Ctrlk
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Truth Oracle

The Intersubjective Optimistic-Oracle Protocol

white paper

Introduction

Truth Oracle is a protocol that can unlock new applications through a novel approach to decentralized micro-coordination.

The innovation behind this protocol lies in its decentralized dispute resolution mechanism: an intersubjective optimistic oracle framework which offers practical, as well as sophisticated, guarantees between disputing parties. Truth Oracle employs a multi-stage dispute resolution process. The protocol leverages specialized committees to resolve complex and nuanced disputes before allowing escalation up the protocol's hierarchy for further assessment. This approach adds granularity to the oracle and enables the protocol to scale beyond crude token voting constructs, to meet the demands of novel applications.

By utilizing a combination of committees, economic security and provably reputable Attesters, for the first time, we can build a permissionless optimistic oracle that provides the necessary ingredients to resolve a diverse range of disputes without leaving the protocol vulnerable to economic attacks. Notably, the participants in Truth Oracle enhance its credibility by contributing a diverse set of interests and expertise from multiple disciplines.

Status quo

Current standards for decentralized dispute resolution amount to various flavors of economic security and token voting schemes. Although useful for preventing spam and griefing attacks, these approaches leave much to be desired whenever there are ambiguous disputes.

One limitation of typical token voting arbitration is the bottleneck inherent in its single-step framework. In these naive designs, a dispute is immediately escalated to the highest arbiter who's responsible for making the first and final decision on the outcome of a dispute. Single-step arbitration frameworks don't enable the application of specialized insights, or provide granularity, to the dispute resolution process. Additionally, in case there are subjective disagreements, these oracle systems don't offer a path to escalate matters to more neutral arbiters.

In modern day optimistic oracles, a resolution can be proposed by anyone (or in some cases whitelisted parties) willing to pledge a bond behind their proposed resolution. These claims are “optimistically” considered valid unless otherwise disputed. Traditionally, there is a short challenge window after a claim is made, during which anyone can dispute the claim by pledging a bond equal to that of the resolver. Upon dispute, the matter is then escalated to a set of token holders for a final resolution.

Illustration of common optimistic oracles

Limitations of crude token resolutions

The above model is effective in preventing obviously malicious resolutions or disputes where the outcome is trivially verifiable. The more objective the nature of a dispute, the more effective those schemes are - e.g.) “Y/N did a specific transaction occur onchain as is being claimed?” - However, if a dispute is genuinely contentious, nuanced, or ambiguous, naive token voting oracles fall significantly short of acceptable performance.

Providing optimistic oracle services for subjective disputes is a very complex process. When a result isn't black and white, and legitimate arguments can be made for both sides of a dispute, it's naive to think that ordinary token holders would be well-equipped to accurately resolve these matters on a consistent basis. The ability to purchase a token does not equate to the qualifications necessary for dispute resolution. Even if, on average, a token vote reflects what should be the 'correct' outcome, this process is highly subjective and lacks the guarantees necessary to function at scale.

Aside from the limitations described so far, there are other obvious reasons for the inadequacy of naive token-voting oracles. The most critical reason is that plutocratic dispute resolution is vulnerable to economic manipulation. In such systems, nothing prevents a well-resourced bad actor from manipulating arbitrary outcomes in obvious attacks, or using their resources to gain a slight long term edge in less-obvious attacks.

Limitations:

  • Single-step arbitration

  • Absence of expertise

  • No escalation path

  • Lacks credibility at scale

  • Vulnerable to economic manipulation

Effective for:

  • Objective disputes

  • Easily verifiable claims

Ineffective for:

  • Subjective disputes

  • Nuanced / ambiguous outcomes

  • Complex claims

Proposed solution

To address the issues mentioned above, Truth Oracle employs a multi-stage optimistic oracle process. The various modules within the oracle are governed by the protocol token.

In recent years, the need for decentralized onchain coordination has led to the emergence of onchain sybil-resistant identities within the Ethereum ecosystem. Many of these individuals are delegates and governance participants, who demonstrate the skills and knowledge necessary to consistently coordinate and problem-solve in decentralized groups in order to reach consensus - a process very similar to dispute resolution.

In this paper, these large public clusters of sybil-resistant, reputation-based identities are referred to as: Proof of Reputation based Attesters (“PoR-Attesters” or just “Attesters” for short).

Attesters are assigned the status of highest arbiter in Truth Oracle. Initial disputes, however, are resolved by special committees. In this protocol, governance appointed specialists serve as the first line of arbitration and offer subjective insights in the resolution process. These specialist committees provide informed perspectives and enable the granularity necessary to offer oracle services at scale.

TRUE

The element that contributes economic security to Truth Oracle and its respective applications is the protocol’s governance token. TRUE provides economic utility to the oracle, governs the protocol, and designates resolution paths enabling the creation of optimistic markets.

