The text below attempts to dissect the phenomenon of interpersonal trust: what we mean by “trust” today, how we arrived here, and how it has evolved in the digital era. If you’re only interested in crypto protocols building their products around this term, feel free to jump to the last section. Additionally, every major step on the path to an understanding of trust will be accompanied by a picture.
We will discuss:
simplified evolution of interpersonal trust irl: early societies, pre-internet states, digital era
attempts to establish trust in digital environments: trust networks, credit networks (including many useful links to academic research)
novel approaches to trust modeling in digital environments: a brief review of the new cohort of web3 protocols
this paper is a part of research done jointly with R Alter, D Mukh, and M Volkh
As Wikipedia suggests, “trust is the belief that another person will do what is expected. It brings with it a willingness for one party (the trustor) to become vulnerable to another party (the trustee), on the presumption that the trustee will act in ways that benefit the trustor. In addition, the trustor does not have control over the actions of the trustee”. Some say that trust “is often viewed as the glue that holds society together”, and it is hard to argue against it. Indeed, trust is something we encounter on a daily basis throughout our lives. It is explicitly observable in direct interpersonal relationships, but it also forms the intangible basis for numerous systems (political, economical, and generally societal) that we interact with, while being hidden under layers of abstractions. Trust relations are so deeply integrated in our society that one often cannot simply identify them. Therefore, before we attempt to understand the nature of trust in digital realms and move to the truly astonishing “trust-related things”, let’s skim through the simplified history of the evolution of trust and try to point out some of the locations of trust in the fabric of human society. At this point it is important to articulate the key properties of the phenomenon we call “trust”, rehashing the definition above:
it is a person’s attitude towards another person — an expectation that the trustee will act in a specific way;
the person is uncertain about the outcome of the trustee's actions;
trust does not imply control over the trustee's actions.
It is also important to clarify that we will not talk about trust towards things or trust in ourselves – interpersonal trust (directed towards a specific human being — primarily from another human being, but sometimes from the collective of other human beings) will be at the core of our discussion. However, since different types of trust often overlap (e.g., I trust Alex because I trust the system he’s part of), I will occasionally discuss other types of trust as well. Additionally, the concept of generalized trust (which is more about one’s state of mind rather than interpersonal attitudes) is outside the scope.
By “early societies” I mean collectives of gatherers, hunters, and/or barbarians that existed before the emergence of states, including those of the ancient world. As far as we know from history and anthropology, these societies were often isolated from each other, sometimes fighting among themselves, but more often with nature. They could be characterized by primitive, pre-money economies, low levels of division of labor, a high level of interpersonal trust, and strong communal ties. These communities relied heavily on shared resources and mutual aid to survive. Leadership structures were often informal and based on kinship or merit rather than formalized hierarchies. Social norms and customs were typically transmitted orally through generations, helping to maintain cohesion and continuity within the group.
In the table below I mention some of the activities typical for these kinds of communes and the role of trust in them. Since the goal here is not to give an extensive historical perspective on early societies but to identify the role of interpersonal trust within them, I intentionally modernize, simplify, and let go of some details.
We could elaborate further, but the bottom line here is that trust-requiring interactions in such societies were based on past interactions and/or assessments done by trusted third parties (family, known members of the community, etc). Not a very interesting table, to be honest. At this point, it is difficult to find scenarios where one has to trust someone without having visible direct or indirect (through trusted third parties) connections (or “trust links”) with the trustee. Again, the whole picture is very simplified, but the core idea here is that trust is rarely directed toward strangers — no system capable of providing a basis for trust in strangers exists (yet).
With the advent of banks and the centralization of power and culture, we are slowly but steadily approaching the modern state of affairs, which shifts its focus away from private and direct trust links.
By “states” in this paragraph, I mean societies (both ancient and modern, post-industrial revolution) characterized by the development of centralized governance, legal systems, and administrative structures, including 'proto-states' such as city-states or Greek poleis. The distinct feature of interpersonal relationships within states is the higher frequency of interactions among strangers compared to early societies. Acquiring goods or services, storing money in the bank, or hiring an employee — all of these activities imply some level of trust. However, in the absence of past personal or third-party experience, the logic of early societies doesn’t work anymore. Thus, other sources of trust (or, at least, instruments to reinforce the personal assessment of a potential trustee) arise. This was the period for the first cohort of “trust systems” to flourish.
At a very high level, we can describe them as systems that replace the need to trust a stranger with trust in the system or provide us with extra information to form a personal assessment of a potential trustee. Thus, the system must take something as an input to assess a potential trustee and produce something as an output. Some notable examples of such systems include legal systems, regulatory bodies, and certification processes. But to be more precise, let’s draw a new table — this one will be a little bit more complicated.
There are some interesting conclusions we can draw from this table:
In pre-internet states, trust tends to concentrate around trust systems. More and more often, trust is directed toward a system, and interpersonal trust (toward a specific human being) emanates from the trust in the system (of state’s rules, self-regulation, religious institutions, guilds and other non-state entities). Though the system itself still needs to assess the trustworthiness of this human being somehow — in some sense, we can say that we delegate the process of the trustee’s assessment to the system.
Trust systems can play different roles. Some of them can reinforce our personal assessment of a potential trustee (legal contract with an employee), others can relieve us from making this assessment (usually it is connected to society's well-being — for example, uncertified or unlicensed medicines merely can’t enter the market). When we are talking about the collective assessment of trustees, the goal of the trust system is to provide a framework for this process.
Trust systems are usually disconnected from each other – they operate each under their own set of rules and pursue their own specific goals. In other words, no system can provide a generalized framework for trust assessment, potentially adaptable across different domains.
An increase in contextual information relevant for forming a personal assessment. As society becomes more complex, one can form some level of “basic trust” towards a potential trustee from the general understanding of how the world works and indirect clues in the form of reputation — for example, being members of the same community (soccer fans or university students) or the process called the “shadow of the future”.
The growing role of the state. In the absence of an objective opportunity to rely on personal experience or reviews from friends, states and their norms start to play a huge role in trust assessment. By “state” here I mean not only directly the rules of law but also their “derivatives” — banking regulations, mandatory certification (even if it is carried out by the community), etc. Systems provided by private entities and international organizations also start to play significant role in trust assessment (e.g., ISO certifications, financial credit ratings).
Below, you can see a visualization of the trust flow in modern society for the "hiring a lawyer" case.
At the same time, we can find examples of systems trying to grasp and manipulate a more generalized notion of interpersonal trust as “the belief that another person will do what is expected”. These include the concepts of mutual credit systems and their close relatives: LETS (local exchange trading systems) and local currencies, CES (community exchange systems) and community currencies, time banks and time-based currencies. The most prominent example is Sardex (which is post-internet phenomenon, but since it’s a “child” of the aforementioned concepts, let it be in this section), which could be thought of as a mutual credit system, taking as input promises to provide goods or services from various parties and issuing in exchange for them a local Euro-denominated complementary currency circulating within Sardex members in cycles or loops. While the goal of the Sardex system is focused on facilitating local economic development and resilience, the underlying logic represents a trust system, providing the ability to transfer and net mutual trust-based commitments towards other members of Sardex.
