator regarding the CIP and Hard For Combinator updates. While the goal is to transition during Voltaire to a member-based organization that manages the genesis keys, this example demonstrates how a slight advantage in the voting ratio can lead to a substantial advantage in voting power. And an early advantage in voting power can be used over time to further increase this disparity.
This example provides a power illustration of the importance of monitoring and controlling the distribution of Banzhaf voting power in the Cardano community processes.
Charles Hoskinson on the Cardano Roadmap and Governance. 30th June 2022
adanamics ()