“The machine does not isolate man from the great problems of nature but plunges him more deeply into them” – Antoine de Saint-Exupery

Technology and philosophy do not seem to have much in common. One talks about machines and code, while the other delves into the essence of human thought and existence. How can machines influence our thinking? How can code answer the existential questions of human life? Historically, technology has developed independently of larger philosophical inquiries – until Satoshi Nakamoto dropped the Bitcoin whitepaper in 2008. 

With its proposition of peer-to-peer transfers of electronic cash, the paper created a movement that reimagined how humans interacted with money. The movement did not stop at that. It led to the creation of an industry, web3, that has enshrined privacy, digital sovereignty, permissionless innovation, and user ownership into code. It ensures that these concepts are not luxuries but fundamental rights of the users. Social experiments with pop-up cities such as Zuzalu exemplify this. 

This intersection of blockchain and philosophy has guided us at Advaita Labs since our inception. We have been particularly focused on the philosophy of intersubjectivity – popularized recently by EigenLayer’s latest whitepaper – and its applications in distributed systems since our beginning, through our work on causality graphs. This work has led to the creation of the Hetu Protocol.

The Hetu protocol is based on the foundations of intersubjectivity and verifiable causality, a combination of which allows us to create an attribution layer enabling trust-minimized diverse incentive mechanisms for future microservices. 

In this blog, we will explore one of the founding pillars of Hetu, intersubjectivity, and how it can help create diverse economic incentives in distributed systems. We will then explore how intersubjectivity helps minimize trust in heterogeneous distributed systems, and finally look at some drawbacks in current distributed systems that make intersubjective consensus mechanisms necessary. 

“I am all that I see, I am an intersubjective field, not despite my body and historical situation, but, on the contrary, by being this body and this situation, and through them, all the rest." – Edmund Husserl

Blockchain technology is based on the idea of objectivity. If any data is stored on-chain, it is immutable, attributable, verifiable, and hence becomes a truth that cannot be changed. Much like the scientific fact of gravity. This objectivity ensures transparency and trust.

Subjectivity, on the other hand, is a version of reality that is mediated by the experiences of a single entity. For example, one may think that ETH will be more valuable in the long run when compared with BTC. This assumption is based on personal beliefs and is therefore, a subjective truth.  

Now, let’s understand intersubjectivity. A term first coined by the Austrian-German philosopher Edmund Husserl, it refers to a shared understanding between entities developed through their interactions with each other. Husserl posited that any experience of the self is always mediated through many factors – the body, historical life situations, relations with the other constituents of the world. Intersubjectivity thus forms a mutual understanding essential for human experience, extending beyond Descarters’ “I think, therefore, I am” to “I exist in relation to others and all my experiences are based on a shared understanding with the world”.  

Money is a prime example of an intersubjective agreement. Society collectively agrees to attribute value to coins, paper currencies or digital entries in banks and calls it money. The function of money is a social construct and is based on the fact that people trust it. 

The philosopher Jurgen Habermas highlighted the role of intersubjectivity in communication, arguing that any social norms are established through discourse and discussion. This process of communicative action is inherently intersubjective, as it relies on the participants' ability to recognize and validate each other's perspectives. 

Many Eastern philosophies, particularly Buddhism, also reflect intersubjectivity. The Buddhist concept of dependent origination believes that all phenomena arise in dependence upon a multitude of causes and conditions. Take for example a person named Jane. Now, there is no singular self of Jane. It is not her body, her mind, her relationships, her thoughts, or her internal organs alone that can make up Jane. It is all these diverse elements and many more that harmoniously come together to form Jane. Buddhism then emphasizes the interconnectedness of all beings and challenges the notion of an independent, permanent self.

Now that we understand intersubjectivity, let us look at how this philosophy can help create fairer incentive mechanisms for users of decentralized systems. 

“The first law of tokenomics: don’t get your advice from people who use the word tokenomics” – Vitalik Buterin

Current consensus mechanisms work on establishing data ownership through tokens. For example, Bitcoin’s Proof-of-Work consensus mechanism rewards miners, in the form of BTC tokens and transaction fees, for being the first one to solve a complex cryptographic problem. The incentive structure focuses on the ownership of the infrastructure required for computation and the rewards for validation transactions. The rewards of the miners are directly proportional to the might and computational power of their mining infrastructure. It creates a static and linear economic model.

However, the real economy requires diverse incentive models to cater to the needs and motivations of a diverse, pluralistic user group. In this real economy, we do not need to address data ownership but the right to use that data. This is where intersubjective consensus mechanisms can help. 

