On the Definition of Decentralization: Berkeley Blockchain Xcelerator Perspective Discussion
Host: Ashish Rajendra Sai @ Lero - Irish Software Research Centre & Horizon Globex, Jocelyn Weber @Berkeley Blockchain Xcelerator.
Abstract: Decentralization is one of the most loosely defined terms in the blockchain realm. The founders of some of Berkeley Blockchain Xcelerator’s distinguished start-ups and alumni share their perspectives on the debate surrounding the semantics of decentralization in the blockchain world.
Josu San Martin, Co-founder @ Sixtant
Bette Chen, Co-Founder @ Acala
Liam Lin, CEO @ Nickel
Sota Watanabe, CEO@Stake Technologies
● Sixtant (BBX Cohort 2020): A Mexico-based high-frequency trading system that provides market making and liquidity services in crypto.
● Acala (BBX Cohort 2020): A decentralised stablecoin and staking liquidity platform powering cross-blockchain open finance applications. We’re uniquely set up as a decentralized finance consortium with a vision to build cross-chain defi infrastructure, to provide stability and liquidity for the Polkadot ecosystem.
● Nickel (BBX Cohort 2019): Nickel is an enterprise permissioned network for financial services built on top of Hyperledger Fabric and Hedera Consensus Service with Digital Assets Modeling Language (DAML) smart contracts. In 2020, they launched their first service for fraud prevention in emerging markets FX trades, along with trade capture and confirmations on the distributed ledger.
● Stake Technologies (BBX Cohort 2020): Highly scalable infrastructure for Web3.0 via their new Plasm Network - a Polkadot Parachain and Dapps platform, enabling smart contract functionality
The introduction of Facebook’s Libra in 2018 reignited the discussion surrounding the meaning of “Decentralization” in public blockchain context. The official Libra documentation describes itself as a decentralized cryptocurrency. However, many have raised doubts surrounding the decentralization claims of Libra, most notably co-founder of Ethereum, Joe Lubin, described Libra as “A centralized wolf in a decentralized sheep’s clothing”.
There seems to be a myriad of definitions, conceptualizations, and dimensions used to describe this concept. Vitalik Buterin of Ethereum discussed the confusion surrounding the meaning of decentralization in his blog post titled “The Meaning of Decentralization”. In his description of decentralization, he segmented the definition into three components - Architectural (de)centralization, Political (de)centralization and Logical (de)centralization. This categorization has since been extended to include over 6 different components by Balaji S. Srinivasan (CTO of Coinbase).
However, little has changed since the fine-tuning of the definitions by Vitalik and Balaji in regard to the use of the term “decentralization”. Josu San Martin (Co-founder @ Sixtant - Berkeley Blockchain Xcelerator Startup) discusses the issue of using decentralization as a proxy term for permissionless systems in his blog post. He also talks about how decentralization may have a different meaning in crypto-economics than the pure peer-to-peer networking nomenclature. Tony Sheng and Ben Sparango of multicoin have argued against the use of terms such as “Centralized” and “Decentralized” when describing crypto financial instruments. They suggest using a yet-another fine-tuned multi-dimensional model to describe trust in blockchains.
We would like to take this opportunity to capture the opinions of our Xcelerator community on the debate surrounding the semantics of decentralization.
JM: Josu Martin Co-founder @ Sixtant
BC: Bette Chen, Acala Network.
LL: Liam Lin, CEO@Nickel exchange
SW: Sota Watanabe, CEO@Stake Technologies
AS: How would you define decentralization in blockchains? Do we need multidimensional models such as the one proposed by Vitalik, Balaji, and Tony to describe decentralization? Would you prefer it over a more generic definition of decentralization?
JM: The differentiation between centralized and decentralized systems is enough when talking at a very high level; for example, the banking system uses a centralized model while Bitcoin uses a decentralized model. However, once we start taking a deeper look, we realize that those concepts are not nuanced enough. Let us take a look at the concept of “centralized” by using a bank as an example. A bank is a centralized institution, where a small number of individuals have an enormous power over their clientes. But a bank can also be a public company, and in that case the decisions are taken, or sanctioned, by thousands of shareholders. Banks are also forced to follow all the rules established by various regulators, even by international institutions such as the Bank of International Settlements, which takes the power away from the directors and decentralized it in international committees, which in turn also seek the public opinion.
At this point, it may become apparent that the distinction between centralized and decentralized is more a scale of grays rather than a dichotomy. Going back to blockchains and taking this thesis further, we can observe that it is not the same to decentralize the decision making process with a governance model like Decred, or to decentralize validation with easy to run full nodes like Bitcoin.
This is the reason why using multidimensional models is the right way to approach the analysis of the properties of blockchains. The shift from “centralization” concept to a “trust” spectrum as Tony Sheng and Ben Sparango propose in the article, also becomes very useful. An idea that Nick Szabo had already explored back in 2014 in his article The dawn of trustworthy computing, where he argued that the notion of “‘trustless’ is exaggerated shorthand for the more accurate mouthful ‘trust-minimized’”, hinting at the aforementioned spectrum.
