1. Introduction

1.1. Blockchain: A Trust Machine

For years now, blockchain technology has been making waves. At times, being proclaimed a key driver for the next industrial revolution [1,2]. At others, condemned as nothing more than a useless gimmick powering a gigantic Ponzi scheme [3]. Regardless of the public and market sentiment toward the technology, there are certainly reasons it has received so much attention beyond speculation.

In its most simplified form, a public blockchain is a shared append-only ledger that enforces certain rules while allowing anyone read and write access. This distributed digital ledger allows users to deterministically achieve consensus on a common reality without the need for centralized parties to keep records and govern the process [4]. Blockchain technology was first introduced in Bitcoin, though the term “Blockchain” was never explicitly mentioned in the original whitepaper by Nakamoto [5]. Bitcoin was designed to function as a peer-to-peer electronic cash system with a native, independent currency and monetary system tied to participation in the maintenance of the network and processing of transactions [5]. For the first time in history, there was a globally standardized currency that anyone could use, but nobody could monopolize control over due to the fundamental design of blockchain consensus protocols.

While this in itself is a historic achievement, researchers interested in the technology quickly realized other potential use cases in which such a system might yield significant benefits and potentially revolutionize how we transact, exchange value, and keep records. Early examples mentioned by Tapscott [6] include the standardized recording of land titles, providing financial services to all regardless of factors such as location and access to legal documents, decentralized sharing economy platforms, self-sovereign digital identity, automated royalty payments for digital content, and even technology-enabled systems to support and improve governmental functions. In general, blockchain technology primarily enables two functions: (1) digital scarcity enabling value transfer and (2) decentralized processing and validation of transactions immutably stored in a chronologic ledger; the emphasis concerning more advanced applications beyond basic party-to-party asset transfer is on the latter. While digital scarcity certainly plays a role, decentralized computation and execution of transactions can be used to create systems and protocols for various use cases. The innovation here is that neither trust among participants nor the involvement of a central governing party is necessary for anyone to participate in a system or use an application.

Decentralized computation via a smart contract network allows for guaranteed neutral execution of pre-defined logic. This is useful as it removes the need for trust when engaging in even complex transactions. Decentralized computation is practical when building so-called Decentralized Applications (dApps). User-facing applications utilizing a blockchain in its back-end. Currently, the primary use for dApps is in finance, primarily on the Ethereum blockchain [7]. Decentralized finance (DeFi) is an umbrella term for decentralized exchanges, lending protocols, derivatives, insurance, and more. All logic for such applications is written ahead of time in the form of a smart contract and published on a blockchain. Once published, the code cannot be changed but it can be audited by all parties. The logic outlined in the contract is executed when a party interacts with the contract [7]. This allows parties to engage in complex transactions without the need to trust one another - the operation is trustless [8]. There is no need for a bank, a government, a notary, an escrow agent, or any other middleman whose role traditionally is to help overcome issues of trust. This has the potential to drastically streamline and disintermediate a wide variety of transactions, even beyond finance.

Before going further into detail about the blockchain’s role as a trust machine, it is essential to understand decentralization and how it is achieved and maintained. At the foundation of every blockchain is the consensus protocol - in other words, how network participants agree on a common truth to be immutably recorded on the ledger [9]. Depending on whether the blockchain is public, allowing anyone to participate in consensus, private, allowing only allowlisted nodes to participate, or a hybrid system, different methods are used to ensure network integrity. Underlying these different approaches is the need for Byzantine fault tolerance.

The term Byzantine fault comes from the so-called Byzantine Generals Problem, an allegory describing the fundamental challenge of ensuring a certain minimum percentage of actors behave as expected. The analogy used to describe this problem involves multiple generals planning to attack a fortress. The battle is won if all generals coordinate the attack and follow through. If, however, certain generals fail to participate in the coordinated attack or retreat instead of attacking, the battle is lost. Generals cannot be expected with absolute certainty to communicate honestly. Messengers between them may be intercepted, the other party may inject false messengers, and generals may outright lie [10].

Figure 1. Byzantine Generals Problem.

Figure 1. Byzantine Generals Problem.

The two leading permissionless consensus protocols, meaning anyone can participate and participants cannot inherently be trusted, used in public blockchains to overcome this issue are proof of work (PoW) and proof of stake (PoS). In proof of work, network participants use computing power to solve hash-puzzles [11] while in competition with each other. Economic incentives ensure that participants use their computing power to secure the network rather than attack it [5]. Proof of stake works very similarly, though it does not use computational power but rather direct financial incentives. Participants in PoS consensus commit funds as collateral to contribute to transaction validation. These foundational protocols ensure the network’s neutrality or decentralization [12]. There are other approaches to consensus, some of which build on a certain element of trust achieved by only allowing known actors to participate, also known as permissioned blockchains. This, of course, also creates a limited network that may be easier to target [13].

Byzantine fault tolerance of blockchains allows for a provably fair and independent network able to store and process data without the need for a central authority. In short, blockchains are useful when trust between two parties wishing to transact does not exist, which is quite common in a digital economy.

1.2. The Oracle Problem

Bitcoin was originally designed as a decentralized peer-to-peer cash system able to not only execute transactions between parties but also to utilize a native currency with automated monetary policy [5]. While extreme price volatility [14] and high transaction fees [15] have hindered Bitcoin’s adoption as a digital cash system, as demonstrated when Steam removed Bitcoin as payment method in 2017 [16], some argue Bitcoin has found use as a form of “digital gold” - a relatively uncorrelated digital commodity [17]. With Bitcoin’s continued proof that a decentralized blockchain system can reliably secure value, new projects began emerging. Ethereum, for example, took the concept further by adding Turing completeness - the ability to execute computer code [18]. At Devcon1 in 2015, Ethereum co-founder Vitalik Buterin used the allegory of Ethereum being a smartphone on which Bitcoin is merely one single application [19]. This type of Turing complete Blockchain is also sometimes called a smart contract platform. Smart contract platforms have the ability to execute series of “if-then” statements, allowing for logic to be written onto the network. Using a traditional web front-end and connection to a smart contract platform, developers have the ability to build applications with decentralized execution [20]. Such applications have the potential to offer a tremendous amount of value as they replace the need for trust with cryptographic guarantees.

Figure 2. dApp Architecture.

Figure 2. dApp Architecture.

Blockchains have been operational for around 15 years now, and while the promise of a world with a reduced need for intermediaries has attracted some attention, mainstream adoption, and real-world use cases are still few and far between, at least when compared to the full potential of the technology. While this may largely be attributable to scaling problems found in many blockchain networks, difficult user experience journeys, and an inherent lack of understanding of the technology [21], an argument could also be made that decentralized applications that could make a real positive difference in every day life have not yet been built. Whereas simple decentralized applications are already available and quite successful by some metrics, for example, decentralized peer-to-peer lending platforms and decentralized exchanges [22], more advanced applications that would offer value in the daily life of most average people are not. When developing smart contracts, there is a limit as to what can be done natively within a certain blockchain network. Such networks are isolated systems by design. This provides excellent levels of security and reliability. However, it also makes the development of smart contracts that rely on external data challenging to build. The potential of decentralized execution is limited without the ability to observe and interact with off-chain systems. Hence, the question of how developers may provide data access to their smart contracts without potentially introducing major security vulnerabilities arises. Is there a way to minimize centralization and maximize data integrity in data provision? This challenge is sometimes called the oracle problem [23]. Consequently, third-party services that enable data availability are referred to as oracles.

Figure 3. The Oracle Problem.

Figure 3. The Oracle Problem.

Currently, there are several approaches to allow smart contracts to access blockchain-external data. The most primitive way to enable data access are centralized oracles. Centralized oracles provide data from a single external source and are operated by a single entity. Such oracles are very easy to implement and often purpose-built. However, being centralized, they also present a single point of failure. If the data is corrupted or manipulated, it may lead to significant ramifications [24]. Alternatively, several projects are working on implementing decentralized oracles. Decentralized oracle implementations are the focus of this paper. Among them are Chainlink, Band Protocol, API3, and many others. A brief summary of how these projects tackle the Oracle problem follows.

Chainlink is considered the industry leader, going by various metrics, including the total number of integrations [25]. Chainlink is focused on third-party oracles. Theoretically, anyone can set up and operate a node that makes data available on-chain. Node operators are not automatically considered trusted. Economic considerations are used to create trust, with dishonest node operators being punished. A very similar approach is used by Band Protocol (Band, n.d.-b), though using their own blockchain network. API3, on the other hand, relies on trusted first-party oracles to solve the oracle problem (API3, n.d.). Whereas Chainlink and Band Protocol rely on a network of nodes to fetch data from external data sources, e.g., APIs, API3 enables API providers to operate their own oracle node directly. There are, of course, many other approaches to solving the oracle problem. While this paper primarily focuses on decentralized third-party oracle networks, particularly Chainlink’s implementation, all approaches will be taken into consideration.