By shifting the focus on data usage, intersubjective consensus mechanisms reward diverse contributions, fostering an interactive economy. Let’s illustrate this through a decentralized AI platform where all users contribute in diverse ways. While one user provides a high-quality dataset to train the AI model, the other contributes by providing processing power to run simulators. Even if users are performing the same activity, say contributing data sets to train the AI, they can be contributed based on the quality of the data, as well as the frequency and accuracy of their contributions. In this system, payment methods will also become more diverse, say tokens for data quality and staking rewards for maintaining algorithm integrity. Intersubjective consensus allows for focus to be on supporting services and the interactions between them, rather than solely around ownership. 

Diverse incentive mechanisms support more complex economic models, encouraging robust decentralized systems. By rewarding data usage, intersubjective consensus mechanisms promote pluralistic and diverse economic interactions.

The application of intersubjectivity extends beyond just incentive mechanisms in heterogeneous distributed systems, of which decentralized networks are a part. Let’s explore some more application of intersubjectivity in the following sections.

“A trust-minimized solution does not completely remove the need for trust but reduces it to the point where it’s almost negligible.” –  Vitalik Buterin

Heterogeneous distributed systems are systems that comprise diverse components such as hardware, software, data structures, and protocols while working towards a unified goal. Each component in these systems is designed to perform their specialized tasks optimally, increasing the efficiency of the whole system. This division in work also allows the system to be fault-resistant as even if one component fails, the rest of the system continues to function. 

A great example of a heterogeneous distributed system is interoperability between blockchains. All blockchains such as Ethereum, Solana, and Polkadot have different protocols, consensus mechanisms, smart contract languages, and applications. However, interoperability solutions such as bridges, cross-chain protocols and oracles allow all these diverse components to communicate with each other and reach a common goal, for example, bridging from one blockchain to the other. 

In this context, intersubjectivity provides the connectivity and commonality that helps these diverse, pluralistic systems achieve a common goal. Distributed systems depend on participants for inputs, relying on the shared understanding and mutual comprehension between these participants. Here are some ways that users in a distributed system participate in intersubjectivity:

• Governance Models: Decentralized governance models, such as decentralized autonomous organizations (DAO) rely on intersubjectivity for their functioning and organizational decision-making. DAO participants vote on the rules of the organization and future developments through voting mechanisms that rely on intersubjective agreements. 

• Decentralized Identity and Reputation: In truly distributed and decentralized systems, identity management and reputation systems rely on intersubjective validation rather than depending on centralized authorities. 

• Intersubjective Forking: In the latest application of intersubjectivity, EigenLayer’s proposal for intersubjective forking allows for a blockchain to be forked based on intersubjective agreements when on-chain data cannot determine the cause of malicious behaviour in a blockchain. 

Intersubjectivity fosters plurality in distributed systems, not just in terms of the system components but also in terms of the system’s culture. Vitalik Buterin recently highlighted this in his post, where he explored how a blockchain’s culture can be as powerful as economic incentives in attracting users. 

“Culture…affects what kinds of actions people are motivated to do, and what kinds of actions people can do. It affects what is considered legitimate - both in protocol design, and at the ecosystem and application layer.”

By bringing us closer to embracing pluralism, intersubjectivity can also help create a harmonious relationship between human intelligence and artificial intelligence (AI). In a world that is fast-embracing AI in distributed heterogeneous systems, intersubjectivity acknowledges the differences between the two intelligences while also recognizing the similarities. Both forms of intelligence possess the ability to learn through observation and experience, and to process that learning to derive conclusions. The entire process of learning to conclusions is causal and interactive. Recognizing these commonalities allows us to build a foundation for mutual understanding and respect. 

The important point to note is that intelligence, whether human or artificial, relies on interaction and causality. And causality is indifferent to the form of intelligence, focusing solely on the flow of observations made, experiences and interactions had and influence created. Hetu, with its emphasis on causality, aims to be at the forefront of building a better relationship between different intelligences. Just as Ethereum bridges digital and real-world objects, Hetu connects different data structures and intelligences, both human and machine. Through Hetu, we aim to build an attribution layer that connects diverse elements of heterogeneous systems into a cohesive, harmonized energy state.

We now understand intersubjectivity from its philosophical roots as well as how it applies to distributed systems. Now let’s turn our attention to why we need intersubjectivity in distributed systems. 

Distributed systems today face challenges such as oracle discrepancies, transaction ordering issues, and byzantine faults. Intersubjective consensus mechanisms, which emphasize shared understanding and plurality, offer a robust approach to these problems.

1. Oracle Discrepancies

   Oracles are data providers that supply external, off-chain data such as asset prices, weather information, government IDs to a blockchain. Since oracles rely on bringing off-chain data on-chain, it can lead to inconsistencies, biases, and errors. Here, objectivity fails to work since for most of these types of data, there is no simple way (or in some cases it may be totally impossible) to verify their correctness by looking at the data alone. Intersubjective consensus mechanisms can address this problem by using data from multiple independent oracles, ensuring no single oracle unduly influences the system. To reach a single, trusted data point from multiple oracles, these consensus mechanisms can determine their own rules and leverage the data, say by using a weighted average or a median of all oracle data. 