BC: The real goal may not be decentralization, but to build better infrastructure for the digital economy, where there is “more truth and less trust” (a phrase coined by Ethereum Co-Founder and Polkadot Founder Dr. Gavin Wood). Trust may generally seem like a positive term, but often it is a necessary evil that forces us to place our affairs in the hands of third parties. Over-reliance on it opens us up to abuse and risk. Web3 is a concept to use technology to empower us moving towards “truth” - where no 3rd party is required to arbitrate when information is cryptographically verifiable. While decentralization is a critical component in this quest, the reality is never binary. Often “is the system decentralized?” is the wrong question to ask, rather a better question is “what are the risks of trusting this system?”  This can be better understood as a multi-dimensional trust spectrum for the underlying ledger as well as applications (protocols included) built on top, wherever trust is still required, no less effort, capital, regulation, and other protections are required to counter the risks involved that what the current (so-called “centralized”) world is doing today - and that’s the price we pay for “trust”. These trust properties are worth considering when answering that “better” question: censorship resistance, platform security, immutability, custody, verifiable security, legal and regulatory protection, risk hedging schemes e.g. insurance .
For example, Ethereum as a ledger would rate quite high in censorship-resistance, platform security and immutability, while there will always be risks trusting someone or some system, I would argue at this point people do trust information recorded on Ethereum. While smart contracts deployed on Ethereum would inherit these platform-level trust properties, it doesn’t mean they are inherently trustworthy. Some applications are not so immutable where an admin key exists and an operator can change many aspects of the system. This is not good or evil judgment, but a more clinical and sensible approach to evaluating ‘trust’ and the risk of trusting.
 Why Decentralization Isn’t as Important as You Think by Haseeb Qureshi https://unchainedpodcast.com/why-decentralization-isnt-as-important-as-you-think/
 Multicoin Capital has a similar perspective in looking at this problem and produces a good analysis here https://multicoin.capital/2020/03/24/trust-spectrum/
LL: Decentralization to me refers to leaderless systems in which all members have equal rights and functions. There is no single point of failure and members can enter and exit the system without causing disruption or degradation in overall performance. Note that I didn’t explicitly mention freedom of entry and exits as seen in permissionless systems. A key test is if the foundation/company working on the decentralized system ceases to exist, can the system continue to run and can someone else pick up where it was left off to continue development? There are different categories of decentralization as mentioned by Vitalik et al. There should be a base framework that covers the multi-dimensional model against which decentralized systems are evaluated/scored for more clarity on which aspect of the system is decentralized and how egalitarian a system is.
SW: I guess the definition of decentralization heavily depends on one’s nationality. As a Japanese, I would say decentralization and distribution are different. The former means the situation which is not depending on centralized parties or institutions but it allows a certain point of centralization. The latter doesn’t allow any centralized authorities. In this sense, decentralization is not binary. Yes, I believe multidimensional modes are better than a generic definition because decentralization is not binary and every person has every definition of decentralization. However, we need more discussion on the variables and factors to define the term.
AS: What is your opinion on the decentralization aspects of Facebooks’ Libra?
LL: Facebook’s Libra share many similarities to Hedera Hashgraph, in particular - a governing council comprising of industry leaders and both projects aim to transition from a permissioned system to a public network in due time. Hedera has been widely acclaimed as the next-gen DLT while Libra is viewed with suspicion by everyone, from the blockchain community to the regulators. Libra’s problem is its association with Facebook and with the social media omnipresence mired in numerous personal data scandals, it is no surprise that no-one trusts Mark Zuckerberg to handle their money, not to mention, be the world’s central banker.
AS: Do you agree with the suggestion of Tony and Ben to move away from the binary notion of decentralization?
BC: It is a great start towards truly understanding the problem domain of the decentralization discussions. The term ‘decentralization’ is more like a ‘solution’ than a problem definition. Once we have a holistic view of the problem domain - trust, virtues e.g. fairness and openness and risks that we want to encode into or eliminate from the new systems (e.g. blockchains), then we can take on a more sober approach technically and governance-wise.
SW: Agreed. Decentralization is not just a term but a movement for us. To attract and excite more people, the term should have room for flexible interpretation so that each person is able to consolidate ideas on their own.
AS: Is a truly decentralized exchange platform (to fiat and other cryptocurrencies) possible?
JM: As we stated before, the concept of a fully decentralized exchange platform is even hard to imagine, there will always be some aspects that could be regarded as centralized: the blockchain it is running on, the webpage used to access it, the number of developers, the oracles…
So we would shift the goal post a bit, and try to shoot for a trust-minimized exchange, in which we can take the trust-minimization as far as possible without degrading user experience. This last point is very important, because things like TOR increase trust-minimization but also present a large obstacle for average users and mass adoption.
