Our previous thread explained what blockchain oracles are and why they’re an essential piece of infrastructure for smart contract-enabled chains. In this one, we’ll dive deeper into how they’re architected to ensure data accuracy and the integrity of onchain data delivery. 🧵👇 First and foremost, it's important to recognize that not all blockchain oracles are created equal or serve the same purpose. There are considerable design differences among various oracles, such as @chainlink, @UMAprotocol, @BandProtocol, and @. However, for the sake of simplicity, we will focus on Chainlink—the largest blockchain oracle provider—as our primary example to illustrate how blockchain oracles function. In our previous thread, we explained why DeFi protocols like @aave can’t rely on a single onchain source such as Uniswap for price feeds. If they do, they expose themselves to manipulation. Asset prices on decentralized exchanges are determined by liquidity-pool ratios, which can be distorted by malicious traders making large swaps using flash loans or their own capital. That’s why protocols like Aave turn to oracle providers such as Chainlink, which—at first glance—seem to solve this problem entirely. After all, Chainlink doesn’t rely on a single exchange like Uniswap; it aggregates price data from many venues, filters it, and delivers the median value onchain. But there’s a deeper layer of risk many people overlook. Even if the underlying data—be it asset prices, weather data, or anything else—is aggregated from dozens of sources, it’s all for nothing if it’s delivered by a single oracle. Trusting a single offchain actor to behave honestly and remain online reintroduces the same single point of failure as relying on a single data source in the first place. If that oracle goes down, gets compromised, or acts maliciously, every onchain protocol depending on it is instantly exposed. This is why Chainlink and most other blockchain oracle providers are designed with decentralization in mind. More specifically, Chainlink isn’t a single oracle but a heterogeneous network of many decentralized oracle networks (DONs). Each DON provides a unique oracle service tailored to its users’ needs. For instance, one DON may provide the ETH/USD price to the Ethereum blockchain, another may provide price feeds for Liquid Staking Tokens (LSTs) to Arbitrum, and a third may transfer tokens cross-chain between Polygon and Solana. Furthermore, each DON comprises multiple nodes run by different operators, often using their own infrastructure and data sources. These may include Web2 telecommunications providers such as Vodafone, leading data providers, and Web3 infrastructure providers such as Infura. These nodes independently fetch, sign, and report data before it’s aggregated into a single, consensus-verified value. For example, a DON consisting of 31 nodes may have each node independently fetch the price data for ETH in USD across multiple centralized and decentralized exchanges and generate its own median value. Then, all the nodes’ medianized price data is aggregated into a single data point (an oracle report) and delivered by the DON to the intended recipient—e.g., a blockchain app like Aave. To aggregate the data from multiple nodes into a single data point, Chainlink uses a protocol called Off-Chain Reporting (OCR). OCR is a system that allows nodes to reach consensus offchain on a single data point. Here’s how it works: First, each node in a DON independently fetches data from its chosen sources and shares its signed report with the rest of the network through a peer-to-peer communication layer. For instance, Node A may report that the price of ETH is $3,120, Node B may say it’s $3,125, Node C may say it’s $3,118, etc. Once enough reports are collected, the nodes use a consensus algorithm to agree on a single aggregated price of ETH—typically a median of all submissions. Only one compact report, containing the final result (a single price for ETH in USD) and the aggregated signatures of the participating nodes, is then published onchain. Once the report is submitted, smart contracts verify its authenticity using those aggregated signatures before accepting it as valid input. This closes the loop, ensuring that no unsigned or tampered data can ever be injected into the blockchain. This design drastically reduces gas costs and improves scalability while maintaining full decentralization and verifiability. Of course, decentralization only works if the nodes within a DON can’t easily collude to manipulate the results. If a handful of them could coordinate to submit false data, all of this would be for nothing. To prevent this, Chainlink relies on a combination of cryptographic guarantees, economic incentives, and transparency mechanisms. 👉Each node cryptographically signs its response, making every submission verifiable and publicly auditable onchain. This ensures that any deviation from expected behavior can be traced back to the responsible operator. 👉On top of that, nodes are economically incentivized to report accurately. They stake LINK tokens and earn fees for good performance, but also risk losing revenue, reputation, and potentially staked assets if they behave dishonestly. Finally, all oracle activity—data updates, node identities, feed parameters, and historical performance—is fully transparent. Anyone can inspect the onchain reports to verify which nodes participated, when updates occurred, and how results were aggregated. If certain nodes misbehave or don’t function properly, their reputation (and potentially their revenue) will take a hit, and nobody will rely on them in the future. In essence, Chainlink’s architecture replaces trust with verification. By decentralizing data sourcing, aggregation, and delivery—and securing each layer with cryptographic proofs, economic incentives, and full transparency—it ensures that blockchain apps can interact with the real world without sacrificing security or reliability.