A standard Truth Oracle implementation contains three stages of escalation, with the following hierarchy:

  1. Committee

  2. TRUE holders

  3. PoR Attesters

Illustration of Truth Oracle escalation process

  • “Resolution Paths” represent the method by which disputes are escalated. They are somewhat analogous to the supply chain behind a data feed in traditional non-optimistic oracles, like that of Chainlink.

  • In the context of this paper, the word “market” will often be referred to outside of its traditionally defined meaning, as an object that is resolvable based on a set of predefined rules (e.g., claim, question, attestation). A collection of markets, in the same context, is defined as an “application” that serves a particular audience, use case, or function.

—

Intersubjectivity

Truth Oracle incorporates intersubjectivity in its framework by overlapping the insights of specialist committees with the credibility of provably reputable Attesters. Additionally, the protocol leverages the subjective economic interests of TRUE holders to coordinate in keeping the oracle live. TRUE holders are incentivized to act honestly when they vote on outcomes, implement resolution paths, and govern the protocol, because TRUE only retains value if the protocol successfully functions as a credible, accurate, and live Oracle.

TRUE holders can add or remove various modules in the dispute resolution process. Resolution paths can include any combination of committees, token votes, or Attesters to attain a desired level of neutrality and intersubjectivity from the oracle.

The ability to alter resolution paths could be limited to markets created in the future, such that all active markets always settle with their originally designated paths. By providing guarantees around the terms of resolution upfront at the time of market creation, the protocol offers the assurances necessary for markets to function at scale with minimal trust assumptions.

Potential Applications

Truth Oracle comes with capabilities that can be used to unlock new decentralized applications. The first iteration of a Truth Oracle application is an event-based prediction market. Prediction markets are not novel themselves, but they offer a proven use case for dispute resolution which will condition the protocol in preparation of future applications and novel micro-markets.

Potential Truth Oracle applications:

  • variations of community notes for social platforms

  • optimistic provenance for filtering human/ai content online

  • escrow services for decentralized P2P markets

  • Futarchy or “governance as a service” experiments

  • real-time data feeds with a combination of economic and reputational guarantees

Protocol Description

Various markets require different types of expertise to resolve accurately. Often, disputes among market participants can be highly nuanced and ambiguous. For that reason, in a default Truth Oracle implementation, specialist committees are appointed by TRUE holders to serve as the initial arbitrators of disputes.

In the ideal case, markets are resolved optimistically, in that there are no disputes with the proposed resolution, and after a challenge period elapses, the market can be considered resolved, matured/expired, executed, etc., depending on the context of the market.

In order to dispute the above mentioned resolution, a disputer simply puts up a bond and files a dispute. Once the original resolution is disputed, it is then up to an appointed committee to resolve the dispute and decide the slashing conditions for the posted bonds. After another short challenge window passes, the market automatically resolves.

In the event that the committee proposed resolution is disputed, the matter is escalated for TRUE holders to weigh in. TRUE holders will then vote on the outcome of the market and set the necessary slashing conditions.

The above solution, however, is insufficient on its own. It’s entirely possible that token holders (being plutocratic in nature) are somehow compromised, acting out of self interest when resolving markets, or otherwise behaving dishonestly. In order to prevent token holders from being able to manipulate results through sheer economic force, another sybil resistant mechanism must be used to offer final verdicts which keep the dispute resolution process honest.

Honesty mechanism

As a last line of defense, Truth Oracle taps into pools of onchain reputation-based identities who take part in day-to-day governance for protocols within the ethereum ecosystem (PoR Attesters). These onchain identities represent individuals and organizations who possess the skills and knowledge necessary to vote on nuanced, complex matters.

The benefit of Attesters with Proof of Reputation is that these individuals are selected through the governance of the respective protocols they contribute to. Those protocols often have strong sybil-resistant delegation mechanisms, thus representing identity clusters that are infeasible to manipulate. Compromising such a system would require infiltrating the governance of several highly active projects.

The only way to be eligible for Truth Oracle Attestership, is to execute the human output (work) necessary to become a reputable governance delegate for a large Ethereum ecosystem project, in hopes of being recognized by TRUE holders and admitted into Attester Pools.

By leveraging provably reputable Attesters, Truth Oracle ensures that markets can be credibly resolved even in the event that the majority of token holders are compromised. Meaning that the protocol can remain honest for all existing resolution paths, even if it’s governed by Byzantine token holders, as long as a supermajority of all Attesters are honest.

Consistent with lower level disputes in the protocol, escalation must occur within the designated challenge window after a token holder vote. In order to escalate disputes to an Attester vote, a substantial bond is necessary. The purpose of the large bond is to disincentivize market participants who are simply unsatisfied with market results from easily griefing the protocol. In the event that a dispute is dishonest in nature, the pledged bond would be slashed.

Depiction of an example Truth Oracle implementation

Following Precedent

It should be noted that the primary purpose of Attesters is to ensure that token holders are unable to collude and manipulate results, not necessarily to offer deep insights into the intention of a market (that’s the function of specialist committees). This means that in the ideal case, where the game theory of the protocol is sustained, Attesters would mostly be deferring to the decisions of token holders, who would likewise be deferring to the resolutions proposed by committees.