Framed it in that way, we already have some trust-minimized exchanges, which do present many qualities of decentralization: bisq for fiat, and Uniswap for crypto, are great examples. They still have many limitations, and they struggle to attract liquidity outside of a few markets, but their use is growing, and we are already seeing very promising iterations of the product that try to overcome these issues.
LL: Yes, for cryptocurrencies but not for fiat currencies. Fiat is widely accepted as money everywhere in the world, whereas cryptocurrencies are treated almost like an alternative investment. A decentralized exchange in its purest form involves permissionless transactions which means no Know-your-customer compliance checks. Dealing with fiat opens up the exchange to the risk of money-laundering and funding a whole slew of criminal activities like human, arms, and drug trafficking. Not to mention terrorist organizations and pariah states like North Korea looking to sidestep sanctions. When cryptocurrencies gets adopted mainstream, you might see a change in policy towards permissionless exchanges to curtail illicit activities.
AS: How does the notion of decentralization impact your startup? If you are working on a public blockchain, how significant is the decentralization component for your value proposition?
JM: Due to the nature of our firm, most of our assets are always deposited on centralized exchanges, so we take centralization as a given. That being said, we are always exploring the latest developments in the space of less centralized models, non-custodial exchanges and DEXs, among others.
However, what we have seen so far is that most of the liquidity concentrates in centralized exchanges. There are some very promising projects suchs as the Kyber Network, Uniswap, the 0x ecosystem, the Bancor Network, or more recently Balancer Labs, to name a few, that are trying different approaches to take some of that volume away from centralized exchanges.
Regarding the centralization of the blockchains of the tokens that we use to trade, we take a similar approach to the trading venues we use. The possibility of an exchange being attacked and losing user funds is our main operational risk that we are always evaluating and trying to mitigate. With blockchains the problem is similar, what is the likelihood that centralization will lead to a destruction of value, be it a deep chain reorg, such as the one that happened to Ethereum Classic or Bitcoin Gold; a hostile takeover like the one currently happening on Steemit; or any other risk arising from centralization.
BC: Highly important. We have the Web3 ethos imprinted in our values - “Can’t Be Evil” (versus Web2.0 Don’t Be Evil). We are building a stablecoin for Polkadot’s network of blockchains, and decentralization is in every aspect of our work. Since inception Sep 2019, we have intentionally chosen a progressive decentralization pathway towards our ultimate goal - making ourselves redundant while the network is thriving. A16z summarizes quite well what we are doing at the organization and technical level:
Phase #1: Initial start-up: no pretense of decentralization, competency trumps. The research and development of the Acala Network are led by the Acala Foundation, which is set up as a decentralized consortium with two founding members - Laminar and Polkawallet.
Phase #2: Community participation: significant portion (>40%) of the network token (ACA) are reserved for participant incentives and funding for community development which are encoded and released on-chain. Acala has on-chain governance to manage treasury, protocol, and technical upgrades. It can evolve from a PoA governance to elected council and referendum to progressively giving more power to the community as it matures.
Phase #3: Sufficient decentralization: according to Acala’s economic model encoded on-chain, at this stage ownership of the network would be in the hands of the ACA holders, and founders would lose majority control. Acala would be self-sufficient in terms of community governance, funding (through a trustless wealth fund owned by the network), and development.
LL: We built a decentralized exchange for institutional FX trading for emerging markets. We chose Hyperledger Fabric’s permissioned blockchain as my clients include banks and corporates and they did not want their transactions to be stored in public networks. Know[ing]-your-customer (KYC) was also a key requirement for operating in a highly regulated currency regime. One tradeoff that we made was to sacrifice decentralization for speed and high throughput. Our initial design involved onchain order books and smart contracts for order execution but it was taking too long compared to a central limit order book setup. We eventually settled on a hybrid design that involves a centralized trading engine with onchain order books to retain immutability of the order sequence. We also switched our consensus algorithm from a Kafka cluster to Hedera Consensus Service in order to achieve trustless consensus from a public network and the privacy of a permissioned network.
SW: We are making a public blockchain itself called Plasm Network. It is a potential Polkadot Parachan and a scalable DApps platform with layer2 technologies such as Plasma and Lighting Network. Decentralization matters for us because our goal is to re-democratize the Web and society by making Web3.0. Plasm Network is the scalable foundation for it. The most important component of Web3.0 is sustainability. History has shown that power inevitably collapses. Ideally, we would like to make a system running without us, a centralized single point of failure or even people. But, from the startup view, decentralization doesn’t matter at the early stage because it can hinder rapid implementations or drastic changes that are needed to make the product more beneficial for the community. Decentralization is not a means but a purpose. Regarding Plasm Network, it can be centralized at the first point and it will be decentralized because we have a proper governance model and a method to distribute token fairly. These features must be the value propositions.