Third-party-based oracle systems approach to overcoming the oracle problem is called a decentralized oracle network, DON for short. As briefly explained above, a decentralized oracle network comprises a committee of nodes [26]. DONs ingest data through a connection between the oracle node and an Application Programming Interface (API) provided by one of many data providers. Different nodes within a single DON gather data from multiple independent sources, and a median of all the different data reported by the various nodes is made available to the smart contract.

Figure 4. Decentralized Oracle Network Data Aggregation.

Figure 4. Decentralized Oracle Network Data Aggregation.

Aggregation and the ability for anyone to operate a node within a DON or to create a new DON, provide a high level of reliability, redundancy, and a degree of decentralization. Nodes also sign the data sent to the network, proving which node was the one to send it. Even though the data is aggregated off-chain, these signatures remain attached to the data and are also sent on-chain. Since this signature is stored on a blockchain, there is a transparent and permanent historical record of the performance of a certain node. Reputation systems allow smart contract developers to filter nodes by certain metrics [26]. This introduces competition for node operators and is an additional incentive to keep nodes honest. In addition to data sourcing and delivery, oracle networks may offer various services, such as trustlessly triggering actions in a smart contract, creating and providing a source of randomness, and more. Oracle functions are discussed in part V of Theory.

2. Theory

2.5. Oracle Implementations & Functions

This section focuses on oracle solutions available today. First, a comparison of different oracle protocol implementations is presented. Next, a closer look at data feeds, inputs, and functionalities that allow for the cryptographically guaranteed automation of various services. An overview of a hybrid smart contract developers toolbox, so to speak.

2.5.1. Oracle Risks

This section aims to explain the current industry standards when implementing oracles. While a centralized oracle is easy to establish for each use case, such implementations introduce vulnerabilities. Smart contracts take the data received from oracles as it is presented. Even if the information is clearly false, it is used as instructed. Hence, while hybrid smart contracts themselves will always predictably act as they are programmed to, the contract will not act as it should if the data source used is corrupted or manipulated; the contract will not act as it should. When using a single centralized oracle there is a high risk of this happening [88]. An example can be found when looking at the decentralized lending platform Compound. In 2020, Compound was sourcing data regarding the market value of assets from a centralized oracle run by Coinbase [89]. Compound allows users to borrow on-chain assets so long as they put up other on-chain assets as collateral. If users are unable to maintain a certain minimum percentage of collateralization, their position is liquidated. On November 26th, the oracle reporting the price of DAI, a US Dollar stablecoin, falsely reported the price of DAI as having spiked to $1.30. As a result, some users were now deemed undercollateralized, and over $89 million ended up being erroneously liquidated [90]. Had multiple oracles connected to different exchanges been used to provide the smart contracts with a median price, this situation could have been prevented. Ultimately, this case highlights the need for a degree of decentralization and redundancy with regard to data provision. Decentralized oracle networks are made up of multiple different independent nodes, all independently incentivized to act honestly. Using decentralized oracle networks instead of single centralized oracles has become the state of the art when building decentralized applications.

2.5.2. Oracle Implementations

Many different projects are working on making data accessible to smart contract developers. The following is a summary of some of the most prominent such projects.

Table 1. Oracle Protocols: Overview.

Table 1. Oracle Protocols: Overview.

Protocol	Architecture	Networks	Token	Launch	Team
Chainlink	Multiplea	11	Yes	May 2019	400+
Band	Algorithmic	18	Yes	Sept 2021b	32
API3	1st Party	10	Yes	July 2021	33
Pyth	1st Party	20	Yes	Aug 2021	25
RedStone	Algorithmic	39	Planned	Jan 2022	14
Tellor	Optimistic	6	Yes	Aug 2019	12
Witnet	Algorithmic	25	Yes	Nov 2020	10
DIA	Algorithmic	8	Yes	2018c	25
Supra	Algorithmic	N/ad	Planned	N/ad	90
UMA	Optimistic	2	Yes	May 2021	136

^aAlgorithmic & 1st Party; ^bLaunch of current version; ^cUnclear whether this is start of development or mainnet launch; ^dData Feeds not yet live.

Chainlink

Chainlink is the apparent market leader for oracles by crucial metrics such as the number of integrations, value secured, number of DONs, and several others [25]. Chainlink was the first oracle project launching on the Ethereum main net in May 2019 [91]. Currently, Chainlink DONs are integrated into over 1600 projects, they have delivered over 5.8 billion data points to smart contracts enabling over $6.9 trillion in transactions associated with these smart contracts across 11 different smart contract platforms, including Ethereum, Polygon, Arbitrum, and others [25]. Chainlink DONs are used by many prominent protocols, such as the lending platforms Aave and Compound, the insurance protocols Nexus Mutual and Etherisc, and the Loopring exchange [92]. Chainlink DONs comprise several nodes, each using different data sources. There is a minimum number of nodes that can be set for the data to be used and a maximum deviation from the mean that each node should not surpass. Chainlink employs a combination of a reputation system and staking to ensure nodes act as they should [93]. Chainlink DONs are used to bridge off-chain data and provide off-chain computation.

The team behind Chainlink is the largest in the oracle space, currently consisting of over 400 people [75]. The team is supported by advisors such as Ari Juels (professor at Cornell Tech), Tom Gonser (founder of DocuSign), Eric Schmidt (former CEO of Google), Jeff Weiner ( former CEO of Linkedin), Balaji Srinivasan (former CTO of Coinbase) and Dan Boneh (professor at Stanford University and the head of its Applied Cryptography Group) [55].

Band Protocol

Band Protocol was launched on the Ethereum main net in 2020. However, this first version received little attention from developers, and the team abandoned the initial design to build Band v2. Band v2 relies on its own blockchain, where node operators also act as validators for this chain’s consensus protocol. Nodes combine data retrieved from external APIs to provide data on-chain. However, the types of APIs that can be connected are limited to freely accessible feeds. Password-protected or paid APIs can currently either not be connected or require the cooperation of API providers [94]. Some off-chain computation features are built natively into the protocol. Band protocol currently has limited adoption and is struggling to take advantage of network effects. The team behind Band Protocol consists of roughly 32 people [95].

API3

API3 was launched in July 2021. Included in the founding team are several former Chainlink employees. API3 hopes to disrupt the oracle space by allowing Web2 API providers to make their data available directly to on-chain applications. Rather than using oracle nodes to collect data from various data sources, API3 focuses on first-party oracle nodes. While this design offers some potential advantages, such as transparency of data sources and vastly lower latency [96], it could be argued that this approach brings some of the concerns of centralized oracles. While the need to maintain reputation will likely deter node operators in this system from acting dishonestly, relying on single centralized parties as data sources may introduce certain risks. Currently, API3 is available on ten smart contract networks. Any API can be made available on-chain via the API3-developed middleware Airnode [96]. The team behind API3 currently consists of roughly 33 people [97]. API3DAO, the decentralized autonomous organization used in some governance functions, currently comprises 6538 members [98].

Pyth

Similar to API3, Pyth has taken a first-party oracle approach focusing on low latency, high confidence market data, primarily price data of currencies, crypto assets, and publically traded equities [99]. Currently, Pyth serves over 250 data feeds across over 20 blockchains. Data is served directly by trading firms, cryptocurrency exchanges, and other financial services providers such as Jane Street Capital, Gemini, Cboe Global Markets, and roughly 80 others [100]. Data providers operate a fork of the Solana blockchain named Pythnet. On this chain, data is aggregated and made available to other networks [100]. Having launched in August of 2021, Pyth is on the newer side of oracle platforms. Prominent users of the Pyth network include the synthetic assets platform Synthetix and Tradingview [100]. Pyth currently has around nine employees [101]. Pyth does not currently have an associated token but is planning to launch one for certain governance functions [99].

RedStone

RedStone primarily provides price data regarding cryptocurrencies, stocks, currencies, and commodities, though this platform offers flexibility regarding data types. For example, there is an oracle reporting the number of Refugees leaving Ukraine, the floor price of certain NFTs, and the prices of grains and livestock. Data is primarily sourced from cryptocurrency exchanges and publicly available APIs [102]. RedStone is focusing on both push and pull-based oracles. In push-based oracles, nodes must “push data onto the chain in certain intervals [103]. In pull-based oracles, data is not continuously delivered on-chain but only when requested. Data is stored on the decentralized storage protocol Arweave and is delivered on a specific blockchain by a network of nodes upon request. These oracles are available on over 39 Blockchains [103]. The protocol was launched in January of 2022 [104], currently has around 17 employees [105], and has 1190 different data feeds [102].