That philosophy draws from decades of precedent set by judicial systems in developed countries, where the higher courts often defer to lower court decisions, unless there’s a clear deviation from accepted norms.

Chronology of dispute resolution process:

Stage 0)

Resolution is proposed permissionlessly:

  • Resolver puts up a bond and proposes a resolution

Stage 1)

Resolution is disputed within the challenge window:

  • Someone disagrees and disputes the result by putting up an equal bond

Stage 2)

Appointed committee resolves dispute:

  • Committee votes on resolution

  • Committee decides slashing conditions for the resolver

  • Committee decides slashing conditions for the disputer

Stage 3)

Escalation occurs within the challenge window:

  • Someone puts up a bigger bond (than the original dispute) to escalate the proposed resolution

Stage 4)

Token holders resolve disputes:

  • Token holders vote on outcome of the market

  • Token holders vote on slashing conditions of resolver

  • Token holders vote on slashing conditions of 1st disputer

  • Token holders vote on slashing conditions of 2nd disputer

Stage 5)

Escalation occurs within the challenge window:

  • Someone puts up a substantial bond to dispute the Token Holder result

Stage 6)

A random selection of Attesters from a preset pool submit a final verdict:

  • A function is called which randomly selects a set number of active addresses from a provably reputable Attester Pool

  • If an Attester is non-responsive, another is selected from the Attester Pool

  • Attesters vote on outcome of the market

  • Attesters vote on slashing conditions of resolver

  • Attesters vote on slashing conditions of 1st disputer

  • Attesters vote on slashing conditions of 2nd disputer

  • Attesters vote on slashing conditions of 3rd disputer

Economics and game theory of protocol actors:

Truth Oracle has intersubjective elements that involve a diverse set of actors such as specialist committees, token holders, and attesters. The economic incentives of these various actors differ greatly from one another and must be independently accounted for.

The various actors in the protocol:

  • Market creators/participants

  • Resolvers

  • Disputers

  • Initial resolution committee

  • $TRUE holders

  • Attesters

  • Moderators / Guardians - *optional

Each of the actors above must be incentivized to fulfill their respective role within the protocol. In the ideal case, all economic incentives for these participants come from fees generated by the protocol via resolved markets. In the early days of the protocol, however, it’s possible that extra incentives will be required to ensure that resolutions and disputes occur in an orderly manner.

Truth Oracle is an Optimistic Oracle, meaning that in the ideal scenario, markets resolve without the need for any intervention on the part of the protocol. The oracle mechanism only activates in the event that there is a dispute.

Resolvers:

The resolver role is permissionelessly carried out by independent actors within the system who monitor the results of unresolved markets. By resolving a market correctly, resolvers earn a modest reward that can either come from a set fee baked into the market, or a small incentive in TRUE tokens from the protocol.

In order to resolve a market, resolvers must put up a small bond that is held until a challenge period elapses. During that challenge window, the resolver’s bond is potentially at risk of being slashed if the proposed resolution is successfully disputed.

Disputers:

Like the resolver role, disputing is also permissionless. Anyone who disagrees with a proposed result can put up a bond and file a dispute. Disputers are typically the same actors as resolvers; they’re mostly made up of individuals who monitor the results of markets and react accordingly (either because they have exposure to the underlying markets or because the economic incentives compel them to).

In Truth Oracle, disputes can happen at various stages in the resolution process. In a standard Truth Oracle implementation, there are three stages of potential disputes and four stages of potential resolutions.

  • Stage 0 = initial resolution

  • Stage 1 = first dispute

  • Stage 2 = committee resolution

  • Stage 3 = second dispute

  • Stage 4 = TRUE holder resolution

  • Stage 5 = highest escalation

  • Stage 6 = Final verdict

The first dispute requires a bond equal to the bond put up by the initial resolver (e.g., ~$200). The second dispute, directed at the resolution proposed by an appointed Committee, requires a more substantial bond for escalation (e.g., $2000).

The reason a larger bond is necessary in the second dispute, is to disincentivize low-cost griefing of the protocol. As mentioned earlier, Truth Oracle expends resources most when it must respond to disputes, if the economic cost to grief the protocol is trivial, then the sustainability of the dispute resolution process could come into question. Thus, with every higher escalation, the bond required to dispute the matter increases.

In order to dispute a TRUE holder decision (escalate the matter to an Attester vote), a significant bond should be required. The idea is that the large bond would be enough to filter out illegitimate disputes that don't have strong merits, and also potentially cover the cost incurred by the protocol for processing the dispute by allowing the ability to slash the bond if the dispute is deemed malicious.

The incentive for an initial dispute can come from the bond of the original resolver, or it can be in the form of token rewards directly from the protocol. However, incentives for later stage disputes will need to be accounted for in the economics of the token because these disputes do not have opposing bonds that can be redistributed as a reward.