What truly distinguishes Sardex from common trust systems in the pre-internet states is the role of interpersonal (or “intercompany,” since we are talking about B2B relations here) trust. As explicitly stated in Sardex original paper, “Trust is a fundamental feature of Sardex. It is community-specific and it relies on values and reciprocal expectations... Trust works at two levels: as a starting point to join the network and towards the Sardex organization (direct), and as a transactions’ lubricant between members (indirect).” As we will see later, this idea of utilizing interpersonal trust (local trust links) for the creation of a generalized trust system (in contrast to the state’s vague sparse trust systems, where local trust links often are not taken into account at all) will find new life.
But at this point, let’s finish the description of trust systems in pre-internet states with the following question: do the state’s norms (and their “children and grandchildren” — banking regulations, municipal rules, and other imperative frameworks) form the best possible framework for assessing trustworthiness, or were they merely a good enough solution for a pre-internet era when coordination of people on a large scale was impossible without a centralized actor?
In the era of the internet, the table above is still relevant. States are still “the kings,” and their trust systems still play the dominant role. However, new kings and new systems are emerging.
Let’s present another, revised table. Since the pre-internet trust systems from the previous table (government licenses, community certificates) are still in place, I will not repeat myself.
What structural changes and patterns can we see from the table:
The constant emergence of new environments where trust is essential. In the pre-internet era, we learned to assess the trustworthiness of strangers based on the context and environment around us. In the digital world, we constantly find ourselves in new settings where we are encouraged to interact with strangers — new apps, online stores, course sellers, and so on. However, in the digital environment, there is no familiar and intuitive context for most of us, nor are there established trust systems.
Replacing opaque and subjective systems (like legal norms) with more objective algorithms. Today, not only symbols from the government and its institutions (certificates, licenses, etc.) can provide some level of trust, but also outputs by Google’s or other tech companies' algorithms (reviews in Google Play, alternative credit-score systems provided by fintech startups, Community notes in X, etc.).
A sharp increase in domains with available trust sources. Now, before entering into any interpersonal interaction where previously we would have faced a lack of information to formulate a personal assessment of a potential trustee, there is help from tech companies (reviews in Google Maps, ratings on specialized websites, etc.).
Back to interpersonal trust. Modern technologies have enabled more sophisticated trust systems that integrate both interpersonal and system-based trust mechanisms, such as online reviews and ratings. In the post-internet era, once again, we are paying more attention to other people's opinions rather than the opinions formed within state’s systems. The internet brings us the ability to grasp and “digitize” millions of individual inputs. In some sense, this is similar to early societies, where before entering a trust-requiring interaction with a stranger, we asked other people’s opinions. However, these people are mostly strangers too.
Diversity of trust systems. While in the pre-internet era, the collective assessment of trustworthiness was a privilege of complex and often vague trust systems (like state’s norms, certificate processes), new technologies enable the creation of novel systems with custom logic and algorithms for the trust assessment. Some of them (like review platforms) take as input the assessment of a potential trustee from hundreds or thousands of individuals, thus, again, in some sense bringing us back to old societies where only trustees’ assessments made by other humans mattered.
While these novel approaches to trust systems provide the opportunity to bring the output of the system (rating, top reviews, etc.) closer to the real trustee’s assessment made by other human beings (in contrast to pre-internet systems, where interpersonal trust and people’s opinions were often disregarded), we should note the potential for algorithmic bias and manipulation in modern trust systems. For example, online review platforms can be gamed, and algorithms can reflect and perpetuate existing biases.
One could ask, "Is it possible to abstract away from specific trust-requiring interactions and create a general framework for digitizing and transmitting trust”? In other words, could we construct a trust system without answering the question of "What is this system needed for”? If yes, what should this system take as input, and what is the best source of information about potential trustees? These are the questions that led to the creation of what we call “trust networks”.
A trust network is a term that encompasses any network where trust relationships are established, maintained, and propagated among members. These networks can be applied in various contexts, such as social networks, business partnerships, and online communities. The primary focus of a trust network is on the overall structure and dynamics of trust among its members. In such a network, trust relationships are built based on past interactions, reputation, and shared values or objectives. We can find first attempts to formalize the notion of trust in digital realms in research papers from the early 1990-x, leading to a wide range of various approaches to trust modeling and management in digital environments (check researches here and here).
Key concepts related to trust networks are:
Graph-based models. Trust networks are often modeled as a graph — a set of nodes and connecting edges, where nodes represent individuals or entities, and edges represent the trust relationships between them.
Trust propagation. Trust can be propagated through the network, meaning that if person A trusts person B and person B trusts person C, person A might have some level of trust in person C by extension.
Reputation systems. These systems often underlie trust networks, where the reputation of nodes (often derived from the history of node’s interactions) is tracked and used to influence trust decisions.
Decentralization. Trust networks can be centralized or decentralized, with decentralized networks often relying on distributed trust models.
Let’s put it in this way — trust networks provide the framework and mechanisms for trust to flow, be evaluated, and be leveraged within a community, much like how infrastructure supports the flow of goods or information in other types of networks. These systems can be seen as generalized trust systems, where the level of trust in any specific node is derived from the other nodes’ inputs (in the form of explicitly stated assessments or information about past interactions) regarding the trustworthiness of their closest neighbors.
Some prominent examples of trust networks include:
PageRank. While not being a trust network per se, it models a kind of implicit trust by interpreting hyperlinks as votes of confidence or relevance. Pages linked to by many others, especially by highly linked pages, are considered more "trustworthy" or important within the network.
Eigentrust. The system assigns a trust score to each participant based on the trustworthiness of their connections. The score considers not just who they trust, but also how trustworthy those connections are themselves (based on past interactions). It uses mathematical calculations to analyze these connections and arrive at a global trust vector for the network.
Web of Trust (WoT). Used in public-key cryptography like PGP (Pretty Good Privacy) and S/MIME (Secure/Multipurpose Internet Mail Extensions). Unlike Eigentrust's mathematical scoring, WoT relies on manual trust assignments. Users directly verify the identities of their contacts and sign their public keys, indicating trust. Trust spreads through connections as users sign each other's keys.
PeerTrust. PeerTrust is a proposed trust model for the P2P environment. It evaluates peer trustworthiness rating based on 5 elements: the feedback obtained from other peers about the target peer, the aggregated number of total transactions that the target peer has with others, the credibility factor of the recommender peers, the transaction context factor for discriminating the mission critical transaction from less or non-critical transactions, and the community context factor.
Other (mostly theoretical) proposed models for trust networks:
The parameters of these systems vary depending on whether it is a generalized trust network or a trust network for specific purposes (e.g., file sharing) or environments (peer-to-peer, multi-agent system, e-commerce, etc.). Additionally, the role of node reputation (history of node interactions) and initial node trust (pre-trusted peers) differs in these systems. Despite the various approaches to defining the rules of network functioning and node interactions within it, all these systems share a common fundamental task: to create a common trust system based on inputs from multiple actors/nodes regarding their degree of trust in their neighbors. The main output of this system is information about the collective assessment of trustworthiness for a node within this system.
It is noteworthy that many theoretical models of trust networks have not found application in real life. Moreover, in many cases, the authors did not even set out to determine the scope of their application, mentioning the applicability of their theoretical research for reputation and rating systems (in more modern papers — also in IoT and different types of wireless networks), or leaving questions about areas of practical application to future researchers.
