Steward Selection


#1

I would like to comment on the “Steward Selection” part of the “Sovrin Provisional Trust Framework Working Draft 01” which we discussed on yesterday’s Trust Framework WG call. I am especially interested in the following proposed rule:

“There MUST NOT be more than 25% (PN 40% with Non-Profit Exception) of Stewards in the same legal jurisdiction or nation state.”

I believe that if Sovrin is to be a public digital identity utility for all of humanity, then this rule is not good enough. I also believe it is an impossible task to come up with a steward selection algorithm that accurately reflects all the cultural/political/legal/etc. diversity we have in the world.

At this point, without making a concrete proposal, I’d just like to share a few mixed thoughts, and would like to invite others to comment:

  • Should all nation states be treated equally in the selection rules? Should Liechtenstein (population 40k) theoretically be allowed to run the same number of stewards as India (population 1.2B)?

  • If all of the world’s nation states are treated equally in the equation, i.e. are allowed to run the same number of stewards, then stewards representing only 5% of the world’s population could potentially achieve a 2/3 majority.

  • Should the amount of stewards in a nation state depend linearly on its population? But then small nation states would have almost no influence.

  • When considering the amount of stewards per nation state, there are probably endless factors that could become political tension points. Should the U.S. as a single nation state only be allowed to have the same number of stewards as each one of the 15 former Soviet Union countries could have individually?

  • In the E.U. Council, voting power depends on population, but not linearly. E.g. Germany has 3 times as much voting power as Austria, even though its population is 10 times as large.

  • In the U.S. Congress there’s a mix between population-based voting (House), and area-based voting (Senate).

  • At the U.N., you have one vote per nation state in the General Assembly, but in the Security Council you have 5 permanent veto countries (victors of WW2).

  • How to take into account cultural diversity. There are efforts to establish metrics of cultural distance between nation states, e.g. “Hofstede’s cultural dimensions”. With such a system, you could try to prevent a set of stewards from a “culturally too homogeneous” background. What about religion and language and ethnicity and many other aspects of human diversity.

  • How to take into account non-nation-state entities, intergovernmental organizations, international non-governmental organizations, regional organizations (e.g. United Nations, Red Cross, European Union, Arab League, Commonwealth of Nations, etc.).

  • How to take into account the difference between the concept of nation states and the concept of jurisdictions that was mentioned on yesterday’s call (e.g. Hong Kong).

This is now about steward selection. but perhaps in the future there could be similar considerations also for trust anchors, trustees, etc. All of this is obviously impossible to enforce by software code. Our approach of having a human governance layer to make these kinds of decisions is what sets Sovrin apart from other efforts. The better we can make this human governance layer, the more successful Sovrin will be.


#2

I really like these thoughts, Markus!

I want to point out, however, that they may get at a somewhat different issue from this rule as originally worded. Your focus seems to be on voting/influence/participation for a population/culture/industry. The rule as originally worded talks about legal jurisdictions. Both issues are important.

The legal jurisdiction matters because we imagine scenarios where subpoenas or regulations could have the potential to impact Sovrin. We don’t want any particular legal system or government to be able to do that disproportionately. In that perspective, Liechtenstein’s legal system and India’s legal system are qualitative and quantitative peers, and must be so; otherwise a legal change in India’s stance on privacy, for example, could have a disproportionate effect on the network vis-a-vis Liechtenstein’s. This jurisdictional requirement probably manifests in the proportion of validators, mostly.

The culture/population/industry proportional representation also matters, for matters of fairness and skew. Suppose 75% of the validator nodes come from universities. They’re balanced or safe in terms of which legal jurisdiction they come from, but they’re not balanced by industry; too many are from education. This would allow academia–the mindset, the tech stacks, the priorities–to have undue influence within Sovrin. Substitute other subsets, based on culture or population, for my university example, and you get all the other issues you are raising.

It seems less clear to me what the proper answers are to your questions, Markus. I’m not proposing any. I’m just saying that they don’t invalidate the legal issue. The good news, I think, is that we don’t quite have the same complexity that a government does, because we can assert many requirements and then be constrained by all of them, without reconciling them:

  • We can’t have too many validators in a single legal jurisdiction, regardless of how “big” the jurisdiction is. This can’t be proportional.
  • We ALSO can’t have too many validators in a single industry vertical or a single culture or a single timezone or a single X, but perhaps “too many” is slightly more complex to analyze.

#3

Great discussion.

The purpose of this rule is not to ensure equal representation, but to ensure that a sufficiently large group of stewards to threaten network operations couldn’t be coerced by a government. We should have similar concerns (as Daniel points out) about industry verticals and other groups that could hold too large a sway over decisions.

As I see it:

  • The TGB and stewards are representing the technical operations of the network.
  • Network members (people, organizations, etc.) are represented by the Board of Trustees.

The rules around the latter have not been formalized yet, but the Exec comm of the board is discussing them so we could bring a proposal to the board of trustees sometime soon.

That said, I don’t see either body (BoT or TGB) being sufficiently large to provide realistic proportional representation.


#4

Thanks for these answers. I understand and agree that the “legal jurisdiction” component is very important in steward selection, and basing the selection rules on this component is sufficient for building a digital identity system that is by far superior to anything else we have today. But I also believe that it isn’t completely disconnected from the representation/diversity considerations.

Let’s assume we decide on a 15% limit for stewards per jurisdiction.

  1. We could then e.g. have 12% each in the U.S., U.K., Canada, Australia, and New Zealand for a total of 60% in these countries. This doesn’t sound bad, but they are all all members of the Five Eyes intelligence alliance, and an individual let’s say in Brazil or Germany may not feel comfortable rooting their entire digital identity in this system.
  2. Or you could have 15% stewards each in China, North Korea, Iran, Russia. This would be compliant with the rule, but an American credit union may not want to base the security of its financial transactions on such a system.

If the aspiration is to build a system for all of humanity, then these questions matter. I also don’t have an answer right now. It is possible that my representation/diversity questions apply more to trustee and trust anchor selection than to steward selection. Ultimately, these questions apply to all potential sources of threats to the sovereignty and neutrality of the system.


#5

I have been tied up heavily in travel meetings the past two days (and don’t return to Seattle for one more day), but I just wanted to say this is an excellent discussion and IMHO exactly what we should be debating and deciding here as a community.

I’m digesting the points made so far and will try to post more thoughts on this by Friday.


#6

I asked a few friends about this and got some feedback.

One approach could be to favor/limit places according to indicators like strength of the democratic system, civic freedoms, etc.

E.g. see tools like the following:

Another factor could be diversity, i.e. favor/limit steward locations based on the predominant language, religion, certain cultural indicators, whatever.

I realize even talking about this could mean navigating a political minefield, but hey I’m not afraid to bring it up!


#7

@peacekeeper My thoughts were going in the same direction. What we really need to accomplish with the Steward Selection Policies in the Sovrin Trust Framework is just one thing: sufficient diffusion of high-quality stewards around the world such that no one country or industry coalition or other disruptive force could try to interfere with the network.

This afternoon I am going to be trying to consolidate the suggestions for how to do this into an updated set of Steward Selection policies for the Provisional Trust Framework and will post a link to that within the next 24 hours.