Truth Oracle is a decentralized source of Truth that verifies subjective real-world information onchain.

It enables the provision of Truth to applications in ways that were previously unviable.

If that sounds interesting, you will love this thread.

Let's break it down 🧵
1/n

The Oracle utilizes the power of intersubjective Truth to resolve disputes with strong, yet practical guarantees.

It's the first optimistic oracle that can scale to meet the demands of micro-applications.

It can be used to optimistically verify virtually any offchain data.

2/n

If you're interested in the provable details, I recommend reading the whitepaper.

But for a high level understanding of the protocol, this thread should build the necessary intuition.

Link to paper -  https://truth-oracle.gitbook.io/truth-oracle

Now, let's dive in 👇

So, what are the current standards for optimistic oracles?

Most modern day designs reduce down to token voting and economic security.

Resolutions are determined based on the outcome with the greatest economic weight behind it.

4/n

Typically, claims are optimistically considered valid unless otherwise disputed within a challenge window.

When a dispute is filled, token holders have the first and last say.

When voting ends, the option with the most economic support is set as the final result.

5/n

I know what else you're thinking though; "wait, in that case can't anyone just use raw capital to manipulate outcomes?"

And that is a very good question, to which the answer is; yes, naive token voting is definitely vulnerable to economic manipulation.

7/n

In fact, basically every app that uses optimistic oracles has some multi-sig backstop. Generally, this multi-sig belongs to the team behind the project.

Usually I would criticize such a construct, but in this case these projects have no other choice.

8/n

Leaving their apps vulnerable to manipulation is simply not a viable solution.

However they also want to be credibly neutral in their application's resolution process, so they end up outsourcing the service to crude token-voting providers anyway, despite the issues.

9/n

The limitations I've discussed so far are what motivated Truth Oracle.

This protocol aims for a maximally credible dispute resolution process which is not limited to the subjectivity of a single token holder base or a small group of actors.

10/n

Instead, Truth Oracle permits multi-stage escalation with proposed resolutions from independent, interdisciplinary actors at every step.

The nature of this process is what makes this oracle intersubjective, as apposed to typical single layer dispute resolution frameworks.

11/n

One of the innovative aspects of this protocol is its permissionless approach to Proof of Reputation.

Truth Oracle employs a lottery system to encourage provably reputable identities within the Ethereum ecosystem, who are recognized by the protocol, to serve as Attesters.

12/n

In a typical Truth Oracle implementation, there are 3 separate arbitrators, each proposing resolutions in different stages within a resolution path.

They operate under the following ascending hierarchy:
- Committees
- Token Holders
- Attesters

illustration of the process:

Resolvers/Claimants:

Like with traditional optimistic oracles, Truth Oracle allows permissionless resolution.

Anyone can put up a bond and submit a claim or resolution (depending on the context of the application).

In the ideal case a claim or resolution is optimistically assumed to be valid unless it is disputed.

If a proposed resolution is clearly erroneous, the resolvers' bond can be slashed.

Given how trivial it is to slash the bond of the resolver, in most cases, there shouldn't be too many malicious resolutions.

Once the optimistic challenge window passes, the resolver receives a modest reward.

14/n

Disputers:

Disputers are essentially the same group of actors as initial resolvers. They monitor the outcome of markets and propose resolutions, or dispute inaccurate ones.

They can challenge in three different stages during designated challenge windows, by putting up larger bonds with each escalation.

It starts with a bond equal to that of the initial resolvers/claimants, and increases every time a higher authority needs to weigh in.

This design is meant to ensure the cost of griefing the protocol and forcing it to expend resources is non-trivial.

15/n

Committees:

Responsible for the majority of dispute resolutions, Committees serve as the protocol's initial arbiters.

If an optimistically proposed resolution is disputed, Specialist Committees appointed via governance, resolve the dispute.

These Committees are made up of individuals who specialize in specific types of applications. Ideally they're closely familiar with the apps in question and potentially have other functions within their respective apps as well.

16/n

Something that's relatively common when verifying subjective information is the occurrence of an unexpected outcome.

These ambiguous outcomes can potentially be unexpected for all the participants involved, including the market creators, or whoever defined the rules.

It is in these cases that Committees offer valuable insights to what a resolution "should be" in the context of the market and its respective rules.

If participants refute the resolutions proposed by these Specialists, they can simply put up a bond and file a challenge, escalating the matter to token holders.

17/n

TRUE - The Oracle token:

TRUE serves several functions within the protocol. It provides economic alignment to the oracle, resolves mid-level disputes, governs all the components of the protocol, and plays an important role in maintaining the liveness of the oracle.

Where Specialist Committees offer deep insights with their proposed resolutions, TRUE holders contribute economic subjectivity to the dispute resolution process.

If a Committee proposed resolution is challenged, TRUE holders are next in line to arbitrate.

However, in a full Truth Oracle implementation, their decisions are not final, they simply offer another layer of perspective to the intersubjective framework.

If the community takes issue with a TRUE holder proposed resolution, they can dispute this decision by putting up a substantial bond.

18/n

It should be noted, though, that TRUE holders are incentivized to act honestly because they don't make the final decision in resolution paths anyway.

A maliciously proposed outcome would only damage the Oracle's credibility and the value of the token by extension, making it a pointless endeavour.

For this same reason, the token is not a likely target of bad actors seeking to use it for manipulation, because even though it governs the protocol, it can't obstruct a market or claim once they're live.