At the same time, the concept of a credit network, which is a close relative of a trust network, was born. The credit network model was invented independently by (at least) four distinct groups of researchers, motivated by somewhat different issues and applications, but arriving at the same essential elements:
DeFigueiredo & Barr sought a reputation system with bounded loss from coalitions of malicious users;
Ghosh et al. aimed to support distributed payment and multi-user credit checking for multi-item auctions;
Karlan et al. wanted to construct an economic model of informal borrowing networks;
Mislove et al. were concerned with deterring spam.
A common thread in the objectives of these researchers was to capture a notion of pairwise trust, representable in quantified terms. In each case, the trust measure is grounded by interpreting the quantity as a capacity for transaction. That is, the degree of trust in one agent for another is measured by how much it is willing to expose itself to transactions with that counterpart. In other words, the model operationalizes trust as an extension of credit, in a framework where a credit balance entitles an agent to transact with the agent granting credit.
Credit networks can be seen as systems for trust modeling and management in digital environments, where trust is financialized in its most generalized form — in the form of a credit. It is important to note that “credit” in these systems can represent either a concept close to its traditional meaning (the transfer of money or other value with the trust that the borrower/trustee will return it in the future) or an IOU (I-Owe-You — the promise to provide some work, service, or any other value to the members of the network).
As a consequence, the primary goal of the first type of credit networks (check “Trust and social collateral” by D. Karlan) is to transform multiple nodes’ inputs into a score which represents the amount of value that could be entrusted to a node (with the expectation of returning the value later), while the connections between nodes are viewed as collateral (which likely would be lost in the case of a node’s inability to return the value) — this approach is similar in many ways to how informal borrowing works in real life. Indeed, we often lend/borrow money to/from our friends at the risk of damaging our good relations. Though, in real life, it usually happens between two specific persons, while in this type of credit network the level of trust is determined by the structure of the entire network.
In the second type of credit network (check “Liquidity in Credit Networks: A Little Trust Goes a Long Way” by P. Dandekar), credit (often described as a node’s personal currency) is merely a node’s unconditional promise to bring some value to anyone in the system, and other nodes’ inputs represent the level of trust in the trustee to provide this value. While these ideas can’t be mapped directly into our “real-life” relationships and traditional concepts of credit, they allow the network to serve as a decentralized payment infrastructure, where arbitrary payments can be routed through the network by passing IOUs between trusting nodes in their respective currencies.
The concept of credit networks was further elaborated by Rayan Fugger (Ripple’s cofounder) and other researchers: Pedro Moreno-Sanchez, Pranav Dandekar, Geoffrey Ramseyer, and others.
As with studies dedicated to trust networks, many of these studies did not aim to find practical applications for the formulated concepts, let alone search for product market fit. Unsurprisingly, most of them remained theoretical, though some core concepts of credit networks are used in the Lightning Network architecture, which can be thought of as a type of credit network, where the “credit limit” is the amount of Bitcoin that has been committed to each channel (essentially, “credit” or IOUs here are collateralized by BTC).
The first, and for a long time the only, attempt to implement a “pure” credit network (a credit network where IOUs are literally promises or expectations without any collateral) in practice was the first version of the Ripple network, traces of which (named “Community Credits”) can still be found on Ripple’s website. It was designed to enable people to extend credit to each other, forming a network of IOUs. In this system, users could create trustlines with each other, allowing for the transfer of value based on mutual trust. The idea was to facilitate transactions through a web of trust where people could pay each other through intermediaries who had established credit relationships.
But facing reality is much tougher than conceptualizing it in a paper — the initial Ripple ideas were abandoned in favor of serving banks’ capital. The villages.io website, providing access to the “Decentralized Barter Credits and Collaboration System” (essentially, the first version of RippleNet), is still alive. However, it is more of a historical artifact than a functioning credit network.
After Ripple abandoned the implementation of its original ideas, interest in credit networks (at least in creating projects based on such ideas) subsided for a while. However, today we are witnessing a return to and a reevaluation of the ideas behind credit networks. The underlying architecture still represents a trust system, where trust is manifested as credit flowing through the network, but the details have significantly changed. So, let’s take a look at the key projects developing this idea.
My strong belief is that viable long-term solutions for trust modeling in digital environments can only be based on blockchains with open and accessible data. Yes, there are a lot of custom solutions for trust management in digital domains, where the notion of trust is domain-specific. Just to name a few: credit risk analysis, spam/fraud detection systems, loyalty programs, reputation systems, social media content moderation, identity verification services, knowledge-sharing platforms, etc. And yes, there are plenty of novel, creative techniques used within these solutions. But you won’t see them in the list below.
So, let’s have a look at how developers are trying to create trust systems in web3.
1. OpenRank
EigenTrust-based decentralized ranking and reputation protocol. You can think of it as a combination of smart contracts, off-chain scripts, and frameworks for actors’ interactions, which together analyze the graph and provide trust scores for nodes within it. If this sounds too abstract, imagine finding yourself in an unfamiliar environment surrounded by four friends and hundreds of strangers — how can you determine which strangers are trustworthy and which are not? With OpenRank, you can tell the protocol, “These are my four friends, and I trust them”, and the protocol will calculate trust scores for the strangers based on your friends’ statements and/or previous experiences.
Reality is a bit more complicated because:
trust is always context-dependent and never generalized — therefore, OpenRank operates only in specific environments with defined notions of trust (e.g., GitHub, Farcaster, etc.) and not across the web as a whole (though OpenRank strives to provide developers and protocols the ability to compose reputation from one use case context to another);
different environments demand different approaches to trust management — so, in addition to EigenTrust, other algorithms like “Hubs and Authorities” or “Collaborative Filtering” are also utilized. Each new environment requires custom logic to be developed by engineers with its own trust signals (what traits signal trustworthiness), trust-flow mechanics (how do these trust signals propagate through the graph), seed trust logic, etc.
often, you don’t have explicit “friends” in digital environments — in such cases, seed trust is derived not from your explicit statements but from implicit actions captured from your digital footprint (e.g., likes, transactions, transfers, etc.).
This approach is highly practical and can be implemented today in various contexts: social networks (for search, discovery, and recommendation), marketplaces (to make informed decisions before interacting with goods, software, applications, or sellers), consumer apps and wallets (for search and discovery functionalities), governance and public goods funding (for efficient resource allocation). You can explore live integrations of this protocol (including Farcaster, Lens, Optimism, Metamask Snaps, etc.) in its documentation.
As stated in the documentation, in the longer run, anyone will be able to publish their algorithms to OpenRank and earn rewards if their algorithms are adopted by application developers. Speculating further (in a hyper-long-term perspective), with the availability of a universal (likely, blockchains-based) social graph and a rich variety of digital footprints attached to its nodes, we might see the emergence of "on-demand trust systems" capable of automatically interpreting this data. Such systems could create a "trust layer" in any environment without requiring manual setup. We can even envision a future where personal AI agents generate personalized trust algorithms whenever we interact with others in the digital realm, transforming strangers in any digital environment into… strangers with trust scores! This might sound strange at the moment, but it will likely become the starting point for interactions with strangers in the not-so-distant future and serve as the foundation for novel mechanisms, such as the “dependency graph” proposed by Vitalik.
2. Intuition
Intuition calls itself the “Trust Protocol” but takes a different approach to both trust calculation and trust definition. While OpenRank’s idea is centered around identifying suitable algorithms and adapting them for specific digital domains, Intuition aims to “compose a universal and permissionless knowledge graph, capable of handling both objective facts and subjective opinions”. In other words, OpenRank computes on top of existing graphs, whereas Intuition seeks to create its own universal graph with natively integrated trust-related data.