Tellor

Tellor is taking an open approach to oracles. A user can request that any data be reported on-chain for a small payment. Data reporters then compete to provide the requested data to claim the tip. Any holder of the token associated with Tellor may lock tokens in order to be able to dispute data reports. When a dispute is opened, the community has two days to vote on the correctness of the report. If the user who opened the dispute correctly flagged the data, the node that supplied the data is penalized. Otherwise, the user who supplied the data receives the locked tokens of the user who opened the dispute. While the dispute is ongoing, a new value is reported to the requester [106]. Tellor is live on the six most prominent EVM main net blockchains, currently has around 12 employees [107], and initially launched in August 2019 [108].

Witnet

The Witnet oracle operates on its own blockchain secured by a decentralized network of nodes that earn WIT from fulfilling requests for data and participating in consensus. Developers may use Witnet for its cryptocurrency price feeds, as a source of randomness, or to connect to external APIs. There are 248 price feeds to 25 smart contract platforms [109]. Witnet originally launched in November 2020 [110] and currently has around 10 team members (Witnet, n.d.-b). Currently, Witnet fulfills around 10,000 to 15,000 data requests per week, according to figures published by the Witnet Foundation. By relying on its own blockchain, Witner Oracles benefits from increased efficiency since the Witnet blockchain is able to communicate data to all other integrated EVM chains without needing a separate decentralized network of node operators for each one [110].

DIA

DIA oracles are focused on providing market data. They currently have around 70 data sources, of which around 45 originate on various blockchains, while 25 are off-chain data sources. Data feeds include prices of various crypto assets, non-fungible tokens (NFTs), and liquid staked derivatives (LSDs). This data is sourced from centralized exchanges, decentralized exchanges, and NFT marketplaces. In addition to price feeds, DIA offers functionality for verifiable randomness on eight smart contract networks, including Polygon, Moonbeam, Fuse, and others [111]. DIA consists of three layers: data collection, data storage and processing, and data publishing. All layers run on a Kubernetes cluster hosted on the IBM cloud [112]. DIA launched in 2018 [113]. The DIA Association currently has around 25 employees [114], and the DIA DAO currently receives between 10 and 30 votes on recent proposals, indicating a rather small community.

Supra

The oracle protocol implementation created by Supra is called Distributed Oracle Agreement (DORA). DORA is made up of a decentralized network of nodes that is itself made up of smaller, periodically randomized networks of nodes [115]. This provides a high level of security in transferring data on-chain. As of July 2023, Supra has not yet launched its oracle protocol on mainnet and is only live as a beta testnet, making it difficult to evaluate further factors. In addition to its oracle protocol, Supra offers randomness services separately from data feeds; this feature is live on several blockchain networks. Supra oracles participated in accelerator programs run by the University of California, Berkeley, and Mastercard [116] (Supra, n.d.-a). Currently, their team consists of roughly 90 people [117].

UMA

UMA employs a so-called optimistic oracle. Anyone can request a specific piece of data such as “What was the mean temperature in July?” or “Who won the World Cup in 2021?”. Users then post a bond and provide an answer to receive a reward. In most cases, the result is assumed to be valid and accepted. In case the information provided is untrue it can be disputed. In this case, the data provider may lose the collateral they deposited. This approach is relatively novel and specializes in delivering a wide variety of data types. It is unsuitable to serve frequently updating data feeds instead posting individual answers to specific questions. UMA is currently primarily used to resolve outcomes in the prediction market “Polymarket” [118]. UMA currently has 136 team members [119]. As of August 2023, UMA has settled 5700 requests since launching on mainnet in May 2021 [120]. An interesting use case of optimistic oracles is using them as triggers in case some external event is observed, e.g., a protocol getting hacked. One downside of this implementation is responsiveness, as oracles must be given a dispute period before being used in a smart contract.

Kleros

Kleros, “The Justice Protocol”, is an oracle focused on determining the truth in subjective cases using crypto-economic incentives and a decentralized network of arbitrators rather than digital data links to objective sources. Kleros focuses on disputes, for example, regarding freelancing, e-commerce, content moderation, etc. [121]. Anyone may become a Juror by staking a certain value. Jurors are presented with cases per their indicated areas of expertise and decide based on voting. Game theoretical models incentivize honest behavior and disincentivize malicious behavior [121]. While this approach differs greatly from the objective raw data-based approach of the oracle platforms mentioned previously, it attempts to fill an important niche in operationally necessary on-chain data. Kleros won the European Commission’s Blockchains for Social Good prize and a grant from the French bank BPI France [122]. Today, Kleros has a team of around 32 people [123].

Augur

Augur takes a completely different approach to oracles than the platforms mentioned previously. It is an oracle and a platform for prediction markets. Users may create prediction markets for future events, such as who will win the upcoming presidential election, users may then bet on the outcome of the event using the US Dollar stablecoin DAI. After a certain deadline, the outcome is determined by allowing users to vote. Correct voters are rewarded, while incorrect voters are penalized [124]. In theory, this method not only provides real-world data on-chain but also provides data on the likelihood of future events. Augur was one of the first dApps to launch on the Ethereum smart contract platform in July 2018 (Blockchains could breathe new life into prediction markets, 2018). In November of 2021, control of Augur was handed over to a newly created Decentralized Autonomous Organization (DAO). The DAO, in theory, is meant to democratize and decentralize ownership and decision-making power over the platform by allowing members to vote on actions and fund allocation of funds. The latest completed prediction market listed on their dApp was completed in July of 2021 with no currently running markets. Recently, a new similar dApp called Polymarket has been gaining attention [125].

The Graph

The Graph indexes and serves data from various smart contract platforms via APIs. Rather than bringing off-chain data on-chain, they are focused on doing the opposite in a way that is easy to integrate for Web2 developers [126]. The Graph offers a lot of flexibility regarding data types going far beyond price feeds.

While Chainlink is currently the industry leader and, in many cases, considered the industry standard, there are many relevant approaches to solving the oracle problem. Some specialize in only one function, while others focus on offering a wide toolbox. In this paper, Chainlink’s terminology and examples of their implementation will, at times, be used; however, the scope of this paper remains project-agnostic.

2.5.3. Inputs and Functions

The inputs made available to smart contract developers by oracles can be split into two main categories: data and off-chain computation. Data may refer to making off-chain data or data from other smart contract platforms that are not natively interoperable available on a specific platform. Computation refers to secure data generation or processing outside of a blockchain or smart contract network to fulfill needs for functions that are not possible or prohibitively expensive when done natively within a smart contract network.

2Data

With regard to the former category, data, there is a wide variety of feeds and categories of data that are made available to smart contract developers. In practice, market data is currently by far the most in-demand data category, as this is necessary for many DeFi applications such as exchanges, lending, and derivatives platforms to operate [127]. The market data category can range from simple spot prices of various on-chain crypto assets to liquidity and volume data of commodities and foreign exchange markets. Various oracle projects support different types of this data. While Chainlink provides great flexibility and supports almost all kinds of such data, other, newer oracle platforms have seen the already maturing market for price feeds and focus exclusively on them, Pyth being an example. Market data availability is a key driver of the decentralized finance (DeFi) ecosystem [127]. DeFi is an umbrella term for decentralized applications offering financial services without relying on intermediaries, such as decentralized exchanges, lending platforms, stablecoin protocols, derivatives, and more [128]. MakerDAO, for example, is a protocol allowing users to lock up various crypto assets to mint the US Dollar stablecoin DAI. The locked value must be higher than the amount of DAI minted and if the value of locked assets drops beneath a set liquidation threshold relative to the outstanding DAI, locked assets may be liquidated. All of this is executed by smart contracts (MakerDAO, 2020). Since smart contracts do not natively know the current value of the US Dollar and relevant crypto assets oracles are needed to provide this information. DeFi has multiplied from roughly 720M USD TVL (Total Value Locked - the value of assets deposited in DeFi protocols) at the beginning of 2019 to a peak of roughly 178B USD at the end of 2021. Currently, TVL hovers between 40B and 45B USD [129]. As mentioned earlier, the rapid growth of this sector has led to many oracle protocols building products targeted specifically to serve DeFi applications.

