UMIP-188: Increase Final Fee

Summary

This proposal is to increase the finalFee (a component of proposal and disputing bonds) from $250 to $500. The number of dispute resolution votes are currently a strain on gas rebate costs and voter time and will become prohibitive as the number of requests continue to increase. Increasing the finalFee will increase the cost of incorrect proposals and disputes and therefore reduce oracle dispute rates and dispute resolution votes.

Motivation

Increase in Disputes

While we have seen a downwards trend in the oracle dispute rate in the last year, the increased number of markets has resulted in the number of disputes increasing roughly 3x in the last year.

Unnecessary Disputes

The far majority of disputes are not necessary to correct market resolution, but rather are due to early proposals or incorrect disputes. The values shown below for the month of May are:

  • Successful Dispute resolved too early/P4 (light blue) = 0.9%
  • Successful Dispute resolved Non-P4 (navy) = 0.1%
  • Unsuccessful Dispute (green) = 0.8%
  • Total Dispute rate = 1.8%

UMA’s voting on dispute resolution also shows that 98% of votes reach >90% consensus. This shows that the far majority of disputes are not actually contentious to UMA voters. These obviously incorrect proposals and disputes are largely made due to:

  • Proposers racing to propose markets first resulting in premature proposals
  • User error due to accidents or new proposers
  • Prediction market traders making proposals or disputes to affect betting odds on the market

Cost of Disputes

The high number of disputes puts a higher burden on voters who have to evaluate on average 6 votes per voting cycle and sometimes more than 20. This results in voters spending less time per vote which can lead to a reduced voting accuracy.

Risk Labs also reimburses gas spent by voters resolving disputes and has historically been reimbursed for these costs by UMA DAO. The average 2025 vote costs $584 in gas rebates. This results in average monthly gas rebates of $45k and in times of high gas and Ethereum prices has cost Risk Labs up to $120k in one month.

The above negative effects are challenging right now, but scale linearly with the number of requests and will become prohibitive as UMA scales in the future.

Rationale

UMA will need to address the scaling problems identified above with multiple solutions that reduce the number of disputes voted on by UMA stakers. However, raising bonds is essential to the quality of dispute resolution and the economic feasibility and scalability of UMA. We expect that raising bonds will reduce the number of unnecessary disputes identified in the Motivation section.

The benefits of raising bonds are:

  • The cost of incorrect proposals or disputes intended to manipulate prediction markets is increased.
  • Proposers should be more cautious to wait until the trigger event has passed which will result in less unnecessary disputes about trigger event timing.
  • Less unnecessary disputes result in faster resolution (undisputed markets resolve in 2 hours, but disputed markets resolve in 28-96 hours).
  • Reduced spam disputes. We’ve seen 5 disputes recently that are on poorly formed requests.

Potential downsides of raising bonds are:

  • It increases the risk (and reward of disputing). As shown above the far majority of disputes are:
    • Successfully resolved P4: obvious P4s will still be high EV disputes and marginal P4s (a sports game resolved correctly but 3 minutes too early with a 2 hour challenge period) are inconsequential to market resolution
    • Unsuccessful: it would be beneficial to everyone if these disputes did not happen
  • It could increase the time from a market trigger event to proposal
    • Currently we see objective requests proposed very quickly (hard to measure, but this is usually within minutes for sports games) while the challenge period is currently 2 hours. This means that the time from trigger event to proposal is a very small portion of the overall time to market resolution. Therefore, we strongly expect that increasing bonds for objective markets will not noticeably affect time to resolution, especially for objective markets like sports games or prices. Further, increasing the bonds makes it safer to reduce the challenge period as the cost of an incorrect proposal is higher and there is higher incentive to monitor proposals.

Implementation

First, an explanation of how bonds work on UMA’s currently use optimistic oracles.

Optimistic Oracle V2:
The “totalBond” that proposers and disputers must commit is a sum of two parts:

  • The “finalFee” which is set by UMA governance, which is currently 250 USDC.
  • The “bond” which is set by the integrator, typically varies from 100 to 500 USDC.

In the case of disputes:

  • The losing side loses their “totalBond”
  • The winning side receives their “totalBond” back plus half of the loser’s bond
  • UMA DAO receives the loser’s “finalFee” and half of their bond.

Optimistic Oracle V3:

The “bond” is set by the integration. The minimum “bond” is two times the “finalFee”.