The incentives for disputes at various levels of the system are considered parameter specifications which should be tailored to the circumstances specific to each implementation.

Specialist Committees:

With Truth Oracle, the first stage of dispute resolution is proposed by application-specific committees that are familiar with the types of markets being resolved. These committees are selected by TRUE holders, through governance, to serve as the initial arbitrators of disputes. In theory, this committee should have the best insights to resolve nuanced disputes that arise from ambiguities rooted in market rules.

If the resolutions proposed by these committees are either inaccurate or rejected by market participants, there’s a straightforward process by which to dispute their decisions. The economic incentive for members of these committees would either come directly from the fees generated by the protocol, or from a scheduled TRUE allocation.

Members of these committees are incentivized to behave honestly because they’re selected by token holders and indirectly compensated by token holders. Any infraction on their part could result in removal and loss of future value.

What really assures the committee’s honesty, however, is the seamless ability to dispute their proposed resolutions and appeal to a judgment from a higher authority. If certain members on the committee consistently vote for bad resolutions, removing them from their position would be trivial and they wouldn’t be able to compromise the integrity of the protocol regardless.

Alternative committees

Specialist Committees do not necessarily have to be composed of a small group of people; they can be more general constructs like service providers, escrow facilitators, AI agents, etc. Indeed, it is expected that in future implementations of Truth Oracle, the role of these committees in certain applications can be entirely occupied by AI agents. Not only would agents be more responsive, they would be more cost efficient as well.

Committees can also have additional capabilities within applications. These committees may set the standards and criteria around market creation, or have the ability to incentivize certain behavior.

In general, these committees are meant to be made up of qualified specialists who provide valuable insight in the dispute resolution process, and who are very familiar with the applications they resolve disputes for.

TRUE Holders:

TRUE holders are at the heart of the optimistic oracle framework. The token is an instrument of subjectivity, governance and economic security within the protocol.

In addition to its economic utility in Truth Oracle, TRUE is also used to govern the protocol and its respective modules. It governs the construction of special committees in the protocol and also the composition of Attester Pools. Furthermore, the protocol in its entirety and all of its individual elements, are subject to TRUE holder governance.

Anybody can propose application ideas to Truth Oracle. If an application is accepted by TRUE holders and is successful at scale, the protocol will be able to self-sustain from the economic value created by offering reliable oracle services to that application. In addition, the TRUE token can serve as the governance token for some of the applications which directly benefit from its utility.

Woes of single-utility tokens

A commonly observed issue with single-utility oracle tokens is that they often struggle with voter apathy, stemming from their niche utility that appeals only to a small group of hobbyists or project insiders.

The issue is exacerbated by their vulnerability to adverse selection. Beyond the limited circle of enthusiasts and affiliates, these tokens tend to attract bad actors seeking to use them to manipulate resolutions for personal gain. The result is that single-utility tokens fail to reflect a particularly decentralized or diverse pool of subjective decision-makers, and they are casual targets of collusive behavior.

Multi-utility as a solution

The above rationale is partly why TRUE may also serve as the governance token for some of the applications that it powers.

The synergy that stems from a unified token governing both an oracle and the apps enabled by it, enhances the integrity of the protocol by broadening the potential audience that contributes economic subjectivity to the oracle. This “security snowball effect" means that the economic security of the oracle scales with the viability and success of the applications it enables.

Balancing protocol economics

An initial iteration of a Truth Oracle enabled application is a binary-outcome prediction market that has configurable resolution fees baked into each market. The fees are realized once these markets successfully settle. It is through these vessels that TRUE holders must balance the economic incentives of all actors within the protocol, including themselves, such that the various modules are economically sound and sustainable enough for the protocol to function securely and autonomously at scale.

Attester Pools:

A distinctive feature of Truth Oracle is its innovative approach to permissionlessly integrating sybil-resistant, reputation-based identities into the protocol's dispute resolution process.

With Truth Oracle, it is essential to ensure that Attester Pools are sybil-resistant. Without sybil-resistance, the identities that form Attester Pools could potentially skew or manipulate results. In the event that multiple addresses in an Attester Pool belong to one individual, the protocol’s credibility could be at risk.

Thus, some kind of governance endorsed sybil detection bounty system should be in place to incentivize the community to actively monitor the address lists in Attester Pools and ensure that they are sybil-free. These bounties could potentially even be resolved using various modules within Truth Oracle.

Proof of Reputation sybil-resistance

Sybil-resistance for address uniqueness, however, is insufficient for the purposes of this protocol by itself. There are millions of active addresses onchain everyday. Ignoring the fact that some of these addresses do not represent unique individuals, many of them would not actually be qualified to serve in Attester Pools even if they did.

The rise of DAOs and onchain apps has given birth to a diverse ecosystem of reputable and competent individuals, with onchain identities, that are qualified to serve as Attesters within Truth Oracle. These individuals (and in some cases organizations or groups) often have running track records of contributing to various protocols in the crypto community.