The core idea behind Intuition is to establish an incentivized ecosystem where participants collaboratively build an open and semantic knowledge graph using “identity”, “claim”, and “stake” as its core primitives. Intuition’s 85-page whitepaper contains a wealth of information about the nature of trust, the landscape of ideas surrounding digital trust, and thoroughly developed concepts underpinning Intuition’s approach.
My personal belief is that everything we need to establish digital trust already exists within the digital realm. The challenge lies in extracting the relevant data and interpreting it effectively, rather than creating a universal system for achieving “social consensus on globally persistent canonical identifiers”. Which approach will prove more viable remains to be seen.
Ethos provides a credibility score with an accompanying profile, similar to a credit report but built on open protocols and on-chain records. The core concept of the Ethos Protocol is social Proof of Stake — a decentralized, consensus-based validation mechanism rooted in human values, judgment, and actions. The idea is that people will write reviews about anything in web3, vouch their ETH for others, and face slashing penalties if those they vouched for engage in unethical actions. The Ethos team argues that “In the not-so-distant future, we assume other participants will likely [likely ‘not’ is missed here] take you seriously until you have an Ethos profile”.
4. Legion.cc
Legion is a “merit-based, on-chain fundraising platform” with an integrated on- and off-chain reputation system, represented by the Legion Score, at its core. The concept is that Legion’s algorithms calculate potential investors’ scores based on available data (e.g., number of on-chain transactions, interactions with specific protocols, social graph connections, etc.). Founders can then review these scores, select their most-desired attributes, and present a curated basket of investors with early-stage offerings or token sale allocations.
Union Finance positions itself as a member-owned credit protocol built on Ethereum where members can underwrite lines of credit to other member addresses. Although it is not a classical trust network, it can be thought of as a mechanism to lower the cost of coordinating trust into available credit. Since the role of the system is not to create a mutual trust system but to merely combine individual trust links directed at the closest nodes (thus enabling nodes to “aggregate” trust from their trustors) and enable trustees to use this trust in the form of credit, we can say that Union Finance is a specialized trust system with a specific use case.
Circles calls itself a new money system that is based on individualized cryptocurrencies and a social graph of trust between these currencies. Inspired by the Sardex system and trust network concepts, CirclesUBI encourages its users to mint their own currency and set up trust in the personal currencies of other users. As the system grows, chains of trust (basically, a trust network) are formed within it — when one stranger wants to send money to another in Circles, they automatically search for a transitive chain of trusted currencies connecting them. The idea is that as the social graph of mutual trust becomes more interconnected, these personal currencies converge on one single global monetary system. Circles 2.0 has been launched recently.
Initially proposed by the CirclesUBI team, Circles Entropy is a system that facilitates anonymous trust and credit relations, enabling private transactions among its users. Trust and credit relations are expressed in the form of graphs, that form the fundamental data structures over which secure computation takes place. Focused primarily on the technical aspects, the project can be considered a generalized trust/credit network, creating a digital environment for managing information about the value (credit/IOU) contributed to the system.
The Trustlines Network describes itself as a community-driven project that empowers people to create their own money and access digital payments. Inspired by the ideas of time banks and complementary currencies, it aims to create a decentralized, permissionless, and open platform to host currency networks (an accounting system based on peer-to-peer credit). The value in these currency networks is represented in IOUs issued by its participants. The design builds upon the original Ripple idea.
Heterotopia is a theoretical concept of a world of scale-free credit money proposed by Christopher Goes (Co-founder of Anoma). It provides an overview of fiat money as a coordination mechanism and describes the world of scale-free credit money where control over the issuance of money is re-aligned with trust. In this world, inspired by the trade credit relationships in early societies, credit is personal, trust is distributed, everyone can print their money (though not everyone is willing to accept it) and retroactive trust-funding by institutions plays an important role.
10. Other approaches to digital trust
For those interested in ideas and solutions related to digital trust, I recommend exploring Omer’s articles from Chaos Labs, the TrueMarkets vision of oracles for prediction markets, and writings by Vitalik. While these resources don’t directly aim to create “trust layers” or integrate trust mechanisms into web3, they offer valuable insights and perspectives on the broader concepts surrounding trust and its applications.
In recent months, we have clearly seen a growing interest in trust systems and in identifying the domains where frameworks provided by these systems can be leveraged. The aforementioned projects offer different perspectives on the question of which domains these systems are suitable for:
projects like OpenRank aim to establish trust through the digital footprints we leave across various environments, with their approach focused on finding proper algorithms to interpret these footprints;
initiatives such as Intuition and Ethos Network aspire to create a universal trust layer that could eventually underpin almost every digital communication. Their short- and medium-term tasks involve building the foundation for individuals to contribute the data needed for the creation of such a layer;
Legion takes a more practical, ground-to-earth approach by applying trust frameworks within the narrow web3 investment domain;
Union Finance envisions trust systems as tools for resource allocation, including traditional loans, and for delivering faster, higher-quality liquidity;
Trustlines Network elaborates on Ripple's ideas by concentrating on credit networks as mechanisms for transferring value in the form of IOUs;
platforms like CirclesUBI and the creators of Heterotopia articulate concepts of community trust-based currencies, imagining a future where money is re-aligned with trust;
meanwhile, developers behind projects such as Circles Entropy focus on the development of underlying algorithms for trust and credit network systems, abstracting away from immediate real-world use cases.
The future will show which approach proves to be the most useful and in demand. However, it is impossible not to notice that all these projects share a common idea: rejecting opaque, closed systems with numerous intermediaries in favor of open systems centered around individuals and their opinions (expressed either explicitly or implicitly) about their environments. I am confident that the era of digital corporations and algorithms designed solely for retention or profit will soon become a thing of the past, much like the trust systems offered by states are fading away.
I also believe that the concept of digitizing trust, which has been evolving for decades, will form the foundation of what, in 15 years, we will call the "trust layer" of the internet. Furthermore, I predict that one of the projects mentioned above will become a unicorn and likely serve as the backbone for countless protocols in web3. For now, however, we’re still early in this journey. The best we can do to contribute to a brighter future is to leave our digital footprints and lay the groundwork for trust in our future digital identities.
The text below attempts to dissect the phenomenon of interpersonal trust: what we mean by “trust” today, how we arrived here, and how it has evolved in the digital era. If you’re only interested in crypto protocols building their products around this term, feel free to jump to the last section. Additionally, every major step on the path to an understanding of trust will be accompanied by a picture.
We will discuss:
simplified evolution of interpersonal trust irl: early societies, pre-internet states, digital era
attempts to establish trust in digital environments: trust networks, credit networks (including many useful links to academic research)
novel approaches to trust modeling in digital environments: a brief review of the new cohort of web3 protocols
this paper is a part of research done jointly with R Alter, D Mukh, and M Volkh
As Wikipedia suggests, “trust is the belief that another person will do what is expected. It brings with it a willingness for one party (the trustor) to become vulnerable to another party (the trustee), on the presumption that the trustee will act in ways that benefit the trustor. In addition, the trustor does not have control over the actions of the trustee”. Some say that trust “is often viewed as the glue that holds society together”, and it is hard to argue against it. Indeed, trust is something we encounter on a daily basis throughout our lives. It is explicitly observable in direct interpersonal relationships, but it also forms the intangible basis for numerous systems (political, economical, and generally societal) that we interact with, while being hidden under layers of abstractions. Trust relations are so deeply integrated in our society that one often cannot simply identify them. Therefore, before we attempt to understand the nature of trust in digital realms and move to the truly astonishing “trust-related things”, let’s skim through the simplified history of the evolution of trust and try to point out some of the locations of trust in the fabric of human society. At this point it is important to articulate the key properties of the phenomenon we call “trust”, rehashing the definition above:
it is a person’s attitude towards another person — an expectation that the trustee will act in a specific way;
the person is uncertain about the outcome of the trustee's actions;
trust does not imply control over the trustee's actions.
