How to Design Better On-Chain Governance Models While Protecting Privacy

Transcript

Hello everyone. We still wait 2 minutes more. Okay. Yeah. Hi everyone. So cool to see all of you here. And for those who haven't seen me before, I am Soprina. I'm the the co-managing director of the blockchain souk, the joint research initiative.

And yeah, in the today's world, tech shapes more and more the decisions in our life. So um but how is that governed? Who decides what? And how do we address privacy when some technologies are built up for to be fully transparent and others reveal nothing at all? And this is um our topic over the next 30 minutes. We will tackle three big questions.

what does governance actually means and uh what's working and what's not and what we can um take forward beyond blockchain and beyond DIT in general. And now I'm really excited to have three leading voices here with me. I warmly welcome Shirley Walsh, chief of growth at Paticia blockchain. We have Valentina Andrea, director project management at the Hashgraph Association. And we have Gazi Ben Amour, vice president corporate development at Sama.

And last but not least, we also would like to hear your voice. So uh we reserved some time in the end of the panel for short Q&A. But yeah, let's start and we talk first. Oh, sorry. And let's talk about the term governance. It's such a broad term and um it has not that clear definition. So we are using it for different kind of processes like decision making um or also like in Dows for steering discord channels.

So Gazi what's your take on governance and why it matters can you shed a bit light of on yes so I will take a very theoretical definition of governance as per my view which is you know we take a process we all agree that the process has been executed on the correct data and we all agree that the outcome of the process is correct. has been correctly executed. So is correct. So we are able to expose that process. We're able to expose the inputs to the process and we're able to expose the output of the process and we all

agree that it is indeed how we intended it to be. Thank you. Yeah, because he explained it on a on an overall level very clearly, but in tot's context what we're focused on uh around governance is for example onchain voting where privacy is crucial because uh having onchain voting you want to you don't you want to exclude the uh introduction of bias to the voters uh in the mix. So you want to eliminate the um external uh actors and uh making an impact on their voting uh decisions as well as um um

having like uh maximum fairness in the in the voting process as well as uh verification. And um we also have uh decentralized access management where privacy is built in around um custody of uh digital identity and verifiable credentials and uh and also for custodians on the financial assets. So um we can also have governance around around such topics and uh and uh where privacy is crucial. Thank you. Valentina, do you have something to add? Yeah, I mean maybe if I take it uh even one level higher. Uh in essence, I think governance so

specifically when we're talking about in the context of uh decentralized systems uh in essence governance is responsible management. So making sure that that it's managed in a responsible way. So if we take the parallel like with the web two world and we take a company and it's managed by board of directors executives the management if something goes wrong so if it's mismanaged the company will suffer but it remains at the company or maybe up and down the value chain the suppliers potentially some clients they will be the ones impacted. However, when we're talking about an infrastructure like a blockchain or a distributed ledger, if it's mismanaged, the impact is much

wider because it impacts all the companies so to speak, even potentially markets, industries that are built on top of that. So, the loss can be colossal. That's why governance is so important and yes, that's I would say why it matters so much. Thank you, Shirley. You spoke about privacy enhancing tech under the GDPR this morning. So now when it comes to governance, especially onchain governance, there's always this tension, what should be onchain and what shouldn't?

How do you at Patisia deal with that balance? And where do you see the biggest tradeoffs right now? Um yeah, so we're working on creating this balance between fully anonymous systems and fully uh public systems because you can't create trust in none of them. So that's where the privacy enhancing technologies and specifically MPC comes into play in order to create this uh even playing field. Um uh so yeah, I I believe uh that's that's where the balance um comes from uh using things like fully homorphic encryption or CK or or MPC. Um and the trade-offs do you think trade-offs is exactly if

you have a fully transparent systems then you can't achieve trust because uh the uh counterparty can use your data in a malicious way. For example, if you reveal everything about yourself so you need to have balance there in things which you keep private and which which you keep public. Valentina many assume that protecting privacy requires everything onchain. So if we have a look to hideera the protocol behind the hashgraph association we see that hideera has offchain governance. So what makes this a strong model probably can a little bit shed light about that and what probably we misunderstood about

the concept. Yeah that's that's a good question. Um I think one common misconception um about Hideera or in general about off-chain governance uh is that off-chain necessarily means centralized which goes against everything that we believe in blockchain and the decentralized world. But actually that's that's not true. Like so if we take Hideera for example um yeah so it is governed in a distributed way. So it's governed for those of you who don't know Hideera it's it's a DT that's specifically targeted for enterprises and governments and regulators as well and it's governed by a group of uh 39 I think as of today we have companies so from different industries so it's big