By way of governance, Truth Oracle specifically filters potential identity clusters in search of reputable delegates, prominent figures, and qualified professionals to serve as potential Attesters. TRUE holders have full governance controls over the admission and discharge of Attesters, and they are expected to adjust the composition of Attester Pools over time by increasing the network of quality Attesters and removing inactive ones. Governance must also monitor for potential losses of reputation among Attesters and be able to react accordingly in case the removal of one is necessary. A reference target size for an Attester Pool is between fifty and one hundred Attesters.

The sybil-resistance gadget that is being described, and employed by Truth Oracle, is called Proof of Reputation. In a Proof of Reputation based Attester Pool, candidates would be made up of founders, delegates and multi-sig signers in the Ethereum ecosystem that have a provable history of coordinating in decentralized groups to actively vote on proposals. This behavior uniquely resembles the critical thinking necessary to make subjective assessments about arbitrary disputes through a decentralized process.

“Work” associated with PoR

The viability of Truth Oracle’s PoR implementation comes from the fact that potential Attesters have professional reputations tied to their onchain identities, independent from the protocol itself. The only way to compromise such a system is by generating the human output or “work” necessary to move up the governance ranks of several high profile Ethereum ecosystem projects.

Given the PoR based nature of Attester filtering, it’s reasonable to assume that these independent governance delegates are mostly incentivized to vote honestly, to the extent that the protocol functions expectedly and that these actors engage with the Attester selection process at all.

Participation

It is not entirely obvious, however, that these individuals are even interested in resolving any of this protocol’s disputes. After all, none of these delegates have opted into any sort of arrangement with Truth Oracle, nor are they even aware of its existence at the time of this writing. To address this problem, Truth Oracle pairs incentives that encourage participation in resolutions paths with the permissionless integration of provably reputable actors.

Active Attesters:

For Truth Oracle to function or “maintain liveness,” there must be a way to guarantee that attesters are actively monitoring for disputes and prepared to resolve them. Once the protocol has identified a set of provably reputable onchain identities, it then needs to compel these addresses to join a rolling list of “registered” Attesters.

This means the protocol must compel these provably reputable individuals to engage with the oracle permissionlessly and on a regular basis. In essence, there must be an incentive for these identities to opt into this oracle system, as well as consistently engage with the protocol to maintain responsiveness.

Lottery system

To incentivize Active Attesters, the protocol will employ an epoch-based lottery system that encourages them to opt into Truth Oracle’s resolution paths. By registering for these lotteries, attesters signal to the protocol that they are active and prepared to engage with the oracle. To be eligible for the lottery, attesters must register for Attester duty for that epoch by signing a transaction indicating their intent to serve in this capacity if selected.

Every epoch, one randomly selected Active Attester wins that epoch’s prize. If no Active Attester wins, the prize is carried over to the following epoch. The time of week when the lottery draw happens could also be randomized to encourage attesters to avoid complacency when registering for the lottery (and Attester duty by extension).

Prize Adjustment

Drawing inspiration from Bitcoin's Difficulty Adjustment configuration, Truth Oracle employs a Prize Adjustment system that targets a set number of Attesters every epoch (e.g., 100). If the total number of Active Attesters in a given epoch is above or below the target, the lottery prize 'adjusts' by decreasing or increasing within predefined limits, depending on the magnitude of the deviation from the target.

If the target is 100 Active Attesters per epoch, and the number of Attesters who register for that epoch's lottery is 90, then the total prize for the following epoch would be larger (e.g., by 10%) to make it more attractive for Attesters to opt in. The prize would then continue to increase every epoch (within predefined limits) until the target is reached. This Prize Adjustment allows the protocol to dynamically respond to Attester participation rates, which may be influenced by external factors.

Selection process

In the event that a dispute is escalated to an Attester vote, the protocol will use a source of randomness to select a group of Attesters who have registered for that epoch. Attesters then vote on the outcome of the dispute and the associated slashing conditions within a designated time frame.

If a registered Attester does not respond to a dispute in time after being randomly selected, the protocol will randomly select another Attester to replace the unresponsive one. This process will repeat until enough Active Attesters have voted on the outcome of the dispute.

The number of Attesters that must vote, in order for a resolution to be final, is a governance adjustable parameter which depends on factors such as the size of the entire Attester Pool and the average number of Active Attesters per epoch.

Incentivizing Participation

Another important element is ensuring that Attesters follow through on resolving disputes after they opt into an epoch’s lottery. To ensure disputes are actually resolved, and done so in a timely manner, the protocol economics could include an incentive allocation awarded to Attesters for successfully resolving disputes. The reward could be distributed per Attester and decrease linearly with time, such that the faster each Attester votes, the greater their reward.

These measures guarantee that it is in Attesters’ best interest to respond to a dispute as promptly as possible, as well as condition them to remain alert to protocol developments in preparation for lottery results or unresolved disputes.