It is also important to clarify that we will not talk about trust towards things or trust in ourselves – interpersonal trust (directed towards a specific human being — primarily from another human being, but sometimes from the collective of other human beings) will be at the core of our discussion. However, since different types of trust often overlap (e.g., I trust Alex because I trust the system he’s part of), I will occasionally discuss other types of trust as well. Additionally, the concept of generalized trust (which is more about one’s state of mind rather than interpersonal attitudes) is outside the scope.
By “early societies” I mean collectives of gatherers, hunters, and/or barbarians that existed before the emergence of states, including those of the ancient world. As far as we know from history and anthropology, these societies were often isolated from each other, sometimes fighting among themselves, but more often with nature. They could be characterized by primitive, pre-money economies, low levels of division of labor, a high level of interpersonal trust, and strong communal ties. These communities relied heavily on shared resources and mutual aid to survive. Leadership structures were often informal and based on kinship or merit rather than formalized hierarchies. Social norms and customs were typically transmitted orally through generations, helping to maintain cohesion and continuity within the group.
In the table below I mention some of the activities typical for these kinds of communes and the role of trust in them. Since the goal here is not to give an extensive historical perspective on early societies but to identify the role of interpersonal trust within them, I intentionally modernize, simplify, and let go of some details.
We could elaborate further, but the bottom line here is that trust-requiring interactions in such societies were based on past interactions and/or assessments done by trusted third parties (family, known members of the community, etc). Not a very interesting table, to be honest. At this point, it is difficult to find scenarios where one has to trust someone without having visible direct or indirect (through trusted third parties) connections (or “trust links”) with the trustee. Again, the whole picture is very simplified, but the core idea here is that trust is rarely directed toward strangers — no system capable of providing a basis for trust in strangers exists (yet).
With the advent of banks and the centralization of power and culture, we are slowly but steadily approaching the modern state of affairs, which shifts its focus away from private and direct trust links.
By “states” in this paragraph, I mean societies (both ancient and modern, post-industrial revolution) characterized by the development of centralized governance, legal systems, and administrative structures, including 'proto-states' such as city-states or Greek poleis. The distinct feature of interpersonal relationships within states is the higher frequency of interactions among strangers compared to early societies. Acquiring goods or services, storing money in the bank, or hiring an employee — all of these activities imply some level of trust. However, in the absence of past personal or third-party experience, the logic of early societies doesn’t work anymore. Thus, other sources of trust (or, at least, instruments to reinforce the personal assessment of a potential trustee) arise. This was the period for the first cohort of “trust systems” to flourish.
At a very high level, we can describe them as systems that replace the need to trust a stranger with trust in the system or provide us with extra information to form a personal assessment of a potential trustee. Thus, the system must take something as an input to assess a potential trustee and produce something as an output. Some notable examples of such systems include legal systems, regulatory bodies, and certification processes. But to be more precise, let’s draw a new table — this one will be a little bit more complicated.
There are some interesting conclusions we can draw from this table:
In pre-internet states, trust tends to concentrate around trust systems. More and more often, trust is directed toward a system, and interpersonal trust (toward a specific human being) emanates from the trust in the system (of state’s rules, self-regulation, religious institutions, guilds and other non-state entities). Though the system itself still needs to assess the trustworthiness of this human being somehow — in some sense, we can say that we delegate the process of the trustee’s assessment to the system.
Trust systems can play different roles. Some of them can reinforce our personal assessment of a potential trustee (legal contract with an employee), others can relieve us from making this assessment (usually it is connected to society's well-being — for example, uncertified or unlicensed medicines merely can’t enter the market). When we are talking about the collective assessment of trustees, the goal of the trust system is to provide a framework for this process.
Trust systems are usually disconnected from each other – they operate each under their own set of rules and pursue their own specific goals. In other words, no system can provide a generalized framework for trust assessment, potentially adaptable across different domains.
An increase in contextual information relevant for forming a personal assessment. As society becomes more complex, one can form some level of “basic trust” towards a potential trustee from the general understanding of how the world works and indirect clues in the form of reputation — for example, being members of the same community (soccer fans or university students) or the process called the “shadow of the future”.
The growing role of the state. In the absence of an objective opportunity to rely on personal experience or reviews from friends, states and their norms start to play a huge role in trust assessment. By “state” here I mean not only directly the rules of law but also their “derivatives” — banking regulations, mandatory certification (even if it is carried out by the community), etc. Systems provided by private entities and international organizations also start to play significant role in trust assessment (e.g., ISO certifications, financial credit ratings).
Below, you can see a visualization of the trust flow in modern society for the "hiring a lawyer" case.
At the same time, we can find examples of systems trying to grasp and manipulate a more generalized notion of interpersonal trust as “the belief that another person will do what is expected”. These include the concepts of mutual credit systems and their close relatives: LETS (local exchange trading systems) and local currencies, CES (community exchange systems) and community currencies, time banks and time-based currencies. The most prominent example is Sardex (which is post-internet phenomenon, but since it’s a “child” of the aforementioned concepts, let it be in this section), which could be thought of as a mutual credit system, taking as input promises to provide goods or services from various parties and issuing in exchange for them a local Euro-denominated complementary currency circulating within Sardex members in cycles or loops. While the goal of the Sardex system is focused on facilitating local economic development and resilience, the underlying logic represents a trust system, providing the ability to transfer and net mutual trust-based commitments towards other members of Sardex.
What truly distinguishes Sardex from common trust systems in the pre-internet states is the role of interpersonal (or “intercompany,” since we are talking about B2B relations here) trust. As explicitly stated in Sardex original paper, “Trust is a fundamental feature of Sardex. It is community-specific and it relies on values and reciprocal expectations... Trust works at two levels: as a starting point to join the network and towards the Sardex organization (direct), and as a transactions’ lubricant between members (indirect).” As we will see later, this idea of utilizing interpersonal trust (local trust links) for the creation of a generalized trust system (in contrast to the state’s vague sparse trust systems, where local trust links often are not taken into account at all) will find new life.
But at this point, let’s finish the description of trust systems in pre-internet states with the following question: do the state’s norms (and their “children and grandchildren” — banking regulations, municipal rules, and other imperative frameworks) form the best possible framework for assessing trustworthiness, or were they merely a good enough solution for a pre-internet era when coordination of people on a large scale was impossible without a centralized actor?
In the era of the internet, the table above is still relevant. States are still “the kings,” and their trust systems still play the dominant role. However, new kings and new systems are emerging.
Let’s present another, revised table. Since the pre-internet trust systems from the previous table (government licenses, community certificates) are still in place, I will not repeat myself.