companies like industry leaders such as Google IBM Boeing we have the London school of economics as well and um they basically manage the network so they each run a node uh they manage the the network. They also um make decisions in terms of how the network will be upgraded uh where the funds how the treasury is managed, where the funds are distributed and all of this is managed across different committees. So I think that one um strong suit for that so first is that you have trust because by trusting when you build on on Hideera you trust the companies who are running this network. So this is particularly appealing to to enterprises who want to

build their their applications there. Another thing is I would say because you have dedicated committees, you have the correct people. So you have experts for example in the technology who are part of the committee deciding on the technological upgrades. You have people with more a financial background for the distribution of the treasury. And there's another committee for example for the utilization of the network, how it's used, who are the different um use cases that are built there. So I think this gives you an element of uh expertise as well. So it's not that you just vote because you have tokens so you're allowed to vote. It also ensures that whoever makes the decisions um knows what they're talking about and you

can trust the decision- making process. And maybe one other thing to add here is um the way that it works is that it's not depending on the number of tokens that you hold your vote is bigger or smaller. It's you have one council member equals one vote. So it's pretty equal. Yeah, I would say those are the I could there's more but those are the main advantages. Yeah, I could also come back. We're also believing in this uh hybrid model of public uh and private network. So what we see also from the from the enterprise adoption side they're uh interested in this hybrid model where you have traceability and accountability on the public side but you keep uh all your confidential data

and processes on the private network side uh whilst protecting it with something like MPC uh so um it's similar to how the web 2 has been architected uh to adhere to all the data compliancy um so uh so yeah that that would be the model I think which would be compliant in the long term. Thank you. And now if we take a look into the future, probably we see then everywhere aentic AI. We will probably see everywhere the physical AI means human and machine collaboration will be at some point in in the future super the norm.

So, but what does that mean for governance and um for privacy and therefore also for data sovereignity and um yeah Gazi Zama specialized in the fullomorphic encryption means that during the entire computation the data is completely fully encoded fully encoded. So um could you briefly explain how FGE works in practice also probably with some use cases? So I'm good. I'm happy that you remembered.

Yeah. So FH what it does is an algorithm that allows to manipulate encrypted data without having to decrypt them. So you take two numbers, you can add them, you can compare them, you can multiply them, you get another number without ever knowing what you had before. Now apply it first to blockchain. Now we talked about governance. Governance in a public blockchain is the ability to reach a consensus that the validators have executed a smart contract correctly. So they've taken inputs, they've all executed the same smart contract, they all reach the same output, they reach consensus that the output is correct, they publish it. That's how governance works in a public

blockchain. The issue is that everyone knows the input. Everyone knows the output. With FHC, a validator can execute the smart contract on encrypted inputs. They'll get encrypted outputs. They'll be able to reach a consensus on the encrypted output without ever knowing what they've done. And our technology is super flexible in two ways. You decide in your smart contract what data is encrypted and what data is not. So you can decide in the context of a vote, my vote is encrypted but maybe not the date at which I voted. In the context of a transfer, my balance is encrypted but maybe my nationality is not because I still have to pay taxes.

The second flexibility is you could be on any blockchain. We're a protocol that adapts and integrates into blockchain. Today we're going into the EVMS. In the future we're going for other blockchains. And the third one is it's developer friendly. So a developer who knows in the EVM world how to code just say this data is encrypted. I code I code I code I transfer money. Now that money is encrypted. Now the other part that is important and this is what you mentioned in AI is who gets to see what because now my data is in a public domain. It's it's encrypted.

I may want to share it with others. And so there is an access control that is built on MPC where on the blockchain itself you decide who can see what and who can compute what. And so with that you're able to bring governance to the data that you're exposing on a public blockchain. Why does it matter? I mean AI obviously you can decide what is leverage and what is not for a training. uh but also auditability. If the authorities need to know something about me, I can also transfer the rights for them to check whatever they need to check. So, and that's an important part for clients, banks that have to be

compliant with whatever regulations are in place in their countries. Absolutely. Do you have Yeah, basically we do it the same. It's just with MPC and blockchain combination. And uh we see a lot of interest from uh projects doing AI verification. So how do you verify that the AI model is what it's supposed to do in an ethical way? Also how do you train on a data that should be compliant. So this is where the MPC can be used. Uh we're actually coming out with fully arithmetic uh MPC which allows federated learning on confidential data. Um so that's that's how it's changing game.