Collusion-Resistance

Truth Oracle assumes that the supermajority of PoR filtered Attesters are not byzantine and would not solicit bribes if selected from an Attester Pool.

The Protocol assumes that these Attesters have their reputations staked on honest behavior and that burning it would be detrimental to their positions in other projects as well as their professional relationships. This is not an unreasonable assumption, though it is a strong one, and would likely depend on the amount of value that can be economically extracted from colluding.

With the above intuition in mind, it should be noted that a well designed Proof of Reputation system should be mostly collusion-resistant by tapping into high quality Attester Pools. A high quality Attester Pool would consist of attesters with a significant onchain presence and a provable social reputation. Such a group of individuals would include: prominent founders, high ranking governance delegates, onchain legal associations, and reputable research firms, among others.

Increased cost of collusion

All else being equal, the reputations of these prominent individuals are worth significantly more than that of the average Attester, and thus, the bribe necessary to compromise them would increase the economic cost of collusion. This does, however, also raise the economic cost to incentivize these Attesters to participate in the lottery system and to vote expeditiously. Given that the more reputable an Attester is, the higher their cost of time will be, a balanced approach is warranted.

Determining the exact economic cost of collusion is non-trivial and would require extensive information about each individual attester such as their net worth, immediate circumstance, reputational value, specifics of the market being disputed, etc. However, it should be obvious that bribing an attester who has a reputation at stake, and who has been selected at random to resolve a dispute within a finite amount of time (when Attesters are incentivized to act expeditiously), is a more difficult task than outright bribing token holders in traditional optimistic oracles.

In naive token voting oracles, token holders can be bribed trustlessly with trivial implementations that draw inspiration from existing onchain vote-bribing mechanisms. Token holders also have a higher degree of anonymity than attesters do, which makes accepting bribes more viable. Although it is also possible to bribe Attesters trustlessly, the difficulty of avoiding detection and the resulting damage to their reputations, makes it significantly harder.

Resilience to byzantine attesters

The protocol is designed to be resilient in the presence of a non-trivial number of byzantine Attesters so long as that number represents a minority. The exact probability of honest resolution is outlined in the Probabilistic Guarantees section of this paper, but to build some intuition, if we assume there are 100 total Attesters, the protocol can guarantee honest resolution, with greater than 99% certainty, if less than ¼ of Attesters are byzantine.

What's notable here is that if a quarter of all Active Attesters are Byzantine, it represents a complete failure of the governance mechanism that selected these identities in the first place. It’s reasonable to assume that such a miscalculation in the Attester admission process would not occur without preemptive detection by the community or the applications making use of the oracle. Therefore, it is fair to operate under the assumption that such a high percentage of byzantine actors inside the Attester Pool could only be the result of governance and incentive failures outside the scope of this protocol.

Attester Decentralization:

Attester Pools, like many other blockchain-based systems, benefit from decentralization within their respective networks. By deriving Proof of Reputation from the governance of multiple crypto projects, Truth Oracle can increase the decentralization of its attester lists. It is important that the attester lists being used to compose PoR Attester Pools are not all from a common or small group of crypto projects. The more diverse the set of projects from which reputation is derived, the more decentralized Attester Pools will be.

Another way Attester Pools benefit from decentralization is geographically. A geographically decentralized Attester Pool offers diverse perspectives to the Intersubjective Oracle, and increases the censorship-resistance of the protocol. A globally distributed Attester Pool would pose significant challenges to physical censorship and persecution of attesters to destabilize the Oracle.

Moderators / Guardians:

A layer of protection that adds integrity to Attester votes is the Moderator / Guardian role. Moderators would typically consist of a group of qualified onchain-security and legal experts who contribute to the integrity of an Attester ruling. These individuals ensure that the vote process is collusion-free and that Attesters are aware of all relevant facts and circumstances regarding a dispute.

To improve the collusion-resistance of the protocol, Moderators would monitor onchain activity or remain alert to the community social layer in order to detect potential collusion among selected Attesters. If collusion is detected, Moderators would remove the offending Attester(s) from the vote.

Moderators are limited in how many times they’re allowed to remove Attesters during a single vote. This is because beyond a certain threshold, either the majority of Attesters are byzantine (which would conflict with this protocol's base assumptions), or Moderators are censoring and must be replaced by token holders after the vote finalizes.

This process isn’t foolproof, especially in regards to the monitoring of the social layer, but it would provide improved assurances against trustless bribes. Detecting a trustless onchain bribe is not trivial, but it is feasible and within the expected capabilities of protocol actors like Moderators.

It is also entirely possible that the Attesters voting on a dispute may be unaware of real-world evidence pertinent to the outcome. In addition to the recommendations implicit in Committee resolutions and the subjective votes of TRUE holders, it is vital to ensure that Attesters are also informed of any real-time developments related to the disputed market. A Moderator’s role within the protocol would include verifying that Attesters have access to the latest credible information about an open dispute.