What structural changes and patterns can we see from the table:
The constant emergence of new environments where trust is essential. In the pre-internet era, we learned to assess the trustworthiness of strangers based on the context and environment around us. In the digital world, we constantly find ourselves in new settings where we are encouraged to interact with strangers — new apps, online stores, course sellers, and so on. However, in the digital environment, there is no familiar and intuitive context for most of us, nor are there established trust systems.
Replacing opaque and subjective systems (like legal norms) with more objective algorithms. Today, not only symbols from the government and its institutions (certificates, licenses, etc.) can provide some level of trust, but also outputs by Google’s or other tech companies' algorithms (reviews in Google Play, alternative credit-score systems provided by fintech startups, Community notes in X, etc.).
A sharp increase in domains with available trust sources. Now, before entering into any interpersonal interaction where previously we would have faced a lack of information to formulate a personal assessment of a potential trustee, there is help from tech companies (reviews in Google Maps, ratings on specialized websites, etc.).
Back to interpersonal trust. Modern technologies have enabled more sophisticated trust systems that integrate both interpersonal and system-based trust mechanisms, such as online reviews and ratings. In the post-internet era, once again, we are paying more attention to other people's opinions rather than the opinions formed within state’s systems. The internet brings us the ability to grasp and “digitize” millions of individual inputs. In some sense, this is similar to early societies, where before entering a trust-requiring interaction with a stranger, we asked other people’s opinions. However, these people are mostly strangers too.
Diversity of trust systems. While in the pre-internet era, the collective assessment of trustworthiness was a privilege of complex and often vague trust systems (like state’s norms, certificate processes), new technologies enable the creation of novel systems with custom logic and algorithms for the trust assessment. Some of them (like review platforms) take as input the assessment of a potential trustee from hundreds or thousands of individuals, thus, again, in some sense bringing us back to old societies where only trustees’ assessments made by other humans mattered.
While these novel approaches to trust systems provide the opportunity to bring the output of the system (rating, top reviews, etc.) closer to the real trustee’s assessment made by other human beings (in contrast to pre-internet systems, where interpersonal trust and people’s opinions were often disregarded), we should note the potential for algorithmic bias and manipulation in modern trust systems. For example, online review platforms can be gamed, and algorithms can reflect and perpetuate existing biases.
One could ask, "Is it possible to abstract away from specific trust-requiring interactions and create a general framework for digitizing and transmitting trust”? In other words, could we construct a trust system without answering the question of "What is this system needed for”? If yes, what should this system take as input, and what is the best source of information about potential trustees? These are the questions that led to the creation of what we call “trust networks”.
A trust network is a term that encompasses any network where trust relationships are established, maintained, and propagated among members. These networks can be applied in various contexts, such as social networks, business partnerships, and online communities. The primary focus of a trust network is on the overall structure and dynamics of trust among its members. In such a network, trust relationships are built based on past interactions, reputation, and shared values or objectives. We can find first attempts to formalize the notion of trust in digital realms in research papers from the early 1990-x, leading to a wide range of various approaches to trust modeling and management in digital environments (check researches here and here).
Key concepts related to trust networks are:
Graph-based models. Trust networks are often modeled as a graph — a set of nodes and connecting edges, where nodes represent individuals or entities, and edges represent the trust relationships between them.
Trust propagation. Trust can be propagated through the network, meaning that if person A trusts person B and person B trusts person C, person A might have some level of trust in person C by extension.
Reputation systems. These systems often underlie trust networks, where the reputation of nodes (often derived from the history of node’s interactions) is tracked and used to influence trust decisions.
Decentralization. Trust networks can be centralized or decentralized, with decentralized networks often relying on distributed trust models.
Let’s put it in this way — trust networks provide the framework and mechanisms for trust to flow, be evaluated, and be leveraged within a community, much like how infrastructure supports the flow of goods or information in other types of networks. These systems can be seen as generalized trust systems, where the level of trust in any specific node is derived from the other nodes’ inputs (in the form of explicitly stated assessments or information about past interactions) regarding the trustworthiness of their closest neighbors.
Some prominent examples of trust networks include:
PageRank. While not being a trust network per se, it models a kind of implicit trust by interpreting hyperlinks as votes of confidence or relevance. Pages linked to by many others, especially by highly linked pages, are considered more "trustworthy" or important within the network.
Eigentrust. The system assigns a trust score to each participant based on the trustworthiness of their connections. The score considers not just who they trust, but also how trustworthy those connections are themselves (based on past interactions). It uses mathematical calculations to analyze these connections and arrive at a global trust vector for the network.
Web of Trust (WoT). Used in public-key cryptography like PGP (Pretty Good Privacy) and S/MIME (Secure/Multipurpose Internet Mail Extensions). Unlike Eigentrust's mathematical scoring, WoT relies on manual trust assignments. Users directly verify the identities of their contacts and sign their public keys, indicating trust. Trust spreads through connections as users sign each other's keys.
PeerTrust. PeerTrust is a proposed trust model for the P2P environment. It evaluates peer trustworthiness rating based on 5 elements: the feedback obtained from other peers about the target peer, the aggregated number of total transactions that the target peer has with others, the credibility factor of the recommender peers, the transaction context factor for discriminating the mission critical transaction from less or non-critical transactions, and the community context factor.
Other (mostly theoretical) proposed models for trust networks:
The parameters of these systems vary depending on whether it is a generalized trust network or a trust network for specific purposes (e.g., file sharing) or environments (peer-to-peer, multi-agent system, e-commerce, etc.). Additionally, the role of node reputation (history of node interactions) and initial node trust (pre-trusted peers) differs in these systems. Despite the various approaches to defining the rules of network functioning and node interactions within it, all these systems share a common fundamental task: to create a common trust system based on inputs from multiple actors/nodes regarding their degree of trust in their neighbors. The main output of this system is information about the collective assessment of trustworthiness for a node within this system.
It is noteworthy that many theoretical models of trust networks have not found application in real life. Moreover, in many cases, the authors did not even set out to determine the scope of their application, mentioning the applicability of their theoretical research for reputation and rating systems (in more modern papers — also in IoT and different types of wireless networks), or leaving questions about areas of practical application to future researchers.
At the same time, the concept of a credit network, which is a close relative of a trust network, was born. The credit network model was invented independently by (at least) four distinct groups of researchers, motivated by somewhat different issues and applications, but arriving at the same essential elements:
DeFigueiredo & Barr sought a reputation system with bounded loss from coalitions of malicious users;
Ghosh et al. aimed to support distributed payment and multi-user credit checking for multi-item auctions;
Karlan et al. wanted to construct an economic model of informal borrowing networks;
Mislove et al. were concerned with deterring spam.
A common thread in the objectives of these researchers was to capture a notion of pairwise trust, representable in quantified terms. In each case, the trust measure is grounded by interpreting the quantity as a capacity for transaction. That is, the degree of trust in one agent for another is measured by how much it is willing to expose itself to transactions with that counterpart. In other words, the model operationalizes trust as an extension of credit, in a framework where a credit balance entitles an agent to transact with the agent granting credit.
Credit networks can be seen as systems for trust modeling and management in digital environments, where trust is financialized in its most generalized form — in the form of a credit. It is important to note that “credit” in these systems can represent either a concept close to its traditional meaning (the transfer of money or other value with the trust that the borrower/trustee will return it in the future) or an IOU (I-Owe-You — the promise to provide some work, service, or any other value to the members of the network).