In the case of disputes:

  • The losing side loses their “bond”
  • The winning side receives their “bond” back plus half of the loser’s “bond”
  • UMA DAO receives the loser’s “finalFee”.

The final fee is set for each whitelisted bond currency on every supported chain. Final fees are currently set to be $250 for stablecoins and close to $250 for non-stablecoins. This proposal is to double the final fee for commonly used whitelisted bond tokens on commonly used chains to $500.

If passed via onchain vote, the revised final fees would only affect requests and assertions made after onchain execution of the proposal. The UMA core team will also coordinate with integrations to prepare them for the change and advise if they should adjust their bonds.

1 Like

P1: No

Reasons:

  1. The increase in early proposals is not due to the finalFee being too low. If you review the Polymarket Discord, it’s clear that proposers are enticed by the “free” $5 reward. The easy accessibility of the “propose resolution” button also contributes to this issue. Making the proposal process more complex or the button harder to find would likely reduce early proposals more effectively than increasing the finalFee. Raising the finalFee before addressing these two issues is premature and too extreme of a measure. If early proposal frequency remains high after making these changes, we can further explore this discussion.
  2. Hiding the propose resolution button makes sense since most proposals are now submitted by automated bots and increasingly by the AI-based OptimisticTruthBot. By hiding the propose resolution button, less human users would be inclined to propose early, additionally while being unaware that the AI is actively monitoring and responding. Furthermore, most P4s result from bot errors or first-time proposers who often become discouraged after losing their bond. This is exacerbated by bad actors in Polymarket Discord or market comment sections encouraging others to propose or baiting them into thinking a market can be proposed. Since the mechanisms and reward structure already function as a public good, reducing low-effort proposals through interface changes is a more balanced and effective approach than raising the finalFee.
  3. A more urgent issue is the misalignment between market volume and totalBond requirements. For example, sports markets with $2 million in volume may only require a $350 bond, while mention markets with $100 volume may require $750. Most P4 proposals and disputes are in sports, crypto price, and weather markets.
  4. This proposal primarily serves to offset dispute costs reimbursements for Risk Labs rather than address the root causes of disputed proposals. It relies on disincentivization through higher proposal fees but does little to prevent early proposals.

Alternative Solution (besides making the proposal button harder to find):

Require the finalFee to be paid in a different token than the bond. For example, require the finalFee in UMA while keeping the bond in USDC. Users would need to submit 250 UMA + 500 USDC to propose and receive their reward in USDC.

Benefits:

  1. Adds utility to the UMA token by making it essential for both voting and proposing.
  2. Introduces friction by requiring users to acquire UMA, reducing impulsive proposals.
  3. Maintains UMA’s relevance as dApps begin customizing their proposal and resolution tokens.

Concerns:

  1. UMA’s price fluctuates, so the UMA-denominated finalFee would need dynamic pricing - updated at proposal time, daily, or weekly.
  2. This may complicate things for veteran proposers. However, most proposals are automated or submitted by users familiar with UMA, so this is a minor concern.

A Snapshot vote would help gauge UMA holder sentiment. However, proposal creation is currently restricted to space members. UMA admins should address this.

3 Likes

Also, to address another common cause of P4 proposals, such as misclicks followed by self-disputes, add the option for proposers to self-dispute within a defined timeframe, provided no external dispute has occurred. This mechanism allows proposers to acknowledge and correct accidental or incorrect submissions without triggering the 48 hour voting cycle. This is not a retraction. The proposer would still incur the standard penalty. However, if the dispute is initiated by the same address, the proposal would bypass the voting process. This maintains accountability while reducing avoidable delays and improving resolution efficiency.

4 Likes

Please check out my proposal at Reducing Proposal Disputes Through a Dual-Token Proposal Structure

If the goal is to drive more adoption of the oracle, this proposal is not the right approach. It risks making users more complacent with proposals that are slightly incorrect or subjective, rather than encouraging active dispute and refinement.

UMA is designed to handle subjective truth. That inherently means some claims will be wrong, or at least debatable. Disputes are how the system works through those differences. Making disputes harder to trigger weakens UMA’s ability to uncover what’s actually true. Subjectivity is tricky and asserters have a tough job already. Things are subjective - both asserters and disputers know that going in. It’s risky business.