Moderator sign-off mechanism

A mechanism that allows Moderators to post information, facts, or questions onchain, which can be verifiably acknowledged by Attesters before proceeding with a vote, would be an ideal implementation. Such a mechanism would involve a multi-step handshake process between Moderators and Attesters before a final verdict is made. Moderators would post facts, Attesters would respond with confirmation, and if Moderators accept the response, they would enable Attesters to vote.

Under this implementation, an Attesters would not have the ability to vote unless Moderators are satisfied that the facts have been acknowledged or the necessary questions have been answered.

Optional role

Moderators are optional within this framework because they could introduce a point of centralization in the protocol where a small group of actors are in a position to selectively censor specific Attesters that they dislike.

Additionally, provably reputable Attesters may be capable of resolving most disputes without the need for moderation. This is especially true if the market being disputed is of low economic value, such as determining whether some niche content is AI-generated or if an assertion in a social media post, that isn't of significant public interest, is accurate. In these cases, Moderators provide only minor enhancements to the oracle, and could be considered unnecessary in resolution paths.

The need for Moderators depends entirely on the type of application for which Truth Oracle is being used or the economic risks of potential collusion. If implemented correctly, the presence of Moderators or Guardians could significantly enhance the protocol’s credibility in certain applications. In such use cases, the incentives for Moderators would be very similar to those of Specialist Committees; they would be directly incentivized and governed by TRUE holders. Moderators would need ongoing incentives to serve in this capacity, as their obligations involve more overhead than those of Active Attesters.

Probabilistic guarantees:

In Truth Oracle, given that the highest arbiters of truth are the PoR Attesters, the protocol offers a set of probabilistic guarantees based on a few variables: the total number of Attesters, the number of randomly selected attesters, and the assumptions around their honesty. It turns out that these guarantees are very strong when assuming that the number of byzantine attesters are a minority.

The probability of honest resolution follows a cumulative hypergeometric probability distribution. The probability starts low when honest attesters are a small minority, and increases rapidly as the proportion of honest attesters grows, exceeding 90% as the number of honest attesters approaches a supermajority.

Below is a chart illustrating the odds of a (subjectively) honest resolution if there are 100 Active Attesters and 15 are randomly selected to resolve disputes:

As seen above, once we assume that 65% of Active Attesters are honest, the odds of 8 or more of 15 randomly selected being honest, shoots above 90%.

However, ninety percent may not be enough for a robust oracle system. If Truth Oracle wants to offer the highest guarantees to applications powered by it, the probability of honest resolution should be greater than 99%. As seen in the next graphic, that is indeed the case, even in the presence of a significant number of byzantine attesters!

The illustration below is a close-up of the chart above, focusing on the 98th to 100th percentiles. The odds of a (subjectively) honest resolution are greater than 99% when less than 25% of Active Attesters are byzantine, and greater than 99.9% when less than 20% are byzantine.

An order of magnitude increase in the number of Attesters and the sample size gives us even stronger guarantees with fewer honest actors. As the number of total Attesters and the sample size increase together, the probability of honest resolution also increases due to the law of large numbers.

There are diminishing returns to these increases, however, and the feasibility of these numbers is also questionable given that the 10x increase in total Attesters and the sample size, also likely represents a proportional increase in protocol incentives.

Note that the guarantees are much stronger with the 10x increase, and the odds of honest resolution are greater than 99% when just 62% of Attesters are honest, as pictured below:

Anecdotally, if the sample size of randomly selected Attesters more closely reflects the size of the total Attester pool, the probability of honest resolution increases as well.

The odds can also vary as the Attester numbers change relative to each other. For example, in some cases, if the sample size increases relative to the total Attester count, but the total Attester count decreases, the probability of honest resolution can also decrease.

In the example below with 50 total Attesters, the probability of honest resolution reaches 99% when less than twenty percent of Attesters are byzantine (compared to 25% in the first chart). This is despite the sample size being larger relative to the number of total Attesters compared to the first example.

With an increased sample size and the same number of total attesters in the first chart, the probability of honest resolution can be greater than 99% with as few as 69% honest attesters:

Using the same Attester count, with a reduced sample size, the probability of honest resolution reaches 99% when at least 79% of Attesters are honest:

The conclusion to be drawn here is that the tolerance for byzantine actors within Truth Oracle is quite high when the number of total Active Attesters is greater than ~50 and the sample size is not proportionally insignificant. Using the reference examples laid out above as benchmarks, a supermajority of honest Attesters in any of these cases would provide very strong dispute resolution guarantees.

Assumptions:

There are some base assumptions that Truth Oracle relies on in order to function properly, as every protocol does. For example, every blockchain operates under the assumption that the “internet exists” and that it can be used as a medium of communication for nodes on the network. Every blockchain also operates under the assumption that cryptography exists and cannot be broken. These are foundational assumptions that are reasonable to make and relying on the existence of these things is justifiable.