As a consequence, the primary goal of the first type of credit networks (check “Trust and social collateral” by D. Karlan) is to transform multiple nodes’ inputs into a score which represents the amount of value that could be entrusted to a node (with the expectation of returning the value later), while the connections between nodes are viewed as collateral (which likely would be lost in the case of a node’s inability to return the value) — this approach is similar in many ways to how informal borrowing works in real life. Indeed, we often lend/borrow money to/from our friends at the risk of damaging our good relations. Though, in real life, it usually happens between two specific persons, while in this type of credit network the level of trust is determined by the structure of the entire network.
In the second type of credit network (check “Liquidity in Credit Networks: A Little Trust Goes a Long Way” by P. Dandekar), credit (often described as a node’s personal currency) is merely a node’s unconditional promise to bring some value to anyone in the system, and other nodes’ inputs represent the level of trust in the trustee to provide this value. While these ideas can’t be mapped directly into our “real-life” relationships and traditional concepts of credit, they allow the network to serve as a decentralized payment infrastructure, where arbitrary payments can be routed through the network by passing IOUs between trusting nodes in their respective currencies.
The concept of credit networks was further elaborated by Rayan Fugger (Ripple’s cofounder) and other researchers: Pedro Moreno-Sanchez, Pranav Dandekar, Geoffrey Ramseyer, and others.
As with studies dedicated to trust networks, many of these studies did not aim to find practical applications for the formulated concepts, let alone search for product market fit. Unsurprisingly, most of them remained theoretical, though some core concepts of credit networks are used in the Lightning Network architecture, which can be thought of as a type of credit network, where the “credit limit” is the amount of Bitcoin that has been committed to each channel (essentially, “credit” or IOUs here are collateralized by BTC).
The first, and for a long time the only, attempt to implement a “pure” credit network (a credit network where IOUs are literally promises or expectations without any collateral) in practice was the first version of the Ripple network, traces of which (named “Community Credits”) can still be found on Ripple’s website. It was designed to enable people to extend credit to each other, forming a network of IOUs. In this system, users could create trustlines with each other, allowing for the transfer of value based on mutual trust. The idea was to facilitate transactions through a web of trust where people could pay each other through intermediaries who had established credit relationships.
But facing reality is much tougher than conceptualizing it in a paper — the initial Ripple ideas were abandoned in favor of serving banks’ capital. The villages.io website, providing access to the “Decentralized Barter Credits and Collaboration System” (essentially, the first version of RippleNet), is still alive. However, it is more of a historical artifact than a functioning credit network.
After Ripple abandoned the implementation of its original ideas, interest in credit networks (at least in creating projects based on such ideas) subsided for a while. However, today we are witnessing a return to and a reevaluation of the ideas behind credit networks. The underlying architecture still represents a trust system, where trust is manifested as credit flowing through the network, but the details have significantly changed. So, let’s take a look at the key projects developing this idea.
My strong belief is that viable long-term solutions for trust modeling in digital environments can only be based on blockchains with open and accessible data. Yes, there are a lot of custom solutions for trust management in digital domains, where the notion of trust is domain-specific. Just to name a few: credit risk analysis, spam/fraud detection systems, loyalty programs, reputation systems, social media content moderation, identity verification services, knowledge-sharing platforms, etc. And yes, there are plenty of novel, creative techniques used within these solutions. But you won’t see them in the list below.
So, let’s have a look at how developers are trying to create trust systems in web3.
1. OpenRank
EigenTrust-based decentralized ranking and reputation protocol. You can think of it as a combination of smart contracts, off-chain scripts, and frameworks for actors’ interactions, which together analyze the graph and provide trust scores for nodes within it. If this sounds too abstract, imagine finding yourself in an unfamiliar environment surrounded by four friends and hundreds of strangers — how can you determine which strangers are trustworthy and which are not? With OpenRank, you can tell the protocol, “These are my four friends, and I trust them”, and the protocol will calculate trust scores for the strangers based on your friends’ statements and/or previous experiences.
Reality is a bit more complicated because:
trust is always context-dependent and never generalized — therefore, OpenRank operates only in specific environments with defined notions of trust (e.g., GitHub, Farcaster, etc.) and not across the web as a whole (though OpenRank strives to provide developers and protocols the ability to compose reputation from one use case context to another);
different environments demand different approaches to trust management — so, in addition to EigenTrust, other algorithms like “Hubs and Authorities” or “Collaborative Filtering” are also utilized. Each new environment requires custom logic to be developed by engineers with its own trust signals (what traits signal trustworthiness), trust-flow mechanics (how do these trust signals propagate through the graph), seed trust logic, etc.
often, you don’t have explicit “friends” in digital environments — in such cases, seed trust is derived not from your explicit statements but from implicit actions captured from your digital footprint (e.g., likes, transactions, transfers, etc.).
This approach is highly practical and can be implemented today in various contexts: social networks (for search, discovery, and recommendation), marketplaces (to make informed decisions before interacting with goods, software, applications, or sellers), consumer apps and wallets (for search and discovery functionalities), governance and public goods funding (for efficient resource allocation). You can explore live integrations of this protocol (including Farcaster, Lens, Optimism, Metamask Snaps, etc.) in its documentation.
As stated in the documentation, in the longer run, anyone will be able to publish their algorithms to OpenRank and earn rewards if their algorithms are adopted by application developers. Speculating further (in a hyper-long-term perspective), with the availability of a universal (likely, blockchains-based) social graph and a rich variety of digital footprints attached to its nodes, we might see the emergence of "on-demand trust systems" capable of automatically interpreting this data. Such systems could create a "trust layer" in any environment without requiring manual setup. We can even envision a future where personal AI agents generate personalized trust algorithms whenever we interact with others in the digital realm, transforming strangers in any digital environment into… strangers with trust scores! This might sound strange at the moment, but it will likely become the starting point for interactions with strangers in the not-so-distant future and serve as the foundation for novel mechanisms, such as the “dependency graph” proposed by Vitalik.
2. Intuition
Intuition calls itself the “Trust Protocol” but takes a different approach to both trust calculation and trust definition. While OpenRank’s idea is centered around identifying suitable algorithms and adapting them for specific digital domains, Intuition aims to “compose a universal and permissionless knowledge graph, capable of handling both objective facts and subjective opinions”. In other words, OpenRank computes on top of existing graphs, whereas Intuition seeks to create its own universal graph with natively integrated trust-related data.
The core idea behind Intuition is to establish an incentivized ecosystem where participants collaboratively build an open and semantic knowledge graph using “identity”, “claim”, and “stake” as its core primitives. Intuition’s 85-page whitepaper contains a wealth of information about the nature of trust, the landscape of ideas surrounding digital trust, and thoroughly developed concepts underpinning Intuition’s approach.
My personal belief is that everything we need to establish digital trust already exists within the digital realm. The challenge lies in extracting the relevant data and interpreting it effectively, rather than creating a universal system for achieving “social consensus on globally persistent canonical identifiers”. Which approach will prove more viable remains to be seen.
Ethos provides a credibility score with an accompanying profile, similar to a credit report but built on open protocols and on-chain records. The core concept of the Ethos Protocol is social Proof of Stake — a decentralized, consensus-based validation mechanism rooted in human values, judgment, and actions. The idea is that people will write reviews about anything in web3, vouch their ETH for others, and face slashing penalties if those they vouched for engage in unethical actions. The Ethos team argues that “In the not-so-distant future, we assume other participants will likely [likely ‘not’ is missed here] take you seriously until you have an Ethos profile”.
