Liberal Radicalism?

[lehnberg]

As a follow up to @hashmap’s gitter post, I’m creating this forum placeholder to discuss the Liberal Radicalism paper by Buterin, Hitzig and Weyl.

Having the read the paper, it’s quite ambitious and unusual compared to traditional economics and political science papers. It’s worth noting that it’s an initial effort that do not have holistic answers or solutions yet.

There’s been some attempts to comment on its contents, I’m sure there will be much more to come.

To cut to the chase and discuss how this might relate to Grin’s Governance, the paper is proposing a method to allocate funds between projects, based on voting, that attempts to address some of the shortfalls with traditional allocation models such as:

• capitalism, where whoever has money allocates it to the project they wish to fund, and there’s basically a plutocracy running the show, protecting their own interests to the detriment over other groups;
• one person one vote, where the majority will rule, populism is encouraged, and the minority risks becoming oppressed by the masses.

This is highly relevant for our governance model and how we end up funding different efforts, and there are many variations that could be applied. Consider for example a scenario where donors donate to a central fund, and as part of that receive votes that can be cast on the funding of future projects that can be proposed by members of the community. Or funding projects directly, and based on the amount of public funding, matching is being done from the central pool according to the principles laid out in the paper.

There are a couple of obvious attacks laid out in the paper, specifically around vote buying, collusion, and sybil attacks. Which brings me to what I consider the two most crucial limitations of the paper from the perspective of applying it to Grin governance:

@§3 page 5

we assume that an equitable distribution of basic resources has been achieved in some other manner, such as an equal initial distribution of resources

Practically, how would this be applied in our case? Who gets to vote? How are votes being distributed initially?

@§6.2 page 25

The largest practical impediment to achieving this is the anonymity and pseudonimity that is so per- vasive within [Open Source] communities. Clearly, LR relies heavily on separate human identities and can easily be exploited if identities can be replicated. Whether these difficulties can be overcome is an interesting question that experimentation and technological advances for identity solutions will reveal.

I.e. how would we ensure that each individual in the community is represented in our system only once?

Before these challenges can be addressed convincingly, I’m not sure LR would be superior to more simple and conventional approaches to funding.

[igno.peverell]

Sybil attacks on blockchain-based voting or representation systems is a real issue. Anytime I run across an interesting scheme (like LR), it’s the main roadblock. So I don’t think we’ll be able to experiment until this is solved. That’s why all current systems just rely on total amounts dedicated to a vote, which is of course plutocratic, but sadly the only one we know how to do.

Note that there is one mitigation: delegation. Of course the wealthy will still have more influence, but if 100 plebeians can delegate their votes to a representative, it offers some counterpower.

Here’s a thought experiment directly applicable to Grin I’ve tried to flesh out a few times. Imagine we have confidential assets, with an asset that represents a vote. Each year (or desired timeline), all Grin regular outputs get duplicated into Vote Asset (VA) outputs. Candidates or proposals can get listed and some minimal amount of real grin has to be burnt for them to be considered (to avoid too much noise). Once those are identified, VAs can be moved at will, directly to a candidate/proposal or to a representative who can in turn move it some more if desired. After the vote ends, candidate/proposal private keys have to be revealed for counting and all VA outputs are destroyed.

Overall, I’m still not sure how much value this would bring, but worth thinking about some more.

[monkyyy]

@lehnberg

capitalism , where whoever has money allocates it to the project they wish to fund, and there’s basically a plutocracy running the show, protecting their own interests to the detriment over other groups;

I think the criticism is more nuanced here; the words used is “linear” to put this into a political context he’s not criticism that some players have more control over the outcome, but rather humans don’t function in this game we are playing that well. He stated noble intentions for the supposed plutocracy but that they lacked information on how to effectively achive their ends. Actually I’ll just copy paste:

a sponsor with capital wishes to stimulate and support the creation of public goods
but is ill-informed about the appropriate goods to create.

Taking extreme liberties to explain it my way:(shifting the knowledge problem where I believe it actual is) I’m thinking of a trust-fund baby, experiencing isolation while never working a day in their life would be a good example here.

Not reaching for the normal criticism of greed or unfairness. But rather a “central planning doesn’t work” argument. Once you get so rich there is no such thing as a losing move, you lack the trail and error way to learn about the world that is part of the price system for normal people.

I.e. a poor person can be in a situation where they must choose between eating that week and enjoying a vice; and learning that they prefer eating to the vice. And can therefore learn about their own value system that “food > vice” if your system isn’t tuned and everything is at maximum all the time you don’t ever and no nuanced decision making comes into play; this element of the price system breaks down.

What being suggested then is by making the amount of money in the system quadratic (and therefore the value of that money sub-linear) in theory you never have the trust fund baby never learning to drop a vice.