Another related type of data presently in frequent use is “proof of reserves” data. Stablecoins pegged to currencies, commodities, assets, and their derivatives, are commonly collateralized by assets held off-chain or on another blockchain. In this subcategory, oracles query data sources such as an API made available by a regulated custodian holding funds or assets off-chain or the API of blockchain explorers to verify the amount of funds held by a wallet on another chain [130]. The GBPT stablecoin, for example, is backed by British Pounds (GBP) held in an account at Bank Frick, which provides an API to the nodes of a Chainlink DON. The oracle network pushes new data on-chain every 24 hours or when a deviation of over 2% is reported by the nodes. The oracle network for this PoR (Proof of Reserves) feed currently consists of 16 nodes operated by independent organizations such as Blocksize Capital, Deutsche Telekom MMS, NorthWest Nodes, among others [131].

Aside from finance, there are many custom types of data being made available through oracles. This subcategory is largely made up of less mature data types that rely on rather primitive oracle implementations. The functionality of many oracle protocols to source data from traditional APIs allows developers to easily source many types of data they may need. Chainlink, for example, calls this functionality “External Adapters” [132], while API3 calls it “Airnode” [96]. While often not optimized for decentralization and trust minimization due to the use of a single data source, it offers accessibility and easy implementation. This category of oracle feed enables some of the most innovative use cases by allowing the integration of sensor and satellite data, among other direct data sources. The example of parametric insurance mentioned previously relies on satellite and weather data to determine whether a claim will be paid out [133]. Another example showing an innovative use case of a direct data feed is the Chainlink hackathon project “Link My Ride” which built an external adapter allowing a smart contract to interface with a Tesla vehicle. This interface was used to offer automated rental functionality, including payment and unlocking the car doors for the renter [134]. While this example was highly experimental, it demonstrates the potential of hybrid smart contracts.

The final subcategory of data feeds is cross-chain interfacing. Here oracles are used to read data on one blockchain and provide that data to a smart contract on another blockchain. Beyond crypto stablecoins, in theory, this allows for blockchain interoperability, such as transferring a token from one chain to another, as discussed in Chapter IV a.

Computation & Functions

The second main category of smart contract inputs provided via oracles is computation and functions. In this category, oracles are used to perform functions that are not natively possible or efficient to do on-chain and provide the output for on-chain use without necessarily using data from any external source [130]. The following is a brief summary of various functions currently in use. While many of these functions have different names depending on the oracle platform being used, we will use the terminology used by Chainlink for simplicity.

Automation

In order for a smart contract to execute a certain function, that function needs to be called. In most user-facing smart contracts, this is of no issue as the smart contract only needs to perform an action when interacted with by a user [130]. This is the case with most DeFi apps, for example, exchanges, which execute a function when a user initiates a transaction. Not all smart contracts, however, function in this way. An example includes a smart contract that is supposed to check a condition every 24 hours [130]. Doing this manually or incentivizing users to do this is not usually efficient or reliable. Oracle networks can be used to trustlessly interact with a smart contract in a specified way. Chainlink’s automation, formerly Chainlink Keepers, uses a DON to send a predefined transaction based on a trigger condition. By using a decentralized oracle network, this method offers high reliability and redundancy and incentivizes honest behavior. One example use case for automation mentioned in Chainlink’s documentation is a defi yield harvester that withdraws and reallocates capital, generating yield in a defi protocol, allowing for compounding and the employment of advanced strategies [130].

Randomness

Verifiable randomness is currently one of the most prominent use cases for oracles. Computers have two methods to generate randomness: (1) true random number generators (TRNGs) and (2) pseudo-random number generators (PRNGs). PRNGs use a pre-defined algorithm to emulate randomness, whereas TRNGs employ a software-external phenomenon as a source of entropy for randomness generation. Since this external phenomenon cannot be predicted, it may be used to generate a truly random and unpredictable output. In traditional computers, common things used to generate entropy are, for example, the RPM a certain fan in the computer is spinning, current CPU temperature, mouse movements, etc. [135]. Since Blockchains are composed of thousands of computers contributing to an isolated system, this is not natively possible on most contemporary smart contract platforms. Instead, Verified Randomness Functions (VRF) employ oracle nodes to fill this gap. Chainlink’s VRF, for example, works by “combining block data that is still unknown when the request is made with the oracle node’s pre-committed private key to generate both a random number and a cryptographic proof” [75]. Since both the randomness and the cryptographic proof are generated together, the output provided on-chain can be checked before making it available for use by a smart contract. Due to this verification step, nodes are unable to provide biased or manipulated outputs, as the cryptographic check would fail. This function is currently primarily used in the generation of NFT attributes, gambling, and blockchain-enabled games [136].

Scalability & Off-Chain Computations

While smart contract platforms are essentially distributed computers, they trade computational performance for decentralization. The trade-off between scalability, decentralization, and security in the context of blockchain networks is known as the scalability trilemma [137]. Certain blockchain platforms maintain a high level of decentralization but are limited in computational ability, while others focus on scalability over security. For computationally intensive operations, it makes sense to move execution to smart contract platforms optimized for scalability or to a secure off-chain environment. Using oracles and cryptographic guarantees, the computation can trustlessly be outsourced from blockchain networks with low scalability and high security to those with high scalability, returning only the result of the computation [138].

Privacy & Security

One common criticism of smart contract platforms is radical transparency, to the point of potential privacy issues [139]. Blockchains are pseudonymous, not anonymous - while it is possible to remain anonymous, patterns and behaviors may be used to identify users, their behaviors, and holdings. This presents a significant issue for institutional users who may need to protect trade secrets, such as the identities of stakeholders in their supply chain [140]. Oracles can be used to deploy privacy-enabling add-ons to smart contracts. While these add-ons are both plentiful and modular, with new ones continuously being built, there are currently three primary applications: zero-knowledge proofs used to confidentially yet verifiably share the outputs of computations (DECO), trusted hardware connections (Town Crier) [141], and the decorrelation of smart contract execution and settlement (e.g. Mixicles) [142].

Zero-knowledge proof integrations via oracles, such as DECO, allow users to verifiably share a piece of data without revealing its full context [132]. For example, a user may prove that their credit score is above a certain number via a direct connection from an oracle node to the credit score provider. The oracle node obtains the relevant number and transmits only it, no other details, in a cryptographically reliable way. This simple mechanic has a wide variety of applicable use cases and could become an integral part of next-generation smart contracts. Examples include age verification, creditworthiness, insurance premium calculations, Sybil resistance, fraud prevention, and more. DECO was purchased by Chainlink from Cornell University in 2020 [143].

Trusted hardware connections allow smart contracts to access temperproof and confidential oracle inputs via just a single node. This means sensitive information, such as passwords, and personal data may be securely stored and made accessible when needed [142]. An example mentioned in “Town Crier: An Authenticated Data Feed for Smart Contracts” by Zhang et al. [141] is parametric on-chain flight insurance. In order to verify and pay out a claim, the smart contract must be able to access the claimant’s travel details. If the claimant were to submit his travel details directly to the smart contract, this information would be publicly accessible, infringing on their privacy. Instead, the data could be stored on-chain in an encrypted format. In this case, oracle nodes would be needed to decrypt and verify the information, potentially allowing the node operators to access the information. Using a trusted execution environment (TEE) solves this. The TEE acts as a black box decrypting and verifying the information and transmitting the result, was the flight delayed or not, to the smart contract [141]. Use cases for this technology may be found in health care, cyber security, and beyond.

Mixicles allow users to detach the outcome of a smart contract from a payment output. Consider a smart contract based on a binary query, e.g., “Was flight X on time?”, “Was the price of Apple shares above X at time Y?”, “Did Candidate X win the election?”. Based on these inputs, an oracle network can tell a smart contract to transfer funds held in a smart contract to any predefined address [55]. By using newly generated addresses, users may conceal the financial outcome of that smart contract, creating a disconnect between the smart contract and their associated transactions [142].

Beyond these functions, there are those built to remedy specific problems appearing across smart contract platforms, such as Fair Sequencing Services (FSS), which protect from front-running exploits in DeFI [144], Off-Chain Reporting (OCF) which enable batching of oracle network responses to save gas costs [130], and many other, at times custom-built, functions. All in all, oracles act not only as secure data adapters but also as a toolbox, adding functionalities to blockchain networks that are not natively possible. Not all oracle implementations support all functions at once. Some may specialize in a single function.

2.6. Stakeholders & Economics

This section explains the stakeholders typically involved, from data sourcing to data delivery across various use cases mentioned in the previous chapter using Chainlink’s implementation. The optimal fundamental constellation of stakeholders is a variety of node operators, typically 10-30, sourcing relevant data from a selection of publicly available data feeds [26]. In the case of market data, e.g., cryptocurrency spot trading data, data sources include price aggregators such as coinmarketcap and coingecko, and exchanges such as Binance and Kraken [92]. Data providers for a specific feed are currently curated and allowlisted by Chainlink Labs [93]. Node operators can then choose a data provider to connect to [130]. The network of nodes aggregates this data, and when a trigger condition is met, one node is selected to transmit the data on-chain [26].