4. Legion.cc
Legion is a “merit-based, on-chain fundraising platform” with an integrated on- and off-chain reputation system, represented by the Legion Score, at its core. The concept is that Legion’s algorithms calculate potential investors’ scores based on available data (e.g., number of on-chain transactions, interactions with specific protocols, social graph connections, etc.). Founders can then review these scores, select their most-desired attributes, and present a curated basket of investors with early-stage offerings or token sale allocations.
Union Finance positions itself as a member-owned credit protocol built on Ethereum where members can underwrite lines of credit to other member addresses. Although it is not a classical trust network, it can be thought of as a mechanism to lower the cost of coordinating trust into available credit. Since the role of the system is not to create a mutual trust system but to merely combine individual trust links directed at the closest nodes (thus enabling nodes to “aggregate” trust from their trustors) and enable trustees to use this trust in the form of credit, we can say that Union Finance is a specialized trust system with a specific use case.
Circles calls itself a new money system that is based on individualized cryptocurrencies and a social graph of trust between these currencies. Inspired by the Sardex system and trust network concepts, CirclesUBI encourages its users to mint their own currency and set up trust in the personal currencies of other users. As the system grows, chains of trust (basically, a trust network) are formed within it — when one stranger wants to send money to another in Circles, they automatically search for a transitive chain of trusted currencies connecting them. The idea is that as the social graph of mutual trust becomes more interconnected, these personal currencies converge on one single global monetary system. Circles 2.0 has been launched recently.
Initially proposed by the CirclesUBI team, Circles Entropy is a system that facilitates anonymous trust and credit relations, enabling private transactions among its users. Trust and credit relations are expressed in the form of graphs, that form the fundamental data structures over which secure computation takes place. Focused primarily on the technical aspects, the project can be considered a generalized trust/credit network, creating a digital environment for managing information about the value (credit/IOU) contributed to the system.
The Trustlines Network describes itself as a community-driven project that empowers people to create their own money and access digital payments. Inspired by the ideas of time banks and complementary currencies, it aims to create a decentralized, permissionless, and open platform to host currency networks (an accounting system based on peer-to-peer credit). The value in these currency networks is represented in IOUs issued by its participants. The design builds upon the original Ripple idea.
Heterotopia is a theoretical concept of a world of scale-free credit money proposed by Christopher Goes (Co-founder of Anoma). It provides an overview of fiat money as a coordination mechanism and describes the world of scale-free credit money where control over the issuance of money is re-aligned with trust. In this world, inspired by the trade credit relationships in early societies, credit is personal, trust is distributed, everyone can print their money (though not everyone is willing to accept it) and retroactive trust-funding by institutions plays an important role.
10. Other approaches to digital trust
For those interested in ideas and solutions related to digital trust, I recommend exploring Omer’s articles from Chaos Labs, the TrueMarkets vision of oracles for prediction markets, and writings by Vitalik. While these resources don’t directly aim to create “trust layers” or integrate trust mechanisms into web3, they offer valuable insights and perspectives on the broader concepts surrounding trust and its applications.
In recent months, we have clearly seen a growing interest in trust systems and in identifying the domains where frameworks provided by these systems can be leveraged. The aforementioned projects offer different perspectives on the question of which domains these systems are suitable for:
projects like OpenRank aim to establish trust through the digital footprints we leave across various environments, with their approach focused on finding proper algorithms to interpret these footprints;
initiatives such as Intuition and Ethos Network aspire to create a universal trust layer that could eventually underpin almost every digital communication. Their short- and medium-term tasks involve building the foundation for individuals to contribute the data needed for the creation of such a layer;
Legion takes a more practical, ground-to-earth approach by applying trust frameworks within the narrow web3 investment domain;
Union Finance envisions trust systems as tools for resource allocation, including traditional loans, and for delivering faster, higher-quality liquidity;
Trustlines Network elaborates on Ripple's ideas by concentrating on credit networks as mechanisms for transferring value in the form of IOUs;
platforms like CirclesUBI and the creators of Heterotopia articulate concepts of community trust-based currencies, imagining a future where money is re-aligned with trust;
meanwhile, developers behind projects such as Circles Entropy focus on the development of underlying algorithms for trust and credit network systems, abstracting away from immediate real-world use cases.
The future will show which approach proves to be the most useful and in demand. However, it is impossible not to notice that all these projects share a common idea: rejecting opaque, closed systems with numerous intermediaries in favor of open systems centered around individuals and their opinions (expressed either explicitly or implicitly) about their environments. I am confident that the era of digital corporations and algorithms designed solely for retention or profit will soon become a thing of the past, much like the trust systems offered by states are fading away.
I also believe that the concept of digitizing trust, which has been evolving for decades, will form the foundation of what, in 15 years, we will call the "trust layer" of the internet. Furthermore, I predict that one of the projects mentioned above will become a unicorn and likely serve as the backbone for countless protocols in web3. For now, however, we’re still early in this journey. The best we can do to contribute to a brighter future is to leave our digital footprints and lay the groundwork for trust in our future digital identities.
Increased global incentives and alienation. While the parochial outlook and way of life did not suggest many opportunities for financial optimization, globalization definitely did. Being able to acquire goods and services from more talented people and effective enterprises than available in the economic actor’s vicinity, they started to lose incentives to cultivate interpersonal trust. The ultimate result of that is a certain alienation from the emotional security of knowing a person, replaced by the security of punishment in case of disobedience, guaranteed by the third party.
Automation of trust-related decisions. Bitcoin, Ethereum, and other trustless cryptocurrencies and protocols are reducing the need for trust in many domains — primarily the reliance on banks as trusted counterparties for bookkeeping and value transmission. These systems transform "trust systems" into "reliable data transmission systems" by replacing human trust with contractual trust. Similar patterns — namely, the automated reconciliation of multiple individual trust assessments — can be observed in other contexts, such as P2P trust in BitTorrent and trust in relays in Tor.
Increased global incentives and alienation. While the parochial outlook and way of life did not suggest many opportunities for financial optimization, globalization definitely did. Being able to acquire goods and services from more talented people and effective enterprises than available in the economic actor’s vicinity, they started to lose incentives to cultivate interpersonal trust. The ultimate result of that is a certain alienation from the emotional security of knowing a person, replaced by the security of punishment in case of disobedience, guaranteed by the third party.
Automation of trust-related decisions. Bitcoin, Ethereum, and other trustless cryptocurrencies and protocols are reducing the need for trust in many domains — primarily the reliance on banks as trusted counterparties for bookkeeping and value transmission. These systems transform "trust systems" into "reliable data transmission systems" by replacing human trust with contractual trust. Similar patterns — namely, the automated reconciliation of multiple individual trust assessments — can be observed in other contexts, such as P2P trust in BitTorrent and trust in relays in Tor.
Subscribe to 0xucs.eth
Subscribe to 0xucs.eth
<100 subscribers
<100 subscribers
Share Dialog
Share Dialog
1 comment
Спасибо за статью. Особенно было интересно, учитывая что мы создаем и скоро запустим новую веб4 соцсеть и как раз сейчас улучшаем процесс социальных взаимодействий