And do you know how Hideera will um will use it in the future or how is the outlook there? Um I think yeah I mean so obviously I I agree with what you said regarding the the data privacy. So where is the how is the data used but also where is it coming from and where is it going and I think that's where DTS will also play a role in these kind of models uh in the future is to because obviously the data is output input somewhere else and it just continues down the chain. So I think yeah DTS will play a big role um for that in the future. Uh but maybe also a bit on the governance aspect. So

if we think of um like uh the agents of the future who would be able to take actions on our behalf um so potentially make transactions for us, sign contracts for us, place orders, anything like this if they make a mistake in essence, you know, who is then held accountable or or responsible for this. So I think in the future something that we all have to think about is also how do we make sure that the governance that we have in place also can keep up with this and takes into account both the data element that's brought by the by AI but as well the human element that we can have in place um yeah mainly well the ethical part but also uh the decision making uh

aspect it's also the explanability right you know it's uh being able to explain to humans why their credit score is not allowing us to get a loan. Yeah, absolutely. Um, yeah. And now I would like to open up for you. So, are there already questions? Please raise your hand and my colleagues left and right would manage the mic. So, no.

Then yeah over there. Just a sec. The mic is coming. Thank you very much. Um I I would have one question. Um about eight to nine years ago, I was involved uh in well I was leading a project to uh give back to policyh holders. the complete control and ownership of the their personal health related data.

Um the idea was really to not only a enable to um hide the data but also to reveal it for certain people for a certain amount of time um in view to whatever needed to be done including transactions, including contracting. And um I remember then that back in the time there were other stakeholders or players in the market trying to address the topic. Um but nothing came out of it. And um I know that my project was frozen

because I've been told that um there are other players trying to produce the same outcome that might be faster than I was, which wasn't the case obviously. Um but um I I would like to ask you uh where we stand because if there's anything closest to identity, it's our health and anything related to it. And uh the way we process that data is so unclear that it's even unclear to us where we stand. Okay. So I think it's a good question. To whom would you like?

No, I would like to ask the entire panel. Okay. I mean, that's what we've focused on the past years. I think earlier it wasn't the case because there was no it wasn't that popular or it wasn't there. Um I mean we've been building or developing MPC over past 37 years and we're only now there when it's becoming popular and more aware. Um and it's down to low code level. So basically it allows you to put the healthcare data of of an individual into use and monetize on the use of it not the data itself whilst keeping it all private and uh and then you're able to customize the research uh customize the computation and the only the output will

be public and you can even manage and uh control the access who gets to manage that output. Um so I think we're only now in that phase where the architecture and different privacy enhancing technologies are are facilitating such use cases. Yeah. I mean go ahead. I think so it will touch on two things. The first one is what do you do with the data is not necessarily a technical problem and I'm sure Valentina will be a lot more clear on the answer but you know you may give consent to use certain data for research what kind of research you know if it's to cure certain diseases maybe you're

fine if it's to harm certain people maybe you're not a and that's not technology that's ethics right and and so uh today it is in place. We work with actors, government actors, you know, like in France you have the data health hub that would first analyze and this is humans analyzing what is the actual objective of the project. Is it aligned with what people want to do or not? Do we have the right samples? you know, if you're focusing on a certain type of problems, maybe you don't have enough data to either offiscate the data because it's just that guy or maybe we don't have enough to reach meaningful uh

decisions. So, so that's the first part which is making sure that the request is quotequote ethical. The technical part to be honest there are many technologies FH MPC is one of them but there are many others and it could just be putting it in a private server on a bunker and making sure that there are no USBs you see there are simple ways to get certain data accessible uh we do work with FHE because it's what we do and because it works but there are many others and I would say the first problem is definitely the hardest you know judging whether you know the the ultimate objective of the access of the data is

aligned or not with what the user wants and what you know the ethic wants. Yeah, I agree. I mean I think healthcare data is probably something that speaks well to most people that everyone can understand because it's applicable to us and I think that's also an excellent case for uh decentralized digital identity. So I understand uh surely you also have this u working on your end as too at the hashgraph association where we're developing solutions around that and um yeah I think that's we're actually next week hosting an event in collaboration with the crypto valley um focused specifically on that on