Figure 5. Oracle Data Value Chain.

Figure 5. Oracle Data Value Chain.

The above structure pertains primarily to data feeds with sufficient reputable data sources. Newer and more niche data feeds may need to rely on a single data source either due to uncertain demand, e.g., in the case of highly specific satellite imagery, or there simply not being multiple parties who can access and provide specific data, e.g. in the case of bank account balances. Other functions, such as computation, do not require an off-chain data provider, while others, such as Town crier, do not require a network of nodes. In the following, we will go over the major stakeholders in the sourcing and delivery of data on the chain.

2.6.1. Data Providers

As mentioned previously, oracle nodes source data from data providers. Typically, this is done by ingesting an API [130]. This can be a publicly available API or an API made available for this specific use case. While technically, anyone could position themselves as a data provider, there is a particular reputation, transparent methodology, and assurance needed in order to convince node operators that a data feed is trustworthy and reliable. As with Chainlink, some oracle protocols curate a set of data providers for specific feeds. Nodes serving chainlink price feeds, for example, are able to choose from a limited number of exchanges and market data aggregators. Node operators are incentivized to select reliable data providers as incorrect data transmitted via their node may lead to economic or reputational penalties [26]. Hence, it is vital for data providers to have a demonstrably solid methodology for data collection. Considering the example of Chainlink price feeds, there are several options node operators may choose from. Coingecko aggregates market data from over 700 exchanges using a trust score and outlier detection. Coinmarketcap operates quite similarly. Alternatively, exchanges can be used as a primary source for market data, which is reported directly.

Things get interesting when looking at more specialized data types. One example can be found in the case of macroeconomic data. Truflation is a start-up and data provider that provides independent inflation and consumer price index data. Their methodology relies on over 10 million data points to source price data for various individual goods, composing an annually updated household expenditure basket. This allows Truflation to provide close to real-time inflation data [145]. The U.S. Bureau of Labor Statistics publishes inflation data monthly, roughly 12 days after the month’s end (U.S. Bureau of Labor Statistics, n.d.). Smart contract developers now have the option to choose between relying on official legacy sources, unofficial sources that update more frequently using their own methodologies, or an amalgamation of both. Ultimately, the user experience may play a large role in developers’ decisions. Another interesting example is that of data providers for “Proof of Reserves” (PoR) data feeds. PoR feeds are used to prove the collateralization of certain on-chain assets, such as stablecoins. The usefulness of PoR integration also relies heavily on the methodology employed. In July 2023, an article was published criticizing Chainlink’s PoR feeds for being opaque and varying in data source quality [146]. In some cases, these criticisms may be justified as, according to Chainlink’s documentation, some PoR data feeds use self-reporting as the data source [130]. Other PoR feeds use a data feed provided by a regulated custodian as the data source, and others use Blockchain Explorer data to source the wallet balance of an address on another blockchain. Not all data providers and methodologies for sourcing data are created equal.

2.6.2. Node Operators

Nodes receive, consolidate, and process data that is either received from off-chain sources or on-chain sources [26]. While anyone can operate an oracle node, it is the case that for more prominent feeds, node operators are curated and allowlisted based on reputation, for example, with regards to Chainlink price feeds. Reputation plays a significant role in getting jobs for a node. Some of the most reputable node operators currently include Blockdaemon, Blocksize Capital, Deutsche Telekom MMS, LinkPool, and NorthWest Nodes, among others [131]. Unlike some other blockchain projects, node operations, in the case of oracles, are often done by specialized institutions [147]. Obstacles for retail node operators include the cost of specific data sources, which may be over $1000 per month to access, and reputation, as smart contract developers would likely prefer a node that has run a certain job a million times with a success rate of over 99,9% when compared to a newer node. Customers may also require nodes for a specific job to deposit collateral in the form of the LINK token to be eligible. In the case of push-based oracles, nodes need to be funded with the relevant smart contracts platform’s native currency to pay transaction fees when submitting data on-chain. In the case of pull-based oracles, the requester pays this fee [148].

2.6.3. Customers

Customers of oracle services are the organizations operating the smart contracts sourcing data from oracles. This can be simple smart contracts or complete decentralized applications. One of many examples of a protocol that would not be possible without oracle feeds is Ampleforth. Ampleforth is a cryptocurrency with a targeted market value equivalent to the purchasing power of 1$ in 2019. It uses CPI data and market price data oracles to determine whether Ampleforth is trading above or below the target value depending on which the supply expands or contracts to reach the price target [149]. Oracles allow smart contract developers to not only automate existing business logic but also to implement entirely new logic. As the space is still quite new, it remains to be seen what may be built.

2.6.4. Economics, Incentives, Staking

While some oracle implementations leverage their own cryptocurrency token to incentivize all actors to behave correctly, not all do. This section illustrates how token integration may be used to improve reliability and minimize risk through crypto-economic incentive structures. The Chainlink oracle implementation, for example, uses the LINK token for collateral and fee payment. Node operators put up collateral that is lost if the node acts dishonestly. This allows the network to mitigate the effects of an actor attempting to bribe node operators since the attacker would need to bribe multiple nodes to affect the data output of an oracle network. Meanwhile, the incentives are laid out to exponentially increase the rewards of honesty when other nodes are dishonest. Staking is considered an explicit crypto-economic incentive in addition to which the future fees a node operator stands to earn act as an implicit incentive. This further decreases the incentive for node operators to behave dishonestly since reputational history will impact future earnings [26]. Node operators are considered to be economically rational but not honest by default. Additionally, external actors may also stake LINK to participate in a decentralized alerting mechanism to generate some yield [92]. These stakers are incentivized to send alerts when DONs fail to operate as they should. Cryptoeconomic incentives play a significant role in many oracle protocol designs [150].

