Ethereum and Oracles

The Rise of Oracles in Smart Contracts

Smart contracts have revolutionized the way we think about decentralized applications, but one of the most significant challenges they face is the need for external data and computation. This is where oracles come in – specialized entities that can provide the necessary information and computation to make smart contracts more efficient and effective.

Oracles Are Not Always Better

One of the most common misconceptions about oracles is that they are always more efficient than blockchain-based computation. However, this is not always the case. In fact, there are many scenarios where blockchain-based computation is more efficient and secure.

For example, consider a contract that requires a running record of a withdrawal limit. In this case, the contract can maintain its own state and update it as necessary, without the need for an oracle. This approach is more efficient and secure, as it eliminates the need for external data and computation.

The Primary Feature of Ethereum

The primary feature of Ethereum is not Turing-completeness, but rather state. Ethereum accounts can contain not just a balance and code, but also arbitrary data, allowing for multi-step contracts, long-running contracts, and non-financial blockchain-based applications to emerge.

For example, consider a contract that requires a running record of a withdrawal limit. In this case, the contract can maintain its own state and update it as necessary, without the need for an oracle. This approach is more efficient and secure, as it eliminates the need for external data and computation.

When Oracles Make Sense

Despite the limitations of oracles, there are many scenarios where they make sense. For example, in cases where external data is required, such as in financial contracts that use the price of the US dollar, oracles are necessary.

Another important case is smart contracts that actually are very hard to evaluate, such as in cloud computing applications where verifying computations was done legitimately is not a task that the Ethereum blockchain can cheaply handle.

How Can Oracles and Ethereum Work Together

The question is, are the two paradigms of total on-chain and partial on-chain purely competitive? The answer is, as it turns out, no. To further this point, here are a few particular examples:

SchellingCoin

SchellingCoin is a proof-of-concept that shows how we can create a decentralized oracle protocol that is incentive-compatible. The protocol has a two-step commitment protocol so that oracles do not initially know what each other's answers are, and then at the end have an Ethereum contract reward those oracles that are closest to the median.

Verifiable Computation Oracles

When the oracles in question are executing moderately computationally intensive code, then we can actually go beyond the admittedly flaky and untested economics of the SchellingCoin/TruthCoin protocols. The idea is as follows. By default, we have M of N oracles running the code and providing their votes on the answers. However, when an oracle is perceived to vote incorrectly, that oracle can be "challenged". At that point, the oracle must provide the code to the blockchain, the blockchain checks the code against a pre-provided hash and runs the code itself, and sees if the result matches.

Signature Batching

One of the problems that I pointed out with the multisig oracle approach above is signature bloat: if you have three oracles signing everything, then that's 195 extra bytes in the blockchain and three expensive verification operations per transaction. However, with Ethereum we can be somewhat more clever - we can come up with a specialized "oracle contract", to which oracles can submit a single transaction with a single signature with a large number of votes batched together.

Blockchain-Based Auditing

The concept of oracle-based computation can actually go much further than the "Bitcoin multisig oracle" (or, for that matter, Ethereum multisig oracle) idea. The extreme is an approach where oracles also decide the one thing that the Bitcoin-based schemes still leave the blockchain to decide: the order of transactions. If we abandon this requirement, then it is possible to achieve much higher degrees of efficiency by having an oracle maintain a centralized database of transactions and state as they come, providing a signed record of each new balance sheet as a transaction is applied.

Verifiable Secure Multiparty Computation

In the case where you are using oracles specifically for the purpose of maintaining private data, you can set up a protocol where the oracles securely choose a new secret key using multiparty random number generation every 24 hours, sign a message with the old key to prove to the world that the new key has authority, and then have to submit all of the computations that they made using the old key to the Ethereum blockchain for verification.

Conclusion

The primary raison d'être of Ethereum is not just to serve as a smart contract engine; it is more generally to serve as a world-wide trust-free decentralized computer, albeit with the disadvantages that it can hold no secrets and it is about ten thousand times slower than a traditional machine. The work in developing cryptoeconomic protocols to ensure that ordinary people have access to reliable, trustworthy and efficient markets and institutions is not nearly done, and the most exciting end-user-centric innovation is likely what will be built on top.

Source: https://blog.ethereum.org/en/2014/07/22/ethereum-and-oracles