   Hetu offers a way to generate verifiable proofs for this type of intersubjective consensus. The Hetu causality graph tracks the oracles that participated in the intersubjective consensus process, the source from which they obtained the data, and the way the final consensus result was combined. Any third party can independently verify the final data based on the number of oracles, their trustworthiness, and their source of data (and their causal precedences).

2. Transaction Ordering

   Transaction ordering is crucial in blockchains, with any discrepancies in the correct order of transactions changing the outcome and the subsequent state of the blockchain. One example of the manipulation of transaction ordering is front-running attacks, where attackers exploit transaction order for profit. This undermines the fairness and trust in a blockchain. 

   Unlike properties such as unforgeability and anti-double spending, fair ordering of transactions is hard to determine objectively. Intersubjective solutions like Hetu’s verifiable causality graphs provide mechanisms and semantics to reason about fair ordering of transactions, thereby enhancing trust. Instead of using unreliable sources such as physical time of transaction generation, Hetu provides causal information for each transaction, such as the number and identities of peers in the transaction dissemination graph, causal dependencies with respect to other transaction invocation and commitment. Intersubjective consensus policies can define a fair order of transactions using these reliable, verifiable, and decentralized causal information.

3. Byzantine Faults

   Byzantine faults refer to arbitrary behaviors by malicious participants that do not exactly follow the specified distributed protocols. A major type of Byzantine fault is equivocation, in which an adversary presents different views or information to different parties of the network. Now, this creates the issue that the distributed system cannot reach a consensus on whether to completely lock out the faulty server or to still keep taking information from it. A fundamental issue here is that in many cases, it is impossible to correctly identify that a node is adversary and actively attempting to equivocate, or this divergence in views was caused by other malicious nodes. In such a case, intersubjective consensus can ensure that distributed systems are resilient in the face of byzantine faults.

   The Hetu intersubjective consensus mechanism facilitates distributed systems to achieve consensus even if some participants in the system are behaving maliciously. Every event and information disseminated in the Hetu network is tagged with verifiable causality information. When presented with diverging views, honest participants can correctly identify sources of equivocation or malicious behaviors based on causal reasoning. Causality graphs also offer richer sets of semantics to define agreement rules, beyond the simple majority reasoning in traditional objective consensus. By encouraging diversity and redundancy in the system, intersubjective consensus mechanisms ensure that no node becomes a single point of failure and undermines the integrity of the network.

4. Consistency in Heterogeneous Computing Paradigms 

   The foundation of traditional blockchains is a single, global order of transactions that is achieved via objective consensus. Reaching global agreement on a total order, however, presents numerous issues. First, security guarantees of this global total order limits the overall system scalability. Second, the system becomes unavailable under poor or adversarial network conditions. This is a fundamental limitation due to the famous CAP theorem.

   The totally ordered ledger model fits well with financial applications. However, there exists a rich set of computing paradigms outside financial applications where the scalability and unavailability drawbacks of object consensus are detrimental. Some examples include AI, social, gaming, collaborative editing, and payment systems. Common themes across these application scenarios are high interactivity, high traffic volume, and demand for real-time responses. All are hard to achieve with current blockchain technology.

   Hetu’s intersubjectivity resonates well with these heterogeneous computing paradigms. Unlike financial applications, these systems typically do not require global total order of all events. Instead, they are resilient to weaker models of consistency, such as causal consistency and eventual consistency. Hetu’s causality graphs and logical clocks enable systems to have diverging distributed states. The divergence can be temporary and eventually becomes consistent, or it can be partial where only causally-related events are required to be consistent. Such more flexible consistency requirements allow systems to cache and access data locally, and continue to be operational even under long network partitions. Hetu’s intersubjectivity thus provides better performance, scalability, and availability to a diverse range of applications.

We now know that intersubjectivity is essential to generating consensus in decentralized systems. It is important to understand the diverse, pluralistic parts of an ecosystem, but to also ensure that there is unity in that diversity. Achieving harmony and integrity in the pluralistic parts of an ecosystem helps take a holistic approach to growth. And intersubjectivity facilitates that by minimizing the trust between heterogeneous distributed systems. It helps create a democratic framework for collaborative systems that reflect the diversity of their users. 

This is just the beginning. In the next part of our series, we will dive deeper into the future thought frameworks for intersubjective consensus and explore the other pillar of Hetu’s foundation, causality graphs. Stay tuned as we continue this journey towards a unified and harmonious web3.

If you've read till the end, congratulations! We hope we have helped you understand intersubjectivity in distributed heterogeneous systems. Follow us on Twitter and Farcaster for more relevant content from Hetu. Join our Telegram group to be a part of our growing community!