There is a lot more to be said about the token and I encourage those of you who are curious to read the white paper section on TRUE, but for now, I'll have to leave the rest for another thread.

Below is a graphic illustrating TRUE's various utilities:

19/n

Attesters and The Power of Reputation:

The protocol's last line of defence in the search for truth is Proof of Reputation based Attesting.

Attesters are provably reputable, sybil-resistant identities in the Eth ecosystem that have earned their respective reputations externally from Truth Oracle.

This includes reputable founders, high-ranking governance delegates, and other prominent figures, who have staked reputations and identifiable onchain addresses (this can expand to reputable individuals outside of crypto eventually as well).

20/n

When disputes escalate beyond a token holder decision, Attesters must submit a final verdict. Their role is to provide credibly neutral perspectives to the final layer of this intersubjective protocol.

This human element, which is backed by reputation, adds a layer of nuanced decision making that algorithms simply cannot match.

It enables the discretion necessary to verify virtually any subjective real-world information onchain.

This human element is especially important with the rise of AI.

In the future, we expect that the majority of lower level disputes and resolutions will be carried out entirely by AI agents, with humans optimistically verifying these inputs and serving as the final arbiters of Truth.

21/n

Attester Pools and Lottery system:

The Attester selection process can be carried out entirely permissionlessly by Truth Oracle governance.

The protocol will regularly recognize provably reputable individuals with onchain identities, and admits them into attester pools.

Once candidates are recognized, governance simply adds the address associated with the candidate to an attester pool.

There is no need to consult the PoR Attester candidates directly, although there are no restrictions against this. The protocol simply includes the Attesters' known addresses in attester pools, and the oracle's game theory handles the rest.

The obvious next question is why these individuals would have any interest in participating in this protocol at all?

Who's to say they want to take time out of their day to verify random information about a dispute?

To address this variable, Truth Oracle runs a lottery system which incentivizes attesters participation.

22/n

After a candidate is admitted to an Attester Pool, they are free to opt into this lottery if they like. If not they can pretend it doesn't exist.

The lottery runs every epoch and is similar to staking rewards in PoS blockchains. Every epoch the lottery spontaneously resolves and randomly rewards a prize to a registered Attester.

After that, a new epoch begins and all the Attesters who opted in mid way through the previous epoch are registered for the new epoch.

The lottery recurs every epoch, but the prize for the winning Attester adjusts every round based on a target number of Active Attesters (attesters that have opted in).

This concept draws inspiration from bitcoin's difficulty adjustment, which adjusts the difficulty of solving the cryptographic hash function necessary to mine a block, by targeting a set number of blocks produced per epoch.

23/n

The game theory of the lottery system is designed to ensure that there is a certain level of participation among attesters pools. The Prize Adjustment target is a configurable parameter controlled by governance.

By registering for the lottery, an Attester is signalling to the protocol that they're interested in resolving escalated disputes, and should one occur, they are willing to respond.

Attesters that are not responsive to disputes can be removed over time if they miss too many.

In general, governance is expected to regularly admit new candidates with provable reputations and ensure that attester pools are large enough to make the Prize Adjustment target feasible.

If a dispute is escalated, the protocol randomly selects a jury made up of Attesters who've registered for that epoch to resolve the matter.

This randomly selected jury is responsible for delivering the final verdict in a resolution path. If an attester is not responsive after some time, another one is selected to replace them.

24/n

There is also a direct incentive to resolve these disputes, should they happen. The incentive is designed such that the quicker an Attester votes, the larger the reward.

For more details about this process, I recommend examining the white paper where there are a lot more details about these mechanisms, or tune into some more threads on this topic!

25/n

Probabilistic Guarantees:

In Truth Oracle, given that the highest arbiters of truth are PoR Attesters, the protocol offers a set of probabilistic guarantees based on a few variables:

- total number of Attesters
- number of randomly selected attesters
- percentage honest

The probability of a (subjectively) honest resolution follows a cumulative hypergeometric probability distribution.

The odds start low when honest attesters are a minority, and increase rapidly as the proportion of honest attesters grows, providing very strong guarantees as the % honest approaches a supermajority.

26/n

Below are charts illustrating the odds of honest resolution if there are 100 Active Attesters with 15 randomly selected to resolve disputes:

27/n

The math behind these guarantees is particularly compelling.

With the above referenced values in mind, the oracle offers >99% likelihood of honest resolution when fewer than 1/4th of all Active Attesters are byzantine!

That number is impressive because the idea that a quarter of Attesters are byzantine would represent a catastrophic failure on the part of governance, which is supposed to specifically filter attester candidates for Proof of Reputation!

28/n

That's it for now friends, more to come!

If you got this far you probably have a decent intuition of the mechanisms in Truth Oracle.

If you like the idea, please recast the first cast in the thread for visibility🙏

btw, here's the hello world moment:

That's it for now friends, more to come!

If you got this far you probably have a decent intuition of the mechanisms in Truth Oracle.

If you like the idea, please recast the first cast in the thread for visibility🙏

btw, here's the hello world moment:

https://x.com/llamaonthebrink/status/1839536678005792844

Naive dispute resolution schemes work in preventing spam and griefing attacks because when outcomes are objectively verifiable, the narrow subjectivity of single-step arbitration doesn't cause too many problems.

But that design falls apart the moment outcomes are ambiguous.

6/n