Aside from assumptions around cryptography and the modern internet, blockchains also operate under many other assumptions that include key components of their respective protocols. An example of this is the 51% honest majority hash rate necessary for Bitcoin to satisfy consistency and liveness, or the 2/3rds honest majority stake necessary for BFT type protocols to satisfy those same properties.

Truth Oracle relies on the following assumptions to satisfy its intended utility:

  • Majority of Attesters are honest: this protocol provides very strong probabilistic guarantees when a supermajority of Attesters are honest, but offers weak guarantees if the majority of Attesters are not honest.

  • Governance is capable of vetting onchain identities for Proof of Reputation when admitting them into Attester Pools.

  • The governance of projects where Attester Pools derive their Proof of Reputation from are not compromised or otherwise gameable.

  • In case of an escalated dispute, fewer than half of the randomly selected Attesters from an Active Attester list are byzantine.

  • A minimum threshold of Attesters are responsive: as long as a minimum threshold of Attesters are responsive to the protocol and opt into Attester duty (with less than 1/2 byzantine per random sample), the oracle will eventually reach a resolution.

  • Attester selection is truly random

  • Resolutions will be proposed when market rules are satisfied and inaccurate resolutions will be disputed autonomously.

  • The individuals composing Attester Pools are experienced in decentralized coordination and are capable of casting votes based on a set of predefined rules.

Caveats:

Even if the assumptions laid out above hold up in practice, the protocol may further operate under implementation-specific operational assumptions. Depending on the scope of power defined by TRUE holders for Specialist Committees or Moderators, the assumptions of the oracle might change.

A benefit of Truth Oracle is that once a Resolution Path has been created for a market, participants no longer need to assume the honesty of token holders (majority or otherwise), they only need to assume that fewer than a threshold of Attester Pool members are dishonest and corruptible. However, for the oracle to maintain liveness in the long run, the protocol requires that token holders are honest and able to govern the protocol effectively.

Additionally, similar to how most blockchains must ensure their economics incentivize the long term security of their respective chains, token holders must also sustain the protocol’s economics long term.

Limitations:

Truth Oracle is an Intersubjective Optimistic Oracle protocol. It is not a governance framework or a means to manage and implement protocol upgrades. Truth Oracle is a dispute resolution framework, not a managerial solution. It is an oracle solution that enables certain decentralized applications.

An example of an unsuitable use case for Truth Oracle would be governing DeFi protocols. Just as Chainlink data feeds are not used to govern protocols, Truth Oracle should not be used to govern a protocol. Even within Truth Oracle, the oracle framework itself is not used to govern the protocol, the TRUE token serves that purpose via governance strategies outside the scope of this paper.

Glossary:

Active Attester: A member of an Attester Pool who has registered for attester duty in a given epoch.

Adverse Selection: When incentives of a system attract undesirable actors.

Application: A collection of markets serving a particular audience, use case, or function.

Attester: An individual that subjectively votes on arbitrary outcomes. In Truth Oracle this is someone with a provable reputation.

Attester Pool: A list of individuals with proven reputations who can be selected to resolve escalated disputes.

Byzantine: Actors who may behave maliciously or unpredictably.

Collusion Resistance: The ability to resist coordinated efforts by multiple parties to manipulate outcomes.

Disputer: Actors who challenge proposed resolutions by putting up a bond.

DAOs: Networks of distributed groups of individuals coordinating together.

Epoch: Fixed durations within the protocol.

Futarchy: A system of governance where prediction markets are used for decision making.

Governance: The system by which the protocol makes decisions.

Intersubjective: A collection of unique subjective perspectives.

Liveness: The ability of the protocol to continue functioning and remain responsive over time.

Lottery System: A mechanism used to incentivize attester participation by offering rewards to randomly selected Active Attesters.

Market: In the context of this paper, an object that is resolvable based on a set of predefined rules (e.g., claim, question, attestation).

Moderator/Guardian: Optional roles filled by security and legal experts to ensure the integrity of the Attester voting process.

Multi-sig: A Multiple Signature wallet which is used to allow multiple parties to control an address.

Onchain: Activity that occurs directly on blockchains.

Optimistic Oracle: A system where proposed outcomes are considered valid unless challenged within a specified time frame.

Oracle: A system that enables offchain data to be utilized onchain.

Proof of Reputation (PoR): A sybil-resistance mechanism that leverages the established reputations of individuals in the crypto community.

Resolution Path: The methodology for escalating disputes within a specific market or application.

Resolver: Actors who propose resolutions for unresolved markets by putting up a bond.

Slashing Conditions: Penalties applied to bonds put up by resolvers or disputers in case of dishonest behavior.

Specialist Committee: A group appointed by TRUE holders to serve as initial arbiters for disputes in specific applications.

Sybil-resistant: Resistant to attacks where one entity pretends to be multiple entities to gain disproportionate influence.

TRUE: The governance token of Truth Oracle, used for economic security, alignment, and decision-making within the system.

Truth Oracle: An intersubjective optimistic oracle protocol for decentralized micro-coordination and dispute resolution.

Last updated