healthcare. So this is definitely there's some interesting use cases that we're exploring there and how um decentralized identity can help to solve uh some of those issues. So do we have another question? Yep. Over there. Um I have a question regarding the homorphic encryption voting you mentioned. So first you said uh EVM chains how are the gas gas costs like compared to uh the same calculations on non- enrypted data and the second thing is GDPR um related. So um I mean there's a right to be forgotten and even

encrypted data could be seen as personal uh data. So how does this fit into the yeah into the scheme? Yeah. So the first answer on the gas cost that the gas associated to uh the blockchain on which the smart contract is running and this is fairly comparable to non- encrypted one. It's a bit higher but it's not it doesn't break the the bank. Then we process so when you execute f operations you publish events that we listen to and that we execute as synchronously. We typically don't charge for the execution. we charge for the decryption. So the vote is a great example where it's economical because

there is only one decryption. Did the vote pass or fail? But in sometimes it's a bit a bit more decryptions. Uh the second part uh you were uh um sorry I forgot the second part. So it's a GDPR GDPR. So GPR we actually had I think two three weeks ago there was um an opinion from the European Union about the right to be forgotten in a public blockchain. We're participating to the effort. We're trying to create something called the right not to be decrypted. So it's still you still need you know as a public blockchain you still need all the blocks to make sense that the ultimate state is

correct but we could prevent the decryption of your data. So in essence it is there you can still derive whatever you need to derive from the data but you're not able to decrypt it. You're no longer able to use it in other smart contracts. It's not passed as a law, but this is one of the multiple responses that the blockchain European blockchain industry is presenting as ways to respect the privacy of a person while continuing to function in a public. We have a solution for different solution uh where uh which I which I um proposed in the morning and

showcased in the morning and which we have seen is the most adopted one by by players that we work with. So uh it's a basically a combination of private and public blockchain um which is the private is also distributed network but it's just trusted uh stakeholders and node operators uh where you keep the identifiable data on the private network and then just for verification and um auditability and uh uh traceability you keep it the rest the logs you keep on the public side of things and then we have a smart contract langu language which enables ables you to uh reconstruct the transaction uh if an authority comes in or if the person

comes in and wants to erase a certain data about themselves in a in the further his um future. So there there are technically and cryptographically solutions out there. Yeah. On private blockchain it's easier I would say to re hybrid. Yeah. Are there more questions? No. Then I would have a question and another one. If there is one thing, a principle or also maybe even a warning what we should carry into the next infrastructure or digital infrastructure or next generation. What would that be from your

side? Gazi, would you like to start here? Confidentiality for sure. Uh I think we're still we need to evolve because uh as you see like the new GDPR is coming up to blockchain protocols and technology players in web three field. So uh I mean if you take a look at the Dows at first we thought we can just create an open-ended democracies uh where anyone can vote and anyone and everyone uh with an opinion has a right to vote and that would create a perfect system but it's it has proven to be not the case. Um so I think we should only have um like people with expertise and uh and like intellectual uh voting

rights in a DAO and I I see this landscape already changing. Um so uh yeah try try out things but also improve over the process. Um so yeah I I completely agree. I completely agree. So I share the sentiment. I mean I think we've learned and I think we can all agree here that uh full transparency doesn't equate to accountability. So we have to basically move forward using the best of both worlds and especially while we're still kind of in the web 2.5 era and I'm coming really from the institutional and and you know traditional enterprise standpoint. Um yeah, so it's kind of transition, keep

up with the trends, keep evolving, keep adjusting towards hopefully a fully decentralized one day web 3 uh environment um where everything will be run. Thank you. And yeah, we are almost out of time. So I would like to let you with a thought or a statement by um Elena Ostrom. She's a very known well-known researcher and got a Nobel Prize in economic science in for her work um on governance back in the 90s and um she encountered and she stated or her statement is there is no panacea in governance and um yeah 35 years later we

can see exactly it's still there is no oneizefits-all model. So when it comes to governance and this is equally applicable also to privacy. Therefore I think that's exactly what we've seen here today. We have different architectures. We have also different assumptions and also different tradeoffs. So um means we are not moving into a unified model towards a unified model. It's more that um yeah towards a deeper understanding that different circumstances need also different kind of approaches and actions and uh yeah with that I've come to the

end. Thank you to your to our brilliant panel and thanks to you guys and yeah wish you a wonderful evening. Thanks. Let's give it up for our panelists. Thank you.