3. Literature Review

4. Methodology

5. Results & Discussion

6. Summary

References

1. Anderson, J. ’Ike Pono—designing the political and economic systems of the Internet generation 2019.
2. Khalil, M.; Khawaja, K.F.; Sarfraz, M. The adoption of blockchain technology in the financial sector during the era of fourth industrial revolution: a moderated mediated model. Quality & Quantity 2022, 56, 2435–2452.
3. Corradi, F.; Höfner, P. The disenchantment of Bitcoin: unveiling the myth of a digital currency. International Review of Sociology 2018, 28, 193–207. [CrossRef]
4. Wang, X.; Ni, W.; Zha, X.; Yu, G.; Liu, R.P.; Georgalas, N.; Reeves, A. Capacity analysis of public blockchain. Computer Communications 2021, 177, 112–124. [CrossRef]
5. Nakamoto, S. Bitcoin: A peer-to-peer electronic cash system 2008.
6. Tapscott, A. Blockchain Revolution | Talks at Google. https://www.youtube.com/watch?v=3PdO7zVqOwc, 2016.
7. Zheng, P.; Jiang, Z.; Wu, J.; Zheng, Z. Blockchain-based decentralized application: A survey. IEEE Open Journal of the Computer Society 2023. [CrossRef]
8. Liu, J.; Liu, Z. A survey on security verification of blockchain smart contracts. IEEE access 2019, 7, 77894–77904. [CrossRef]
9. Borge, M.; Kokoris-Kogias, E.; Jovanovic, P.; Gasser, L.; Gailly, N.; Ford, B. Proof-of-personhood: Redemocratizing permissionless cryptocurrencies. In Proceedings of the 2017 IEEE European Symposium on Security and Privacy Workshops (EuroS&PW). IEEE, 2017, pp. 23–26.
10. Lamport, L.; Shostak, R.; Pease, M. The Byzantine generals problem. In Concurrency: the works of leslie lamport; 2019; pp. 203–226.
11. Jakobsson, M.; Juels, A. Proofs of work and bread pudding protocols. In Proceedings of the Secure Information Networks: Communications and Multimedia Security IFIP TC6/TC11 Joint Working Conference on Communications and Multimedia Security (CMS’99) September 20–21, 1999, Leuven, Belgium. Springer, 1999, pp. 258–272.
12. Saleh, F. Blockchain without waste: Proof-of-stake. The Review of financial studies 2021, 34, 1156–1190. [CrossRef]
13. Solat, S.; Calvez, P.; Naït-Abdesselam, F. Permissioned vs. Permissionless Blockchain: How and Why There Is Only One Right Choice. J. Softw. 2021, 16, 95–106. [CrossRef]
14. Athey, S.; Parashkevov, I.; Sarukkai, V.; Xia, J. Bitcoin pricing, adoption, and usage: Theory and evidence 2016.
15. Chohan, U.W. The limits to Blockchain? Scaling vs. decentralization 2019.
16. Liao, S. Steam No Longer Accepting Bitcoin Due to ’High Fees and Volatility’. https://www.theverge.com/2017/12/6/16743220/valve-steam-bitcoin-game-store-payment-method-crypto-volatility, 2017.
17. Uddin, M.A.; Ali, M.H.; Masih, M. Bitcoin—A hype or digital gold? Global evidence. Australian economic papers 2020, 59, 215–231. [CrossRef]
18. Zhang, X.; Li, Y.; Sun, M. Towards a formally verified EVM in production environment. In Proceedings of the Coordination Models and Languages: 22nd IFIP WG 6.1 International Conference, COORDINATION 2020, Held as Part of the 15th International Federated Conference on Distributed Computing Techniques, DisCoTec 2020, Valletta, Malta, June 15–19, 2020, Proceedings 22. Springer, 2020, pp. 341–349.
19. Buterin, V. Devcon1: Understanding the Ethereum Blockchain Protocol. https://www.youtube.com/watch?v=gjwr-7PgpN8, 2016.
20. Khan, S.N.; Loukil, F.; Ghedira-Guegan, C.; Benkhelifa, E.; Bani-Hani, A. Blockchain smart contracts: Applications, challenges, and future trends. Peer-to-peer Networking and Applications 2021, 14, 2901–2925. [CrossRef]
21. Presthus, W.; O’Malley, N.O. Motivations and barriers for end-user adoption of bitcoin as digital currency. Procedia Computer Science 2017, 121, 89–97. [CrossRef]
22. Shah, K.; Lathiya, D.; Lukhi, N.; Parmar, K.; Sanghvi, H. A systematic review of decentralized finance protocols. International Journal of Intelligent Networks 2023. [CrossRef]
23. Caldarelli, G.; Rossignoli, C.; Zardini, A. Overcoming the blockchain oracle problem in the traceability of non-fungible products. Sustainability 2020, 12, 2391. [CrossRef]
24. Beniiche, A. A study of blockchain oracles. arXiv preprint arXiv:2004.07140 2020.
25. Nazarov, S. The Chainlink Network in 2023. https://blog.chain.link/the-chainlink-network-in-2023/, 2023.
26. Breidenbach, L.; Cachin, C.; Chan, B.; Coventry, A.; Ellis, S.; Juels, A.; Koushanfar, F.; Miller, A.; Magauran, B.; Moroz, D.; et al. Chainlink 2.0: Next steps in the evolution of decentralized oracle networks. Chainlink Labs 2021, 1, 1–136.
27. Palmer, M. Data is the new oil. ANA marketing maestros 2006, 3.
28. McAfee, A.; Brynjolfsson, E.; Davenport, T.H.; Patil, D.; Barton, D. Big data: the management revolution. Harvard business review 2012, 90, 60–68.
29. Marr, B. Big data in practice: how 45 successful companies used big data analytics to deliver extraordinary results; John Wiley & Sons, 2016.
30. Davenport, T.H.; Harris, J.G. Competing on analytics: the new science of Winning. Harvard business review press, Language 2007, 15, 24.
31. Lindecrantz, E.; Gi, M.T.P.; Zerbi, S. Personalized Experience for Customers: Driving Differentiation in Retail| McKinsey. McKinsey & Company 2020.
32. Raghupathi, W.; Raghupathi, V. Big data analytics in healthcare: promise and potential. Health information science and systems 2014, 2, 1–10. [CrossRef]
33. Baldwin, R. Netflix gambles on big data to become the HBO of streaming. Wired. Retrieved April 2012, 9, 2021.
34. Alphabet Inc.. Alphabet Inc. Form 10-K For the Fiscal Year Ended December 31, 2022. https://abc.xyz/assets/investor/static/pdf/20230203_alphabet_10K.pdf?cache=5ae4398, 2023.
35. Amazon. AMAZON.COM, INC. FORM 10-K For the Fiscal Year Ended December 31, 2022. https://d18rn0p25nwr6d.cloudfront.net/CIK-0001018724/d2fde7ee-05f7-419d-9ce8-186de4c96e25.pdf, 2023.
36. Microsoft 2022 Annual Report. https://www.microsoft.com/investor/reports/ar22/index.html, 2023.
37. Oracle Announces Fiscal 2022 Fourth Quarter and Fiscal Full Year Financial Results. https://investor.oracle.com/investor-news/news-details/2022/Oracle-Announces-Fiscal-2022-Fourth-Quarter-and-Fiscal-Full-Year-Financial-Results/default.aspx, 2023.
38. Ofulue, J.; Benyoucef, M. Data monetization: insights from a technology-enabled literature review and research agenda. Management Review Quarterly 2022, pp. 1–45.
39. Melendez, S.; Pasternack, A. Here are the data brokers quietly buying and selling your personal information. Fast Company 2019, 2.
40. Giombi, K.; Viator, C.; Hoover, J.; Tzeng, J.; Sullivan, H.W.; O’Donoghue, A.C.; Southwell, B.G.; Kahwati, L.C. The impact of interactive advertising on consumer engagement, recall, and understanding: A scoping systematic review for informing regulatory science. Plos one 2022, 17, e0263339. [CrossRef]
41. Loomis, C.J. BlackRock: The $4.3 Trillion Force. https://fortune.com/2014/07/07/blackrock-larry-fink/, 2014.
42. Haberly, D.; MacDonald-Korth, D.; Urban, M.; Wójcik, D. Asset management as a digital platform industry: A global financial network perspective. Geoforum 2019, 106, 167–181. [CrossRef]
43. Ungarino, R. Here are 9 fascinating facts to know about BlackRock, the world’s largest asset manager. https://www.businessinsider.com/what-to-know-about-blackrock-larry-fink-biden-cabinet-facts-2020-12, 2020.
44. Henderson, R.; Walker, O. BlackRock’s black box: The technology hub of modern finance. Financial Times 2020.
45. Bailly, A.; Blanc, C.; Francis, É.; Guillotin, T.; Jamal, F.; Wakim, B.; Roy, P. Effects of dataset size and interactions on the prediction performance of logistic regression and deep learning models. Computer Methods and Programs in Biomedicine 2022, 213, 106504. [CrossRef]
46. Elallid, B.B.; Benamar, N.; Hafid, A.S.; Rachidi, T.; Mrani, N. A comprehensive survey on the application of deep and reinforcement learning approaches in autonomous driving. Journal of King Saud University-Computer and Information Sciences 2022, 34, 7366–7390. [CrossRef]
47. Chui, M.; Hazan, E.; Roberts, R.; Singla, A.; Smaje, K. The economic potential of generative AI 2023.
48. Szabo, N. The God Protocols. http://web.archive.org/web/20061230075325/http://www.theiia.org/ITAudit/index.cfm?act=itaudit.archive&fid=216, 1999.