If the main issue is too many early proposal disputes, why not add logic that penalizes those more heavily? For example, if a dispute is deemed too early, the disputer loses $500 — but for other types of incorrect disputes, the penalty could be $250.

I’m considering building on UMA with AI-generated assertions. AI will sometimes be wrong — and that’s fine. I want a dispute mechanism to correct those cases. I’m comfortable risking $250 per assertion to uncover the truth. But if the cost is $500 each time, I’d explore alternative systems like EigenLayer that are building lower-cost dispute resolution.

This proposal is too focused on existing UMA use cases, without considering future applications where subjectivity plays an even larger role. This change could discourage builders from choosing UMA for those more subjective use cases.

The UMA team just said things are going to be getting cheaper and then turns around increasing the price to use the oracle…

https://x.com/hal2001/status/1930212204437217591

This proposal risks discouraging disputes and letting incorrect or subjective proposals go unchallenged while hindering more builders from choosing UMA as their infra provider.

1 Like

@dynosawr: Thanks for your well thought out comments!

I think you bring up good ways to decrease disputes aside from the raising bonds.

#1 New proposers do have a very high incorrect proposal rate, especially P4s. We added a warning modal when proposing to try to prevent this. Finding more ways to reduce incorrect newbie proposals is important, but I would like to avoid the trade-off of making UX worse as I believe manual proposing is still a necessary alternative on many markets.


Polymarket and UMA have also been working together to improve their rules (definition of trigger events, resolution sources, etc) to prevent disputes. These improvements have led to a decrease in the dispute rate, but as shown in the proposal the increase in markets has outpaced the decrease in dispute rate. We will continue to look for good ways to reduce incorrect newbie proposals and other sources of unnecessary disputes, but I don’t think this will reduce the dispute rate enough to offset the expected 2-3x in requests over the next year, let alone our goal of scaling requests 10x.

#3. Since the finalFee is fixed, the totalBond is in the control of the integrator (e.g. Polymarket). This is something we’ve discussed with the Polymarket team and they are looking at improving this.

#4. Doubling the final fee will partially offset voting costs, but when you consider the gas costs plus the time and effort for voters (reflected in UMA token emissions), a $500 final fee is still very far from fully offsetting the costs to UMA as a whole.
I agree that I want to improve some of the other root causes of disputes that you mentioned above. However, I view the economic incentives and disincentives of bonds as root causes of disputes as well.

Re: self disputes cancelling out a proposal rather than going to a dispute: I think this is a really interesting idea! It would require a new oracle smart contract which requires dev time, auditing, new contract deployments for UMA AND our integrations which is not easy, cheap or fast. However, I will put this idea on the list of things to consider in our future oracle contracts.

2 Likes

P1 No

This would result in a substantial increase of the cost/risk per proposal (unless of course Polymarket does not agree to increase rewards proportionally, in which case it’d be a whole different story).
As unfortunately UMA cannot directly address Polymarket’s incoherent bond pricing (see dynosawr’s examples), this proposal would result in a lot of markets sitting unproposed for days and weeks, given that the already high risk/reward ratio would skyrocket.
If anything, finalFee should be lower fo many markets, unless UMA is willing and able to implement dynamic pricing, which does not sound likely.

Furthermore, I disagree with disputes being excessive right now, as most of them are human errors or bot mistakes, neither of which would be greatly impacted by this proposal.

PS: Self disputes not going to a vote and just resolving P4 is a very good idea, I approve.
(I know that’d be hard to do, but it would be great to implement a short “grace period” where the proposer can withdraw a mistaken proposal for a small fee rather than losing the whole bond.)

3 Likes

Thanks for you consideration and thoughtful feedback. I’m interested to hear about your AI use case if you want to reach out over Discord.

If the main issue is too many early proposal disputes, why not add logic that penalizes those more heavily?

This is an interesting idea that would require smart contract changes. We can consider this for future versions of the oracle.

This proposal risks discouraging disputes and letting incorrect or subjective proposals go unchallenged while hindering more builders from choosing UMA as their infra provider.

Higher bonds increases the cost of making a dispute which could be a barrier, but it also increases the reward for a correct dispute. So increasing the bond only lowers the expected value of disputes that are likely wrong. Raising the bond also increases the cost of making an incorrect proposal. So overall higher bonds should increase the confidence in an oracle result, not decrease it.