---

Can we all take a moment to appreciate this bit of text

The Henry George Theorem

---

more recently Weyl and Zhang (2018) propose a self-assessed tax on capital with a right of compulsory purchase for any buyer willing to pay the possessor’s self-assessed price.

I believe this was suggested much earlier; ancient Athens actually. Someone should probably point that out before anyone trys it.

http://www.daviddfriedman.com/Legal%20Systems/Athenian%20Law.docx “liturgy”

Maybe don’t spread it to far I’m trying to make a card game on this idea. Its just a bad idea for society.

---

D1, D2 and D3

I think this bit is near pointless dribble; its like they forgot value is subjective. I would suggest a d4 that old “consent of the governed” aspect.

LR, insofar as it solves the liberal-communitarian debate,

You did what now? At best it framed the debate

---

My take is rather simple; eth-guy drank his own koolaid, I believe in organic government yet here he is again trying to design a inorganic one as if he didn’t live thru the dao

His math may however be interesting for charity projects, for reasons he explains exceptionally poorly but are rather simple: big donors don’t necessarily want to have their way and their goals are fuzzy in nature.

You could likely go ahead with his little experiment. I would however suggest two very clear differences:

1. Make it so however you view the data you can move a slider that would let you see the out come of “capitalism” or “democracy”, and everything in between on the fly

2. Don’t promise to implement the results as the math says, have a human make the final call based on the maths.

[lehnberg]

@igno.peverell

Sybil attacks on blockchain-based voting or representation systems is a real issue.

I would even go as far as saying it’s a real issue for any (feasible) system for Grin, whether it being blockchain-based, or a typical centralized service. Because we will not (and want not) to be able to identify ourselves as individuals, we will not for sure know whether multiple accounts / wallets / identities are in fact belonging to the one and the same entity/group.

@igno.peverell

Note that there is one mitigation: delegation.

Yes, I’ve been thinking about this as well. It increases the effort to sybil attack, and adds the benefit of allowing less engaged users to influence governance without having to constantly be active and participate.

@igno.peverell

Imagine we have confidential assets, with an asset that represents a vote. Each year (or desired timeline), all Grin regular outputs get duplicated into Vote Asset (VA) outputs. Candidates or proposals can get listed and some minimal amount of real grin has to be burnt for them to be considered (to avoid too much noise). Once those are identified, VAs can be moved at will, directly to a candidate/proposal or to a representative who can in turn move it some more if desired. After the vote ends, candidate/proposal private keys have to be revealed for counting and all VA outputs are destroyed.

To expand on this, imagine if this was two phased:

1. VA outputs vote an arbitrary amount (let’s say 33 as an example) of candidates into a “Grin Parliament”.
   • At the end of the voting period, all VA outputs are counted, the 33 candidates with the most VA outputs amassed are elected.
   • Any unspent VA has no value, any VA sent to the 34th candidate etc has no value.
   • Actual VA tally of each candidate is not known until after the end of the voting period.
2. Grin Parliament then goes about to actually make decisions related to the project, funding for projects, allocations, development etc.
   • Here, each candidate could be allocated new “Parliament Vote” (PV) Outputs, according to the share of VA received.
   • Or, each parliament candidate could be given an equal share of votes. Depends on what the desired properties are.
   • Vote casting could be secret, to reduce the ability to bribe elected candidates effectively.

Some of the benefits of this two-phased approach:

• This reduces the impact of wealth compared to just voting directly on projects. You would first need to place members into parliament, and then have these members co-ordinate and carry out your wishes.
• This reduces the ability to do Sibyl attacks properly, or they become more obvious at least.
• Because the VA voting is hidden, there’s a co-ordination problem to manipulate parliament elections as they will not know how many VAs will need to be spent in order to be successful.
• The parliament representatives could then use Quadratic Voting to cast votes, as the identity problem is then somewhat addressed in this context, but still not completely (see below).

There are still major issues with this approach. Just to name a few:

• Wealth still holds a lot of impact. If you have a lot of VAs, you can make sure you get a certain amount of candidates in.
• Parliamentary model will encourage corruption and regular politics. Campaign promises, popularity contests, and populism. (Vote for me and I’ll do airdrops!, or whatever other catchy slogan resonates with the masses)
• Sibyl attacks are still possible. An attacker could for example create three sock puppet candidates that all run for Parliament, give them individual personalities and political platforms, and thus hedge their bets, and potentially get all three into parliament and then have them vote whichever way they want.
• Once elected, a candidate could also just sell their account to an attacker and have them cast votes on their behalf. This way the attacker can circumvent the secret ballots and pay the candidate, knowing that they get the votes they paid for as they cast them directly.