49. Szabo, N. Unenumerated: Bit Gold. https://unenumerated.blogspot.com/2005/12/bit-gold.html, 2008.
50. Lee, D.K.C.; Guo, L.; Wang, Y. Cryptocurrency: A new investment opportunity? Available at SSRN 2994097 2017.
51. Bitnodes. https://bitnodes.io/.
52. CoinGecko. Total Crypto Market Cap Chart. https://www.coingecko.com/en/global-charts.
53. Covarrubias, L.; Zadamig, J. Organization, Autonomy and Decentralization in the Information Age. Journal of Legislation Science-Number 2019.
54. Ethereum.org. Ethereum Virtual Machine (EVM). https://ethereum.org/en/developers/docs/evm/, 2023.
55. Chainlink. Understanding How Data and APIs Power Next-Generation Economies. https://blog.chain.link/understanding-how-data-and-apis-power-next-generation-economies/, 2020.
56. Caldarelli, G. Real-world blockchain applications under the lens of the oracle problem. A systematic literature review. In Proceedings of the 2020 IEEE International Conference on Technology Management, Operations and Decisions (ICTMOD). IEEE, 2020, pp. 1–6.
57. Chainlink. Hybrid Smart Contracts Explained. https://chain.link/education-hub/hybrid-smart-contracts, 2023.
58. Kshetri, N. Blockchain-based smart contracts to provide crop insurance for smallholder farmers in developing countries. IT Professional 2021, 23, 58–61. [CrossRef]
59. O’Donnell, A.R. Etherisc Launches Blockchain-Backed Parametric Flight Delay Insurance. https://iireporter.com/etherisc-launches-blockchain-backed-parametric-flight-delay-insurance/, 2022.
60. Hamledari, H.; Fischer, M. Role of blockchain-enabled smart contracts in automating construction progress payments. Journal of legal affairs and dispute resolution in engineering and construction 2021, 13, 04520038. [CrossRef]
61. Ghosh, R.K.; Gupta, S.; Singh, V.; Ward, P.S. Demand for crop insurance in developing countries: New evidence from India. Journal of agricultural economics 2021, 72, 293–320. [CrossRef]
62. Rataj, E.; Kunzweiler, K.; Garthus-Niegel, S. Extreme weather events in developing countries and related injuries and mental health disorders-a systematic review. BMC public health 2016, 16, 1–12. [CrossRef]
63. Brüntrup, M.; Heidhues, F. Subsistence agriculture in development: Its role in process of structural change; Grauer, 2002.
64. Ranasinghe, R.; Ruane, A.C.; Vautard, R.; Arnell, N.; Coppola, E.; Cruz, F.A.; Dessai, S.; Saiful Islam, A.; Rahimi, M.; Carrascal, D.R.; et al. Climate change information for regional impact and for risk assessment 2021.
65. Petram, L.O.; et al. The world’s first stock exchange: How the Amsterdam market for Dutch East India Company shares became a modern securities market, 1602-1700. PhD thesis, Universiteit van Amsterdam [Host], 2011.
66. Cong, L.W.; He, Z. Blockchain disruption and smart contracts. The Review of Financial Studies 2019, 32, 1754–1797. [CrossRef]
67. Attaran, M.; Gunasekaran, A. Applications of blockchain technology in business: challenges and opportunities 2019.
68. Semenzin, S.; Rozas, D.; Hassan, S. Blockchain-based application at a governmental level: disruption or illusion? The case of Estonia. Policy and Society 2022, 41, 386–401. [CrossRef]
69. Young, C.R. A Lawyer’s Divorce: Will Decentralized Ledgers and Smart Contracts Succeed in Cutting Out the Middleman. Wash. UL Rev. 2018, 96, 649.
70. Bush, C. Dealing with the conflicts of interest of credit rating agencies: a balanced cure for the disease. Capital Markets Law Journal 2022, 17, 334–364. [CrossRef]
71. Merle, R. A guide to the financial crisis—10 years later. Washington Post. https://www. washingtonpost. com/business/economy/a-guide-to-the-financial-crisis–10-years-later/2018/09/10/114b76ba-af10-11e8-a20b-5f4f84429666_story. html 2018.
72. Caldarelli, G.; Ellul, J. The blockchain oracle problem in decentralized finance—a multivocal approach. Applied Sciences 2021, 11, 7572. [CrossRef]
73. Chainalysis. Wormhole Hack: Lessons from the Wormhole Exploit. https://www.chainalysis.com/blog/wormhole-hack-february-2022/, 2022.
74. Circle. Cross-Chain Transfer Protocol (CCTP) | Circle. https://www.circle.com/en/cross-chain-transfer-protocol.
75. Chainlink. Verifiable Randomness for Blockchain Smart Contracts. https://blog.chain.link/chainlink-vrf-on-chain-verifiable-randomness/, 2020.
76. Scott, S.V.; Zachariadis, M. The Society for Worldwide Interbank Financial Telecommunication (SWIFT): Cooperative governance for network innovation, standards, and community; Taylor & Francis, 2014.
77. Swift. Swift Explores Blockchain Interoperability to Remove Friction from Tokenised Asset Settlement. https://www.swift.com/news-events/news/swift-explores-blockchain-interoperability-remove-friction-tokenised-asset-settlement, 2023.
78. Sazandrishvili, G. Asset tokenization in plain English. Journal of Corporate Accounting & Finance 2020, 31, 68–73.
79. Kim, S. Fractional ownership, democratization and bubble formation-the impact of blockchain enabled asset tokenization 2020.
80. He, S.; Manela, A.; Ross, O.; von Wachter, V. Fundamentals of perpetual futures. arXiv preprint arXiv:2212.06888 2022.
81. Ng, T.S. Blockchain and beyond: smart contracts. Bus. L. Today 2017, p. 1.
82. Kamath, R. Food traceability on blockchain: Walmart’s pork and mango pilots with IBM. The Journal of the British Blockchain Association 2018, 1. [CrossRef]
83. PYMNTS. Walmart Puts Pricey Blockchain Food Tracking Platform on Ice. https://www.pymnts.com/blockchain/2022/walmart-puts-pricey-blockchain-food-tracking-platform-on-ice/, 2022.
84. Caldarelli, G. Understanding the blockchain oracle problem: A call for action. Information 2020, 11, 509. [CrossRef]
85. Meinert, E.; Alturkistani, A.; Foley, K.A.; Osama, T.; Car, J.; Majeed, A.; Van Velthoven, M.; Wells, G.; Brindley, D.; et al. Blockchain implementation in health care: Protocol for a systematic review. JMIR research protocols 2019, 8, e10994. [CrossRef]
86. Zeiselmair, A.; Steinkopf, B.; Gallersdörfer, U.; Bogensperger, A.; Matthes, F. Analysis and Application of Verifiable Computation Techniques in Blockchain Systems for the Energy Sector. Frontiers in Blockchain 2021, 4, 725322. [CrossRef]
87. Latifi, S.; Zhang, Y.; Cheng, L.C. Blockchain-based real estate market: One method for applying blockchain technology in commercial real estate market. In Proceedings of the 2019 IEEE international conference on blockchain (blockchain). IEEE, 2019, pp. 528–535.
88. Caldarelli, G.; Zardini, A.; Rossignoli, C. Blockchain adoption in the fashion sustainable supply chain: Pragmatically addressing barriers. Journal of Organizational Change Management 2021, 34, 507–524. [CrossRef]
89. Chipolina, S. Oracle exploit sees $89 million liquidated on compound. Decrypt 2020.
90. Qin, K.; Zhou, L.; Gamito, P.; Jovanovic, P.; Gervais, A. An empirical study of defi liquidations: Incentives, risks, and instabilities. In Proceedings of the Proceedings of the 21st ACM Internet Measurement Conference, 2021, pp. 336–350.
91. Chainlink. Three Years on Mainnet. https://blog.chain.link/three-years-on-mainnet/, 2022.
92. Chainlink. How Chainlink Price Feeds Secure the DeFi Ecosystem. https://blog.chain.link/chainlink-price-feeds-secure-defi/, 2022.
93. Berger, B.; Huber, S.; Pfeifhofer, S. OraclesLink: An architecture for secure oracle usage. In Proceedings of the 2020 Second International Conference on Blockchain Computing and Applications (BCCA). IEEE, 2020, pp. 66–72.
94. SmartContent. Band Protocol and Chainlink: A Comparative Analysis. https://smartcontentpublication.medium.com/a-comparative-analysis-of-band-protocol-and-chainlink-54b7d14823b5, 2021.
95. Band. Band Protocol | LinkedIn | Team. https://th.linkedin.com/company/band-protocol.
96. Benligiray, B.; Milic, S.; Vänttinen, H. Decentralized apis for web 3.0. API3 Foundation Whitepaper 2020.
97. API3. Api3 | linkedin | Team. https://www.linkedin.com/company/api3.
98. API3. What are first-party oracles? | Documentation. https://dapi-docs.api3.org/explore/introduction/first-party.html.
99. Pyth Network: A First-Party Oracle. https://pyth.network/whitepaper.pdf, 2022.
