Existing oracle alternatives for getting a price quickly are typically m of n approaches like Chainlink or Pyth. Slower oracle designs that rely on escalation and forking in the style of Augur are not covered because they are not designed for getting prices quickly, and to the extent financial engineering is used to overcome the latency issue you are stuck with the cost of capital internally over the finalization time of the oracle.With respect to m of n oracles, the openOracle approach has several key advantages:
Security
openOracle does not rely on trusting m of n whitelisted and kyc’d participants. The trust model is based on 1 of n participants in the oracle network acting in their own self interest. Our design is a proposed solution to the long-lived oracle problem. It can scale to any amount of money at stake. Dishonest participants can lose an uncapped amount of money, and the best of the network accumulate the most resources. Higher security allows us to scale economically. With m of n oracles, on the other hand, there exists no market mechanism to enforce the cost of corruption is greater than the profit from corruption.
Speed
In m of n oracles like Pyth and Chainlink, the price is only valid once there are m signers. This means they can only be as fast as the slowest of the m signers. In openOracle, our reporting speed is based on the fastest of all n participants, which is a permissionless set. The fastest network participants accumulate the most resources and slow reporters lose money. In m of n setups, there are not brutal incentives punishing latency, and there is a fundamental constraint on speed because of the m signers requirement + consensus latency overhead of aggregating the signatures. While the openOracle design includes some request latency via the settlementTime, the price at time of settlement is temporally as close as possible to the true price at that time.
Accuracy
By having thousandths of a basis point precision in our internal oracle swap fee parameters, we can have arbitrarily high precision on our pricing. While precision alone is not enough to guarantee accuracy, we are the most accurate price source on chain assuming volatility is not extremely high and the settlement time is not too long, from our work on honest dispute barriers in the Other Considerations page. Chainlink for example has a 0.15% deviation threshold on ETH-USD on Base, meaning the price can be up to 0.15% off continuously.Pyth includes an abstract concept of spread in their confidence reporting which applications are advised to account for in execution costs. Pyth confidence tends to be well in excess of the true lowest spread on spot CEX order books (for example with ETH price of $3714 currently there is confidence of $1.5, or about 0.04%). Pyth confidence depends in part on how much the quoters agree with each other so they are fundamentally restricted by the m of n design and latency mismatch.This level of pricing accuracy also lets us query the oracle for a price then offer a swap using that price. Alternatively, you can offer a swap executed later by the oracle. In this model, a matcher can put up liquidity in exchange for a fee. Next, the oracle game is played with the swap executed at settled oracle price. Either way you go, it is typically much cheaper than many places on-chain like Uniswap or Aerodrome. The concentrated liquidity Aerodrome WETH/USDC pool on Base uses a dynamic fee but its lowest possible fee is ~0.033%, and we are doing swaps in production today for much cheaper than that - in calm markets, we can get sub-basis-point execution. Even in volatile markets, our mean swap execution is still very competitive and fully on-chain unlike CoWswap; however, like CoWswap, there is some latency to the trade completing (stochastic with ~10-15 second average, ~95% within ~45 seconds).One of the drags on openOracle-based linear execution (like swaps or perps) is manipulation to bias the settled price. The strongest way to manipulate seems to be to report true prices but simply delay the oracle when the settled price is not in your favor. Assuming a reasonable multiplier and initial liquidity (~1.3x, 10% of notional), the extraction seems to be in the ballpark of 20-30% of settlement time volatility with a completely passive counterparty. We dive into this in the Other Considerations section “A stronger form of manipulation”.