The UMA team just said things are going to be getting cheaper and then turns around increasing the price to use the oracle…

A small distinction here, but the finalFee is not a direct cost to using the oracle. It does increase the capital required to propose or assert and increases the risk of an incorrect proposal.

We do want to offer cheap resolution to serve low value, high volume use cases. However, human dispute resolution has a capacity limit and real costs.

Re: capacity: right now voters vote on up to ~25 votes in one voting cycle. I consider this close to a theoretical voter maximum load after which I think voter quantity or quality of votes will start to decrease. This means for UMA to scale we need to lower the dispute rate (higher bonds helps with this), but this will only help so much as requests scale (as shown in the data above). Further to scale significantly and serve lower value use cases we will need to develop cheap and scalable lower level dispute resolution (likely AI), that can escalate in rare cases to human voters. This is the type of cheap and scalable dispute resolution we are envisioning in the research you linked to.

Re: cost: disputes require UMA to offer inflationary rewards so around 300 voters will stake their capital, face the risk of slashing, and show up on short notice to do research on the dispute and vote. This costing $250, or even $500 is far below the actual cost.

So I think the cost of human dispute resolution will need to increase over time, but I do think your pushback is a good argument for delaying this to focus on growth.

1 Like

P1 - No

While this proposal has a laudable goal, it is misguided and could prove very problematic for the continued integrity of the oracle. It’s important to note that the reason the number of disputes has gone up is not because people are bringing more frivolous disputes or proposals. Instead, it is an inevitable byproduct of the fact that more users are using Polymarket and Polymarket has more markets. There are a number of reasons why we shouldn’t increase the final fee.

They’ve already been allegations that rich people control the resolutions, and that whales in the UMA system are the only one that have a say. By doubling the final fee this would make it harder for common users of polymarket to make proposals or disputes. This is a problem as one of the main benefits of polymarket using UMA is that anyone can propose or dispute a market and get a fair resolution of their dispute. By significantly limiting the people who can propose and dispute the resolution of markets to those who have $1000 lying around this removes the ability for most people to dispute or propose markets that they want resolved. All people being able to resolve or propose markets that they want is a vital aspect to polymarket, using UMA and making their markets fair. For example, often times when people complain about markets not being resolved in the comments on the polymarket website, I often respond with a comment to the effect of “anyone can propose a market. If you are that confident in the resolution of the market, you should propose it” however if the bond amount on most markets is increased to $1000 this feels like an ineffective response that ignores the financial constraints that most people have.

Secondly, this could make it longer before proposals can go through given that it’s significantly limits the pool of those who can make disputes. Already this pool is somewhat limited. Given that you need $750 to make a proposal and most poly market users are not willing to put up that money this could significantly limit the number of people who will propose markets which fundamentally will make proposals, and therefore resolutions slower, thereby hindering the usability of our partners market, especially with polymarket

Third, If our goal is “Truth. Onchain.” as we seem to be saying everywhere, a move that makes it two times as hard to dispute things that are untrue will only exacerbate the already existing problem of letting controversial disputes go through when they should be referred to a public vote. These disputes are needed to maintain the integrity of the Oracle, especially now that Polymarket is the official prediction market partner of X.

This problem needs a more long-term solution. Several other solutions have been proposed and well I understand that they require more effort. I think that changing this will have a fundamentally negative effect even if we later implement some of these solutions a few of the solutions that I have heard are the following:

  • Create a system where people are allowed to roll back disputes or proposals if they make a mistake
  • If a market is both disputed and proposed by the same person, the dispute should automatically prevail and should not go to a vote

Additionally, a few proposals that I don’t believe have been suggested yet:

  • One proposal is only allowing people who have been in the community for a certain amount of time to make proposals. This would get rid of proposals that are made without understanding the community and our rules.
  • Another proposal would be to allow the submission of potential market resolutions to the verification team for them to give an "advisory opinion,” but would help guide proposers
  • Make it so that either a proposal or dispute that is clearly incorrect. (some standard such as the proposal reaching less than a 0.75 or 0.5% of the vote could be used) would require the owner of the wallet to either pay the gas costs for their proposal or dispute or not be allowed to propose or dispute again.

@lee One final question: the explanation of the voting procedure says that not only does the final fee get paid to UMA but additionally, half of the bond also goes to the UMA store. Is this money able to be used to pay for gas fees? If so, shouldn’t that make up for the average gas fee for most questions?

In conclusion, you should vote P1 - No