100. Pyth Data Association. https://pyth.network/.
101. Pyth. Pyth Network | LinkedIn | Team. https://ch.linkedin.com/company/pyth-network.
102. RedStone. https://app.redstone.finance/.
103. Redstone Docs. https://docs.redstone.finance.
104. RedStone. Introducing RedStone. https://medium.com/@RedStone_Finance/introducing-redstone-1b79875df4f0, 2022.
105. RedStone. Redstone Oracles | LinkedIn | Team. https://ch.linkedin.com/company/redstone-finance.
106. Tellor. The Tellor Whitepaper | Decentralized Oracle Protocol. https://tellor.io/whitepaper/, 2022. Accessed on 2024-03-20.
107. Tellor. Tellor | LinkedIn | Team. https://www.linkedin.com/company/tellorinc/about/.
108. Core, T. Tellor Launches to Mainnet! https://tellor.io/tellor-launches-to-mainnet/, 2019. Accessed on 2024-03-20.
109. Witnet. Witnet Oracle Docs. https://docs.witnet.io/. Retrieved August 26, 2023.
110. Sánchez de Pedro, A. Witnet After Mainnet: This Is Just the Beginning! https://medium.com/witnet/witnet-after-mainnet-this-is-just-the-beginning-e4b27485c36b, 2020.
111. Dia App | Cross-Chain Oracles for Web3. https://www.diadata.org/app.
112. DIA Technical Structure. https://docs.diadata.org/introduction/dia-technical-structure.
113. Coinmonks. Dia—Open Source Oracles for Web3. https://medium.com/coinmonks/dia-open-source-oracles-for-Web3-c873ddd46a50, 2022.
114. DIA. Dia Association | LinkedIn | Team. https://ch.linkedin.com/company/diadata-org.
115. Supra Research. Dora: Distributed Oracle Agreement. https://supraoracles.com/docs/SupraOracles-DORA-Whitepaper.pdf, 2023.
116. Supra. Supra Oracles | LinkedIn | Team. https://www.linkedin.com/company/supraoracles/people/.
117. Supra Oracles. https://supraoracles.com/.
118. UMA. https://uma.xyz. Retrieved August 26, 2023.
119. UMA. Uma | LinkedIn | Team. https://www.linkedin.com/company/umaproject. Retrieved August 26, 2023.
120. Lambur, H. Introducing UMA’s Optimistic Oracle. https://medium.com/uma-project/introducing-umas-optimistic-oracle-d92ce5d1a4bc, 2021.
121. Lesaege, C.; George, W.; Ast, F. Kleros: Long Paper v2. 0.2. Kleros 2021.
122. Kleros. Kleros. https://kleros.io/.
123. Kleros. Kleros | LinkedIn | Team. https://fr.linkedin.com/company/kleros.
124. Augur. The Augur White Paper: A Decentralized Oracle and Prediction Market Platform 2018.
125. Ramirez, B. A Resurgent Online Betting Market Is Boosted by Crypto and Current Events. https://www.nbcnews.com/tech/internet/polymarket-online-bet-submersible-russia-war-rcna93122, 2023.
126. Tal, Y.; Ramirez, B.; Pohlmann, J. The Graph: A Decentralized Query Protocol for Blockchains. Disponibile all’indirizzo: https://raw. githubusercontent. com/graphprotocol/research/master/papers/whitepaper/the-graph-whitepaper. pdf 2018.
127. Kaleem, M.; Shi, W. Demystifying pythia: A survey of chainlink oracles usage on ethereum. In Proceedings of the Financial Cryptography and Data Security. FC 2021 International Workshops: CoDecFin, DeFi, VOTING, and WTSC, Virtual Event, March 5, 2021, Revised Selected Papers 25. Springer, 2021, pp. 115–123.
128. Aquilina, M.; Frost, J.; Schrimpf, A. Decentralised finance (DeFi): a functional approach. Available at SSRN 4325095 2023. [CrossRef]
129. DeFiLlama. https://defillama.com/.
130. Chainlink. The Chainlink Economics 2.0 Staking Protocol and Staking v0.1 Launch Details. https://blog.chain.link/chainlink-staking-launch-details/, 2022.
131. Chainlink Data. Mainnet Data. https://data.chain.link/ethereum/mainnet.
132. Chainlink Labs Research. Deco Research Series #1: Introduction. https://blog.chain.link/deco-introduction/, 2023.
133. Papacharissiou, H. Build a Parametric Insurance Smart Contract with Chainlink. https://blog.chain.link/parametric-insurance-smart-contract/, 2020.
134. Chainlink Blog. How to Connect a Tesla to a Smart Contract via a Chainlink Node. https://blog.chain.link/create-tesla-smart-contract-rental/, 2020.
135. Stipčević, M.; Koç, Ç.K. True random number generators. In Open Problems in Mathematics and Computational Science; Springer, 2014; pp. 275–315.
136. Solouki, M.; Bamakan, S.M.H. An in-depth insight at digital ownership through dynamic NFTs. Procedia Computer Science 2022, 214, 875–882. [CrossRef]
137. Hafid, A.; Hafid, A.S.; Samih, M. Scaling blockchains: A comprehensive survey. IEEE access 2020, 8, 125244–125262. [CrossRef]
138. Chainlink. What is Off-chain Data and Off-chain Computation? https://chain.link/education-hub/off-chain-data, 2023.
139. Conti, M.; Kumar, E.S.; Lal, C.; Ruj, S. A survey on security and privacy issues of bitcoin. IEEE communications surveys & tutorials 2018, 20, 3416–3452.
140. Finck, M. Blockchains and data protection in the European Union. Eur. Data Prot. L. Rev. 2018, 4, 17. [CrossRef]
141. Zhang, F.; Cecchetti, E.; Croman, K.; Juels, A.; Shi, E. Town crier: An authenticated data feed for smart contracts. In Proceedings of the Proceedings of the 2016 aCM sIGSAC conference on computer and communications security, 2016, pp. 270–282.
142. Juels, A. Town Crier and Chainlink: Enriching the Function of Blockchain Oracles. https://blog.chain.link/town-crier-and-chainlink/, 2019.
143. Jessel, B. Chainlink’s New Acquisition from Cornell University Could Transform Blockchain for Good. https://www.forbes.com/sites/benjessel/2020/08/29/chainlinks-new-acquisition-from-cornell-university-could-transform-blockchain-for-good/, 2020.
144. Zhao, Y.; Kang, X.; Li, T.; Chu, C.K.; Wang, H. Toward trustworthy defi oracles: past, present, and future. IEEE Access 2022, 10, 60914–60928.
145. Truflation. Methodology. https://truflation.com/methodology. Retrieved August 26, 2023. [CrossRef]
146. Kessler, S. Chainlink ‘Proof of Reserve’ Proves Little Beyond Data Going In, Coming Out. https://www.coindesk.com/tech/2023/07/05/chainlink-proof-of-reserve-proves-little-beyond-data-going-in-coming-out/, 2023.
147. LinkRiver. Guardians of the Web3 Economy: Chainlink Node Operators. https://blog.linkriver.io/chainlink-node/, 2023.
148. Mühlberger, R.; Bachhofner, S.; Castelló Ferrer, E.; Di Ciccio, C.; Weber, I.; Wöhrer, M.; Zdun, U. Foundational oracle patterns: Connecting blockchain to the off-chain world. In Proceedings of the Business Process Management: Blockchain and Robotic Process Automation Forum: BPM 2020 Blockchain and RPA Forum, Seville, Spain, September 13–18, 2020, Proceedings 18. Springer, 2020, pp. 35–51.
149. Ampleforth docs. https://docs.ampleforth.org/.
150. Murimi, R.M.; Wang, G.G. On elastic incentives for blockchain oracles. Journal of Database Management (JDM) 2021, 32, 1–26. [CrossRef]
151. Ellis, S.; Juels, A.; Nazarov, S. Chainlink: A decentralized oracle network. Retrieved March 2017, 11, 1.
152. De Collibus, F.M.; Partida, A.; Piškorec, M.; Tessone, C.J. Heterogeneous preferential attachment in key ethereum-based cryptoassets. Frontiers in Physics 2021, 9, 720708.
153. Egberts, A. The oracle problem-an analysis of how blockchain oracles undermine the advantages of decentralized ledger systems. Available at SSRN 3382343 2017.
154. Eberhardt, J.; Tai, S. On or off the blockchain? Insights on off-chaining computation and data. In Proceedings of the Service-Oriented and Cloud Computing: 6th IFIP WG 2.14 European Conference, ESOCC 2017, Oslo, Norway, September 27-29, 2017, Proceedings 6. Springer, 2017, pp. 3–15.
155. Xu, X.; Weber, I.; Staples, M. Architecture for blockchain applications; Springer, 2019.
156. Al-Breiki, H.; Rehman, M.H.U.; Salah, K.; Svetinovic, D. Trustworthy blockchain oracles: review, comparison, and open research challenges. IEEE access 2020, 8, 85675–85685.
157. Lo, S.K.; Xu, X.; Staples, M.; Yao, L. Reliability analysis for blockchain oracles. Computers & Electrical Engineering 2020, 83, 106582.
158. Nazarov, S.; Shukla, P.; Erwin, A.; Rajput, A. Bridging the governance gap: Interoperability for blockchain and legacy systems. In Proceedings of the World Economic Forum whitepaper. https://www. weforum. org/whitepapers/bridging-the-governance-gap-interoperability-for-blockchain-and-legacy-systems, 2020.
159. Caldarelli, G. Wrapping trust for interoperability: A preliminary study of wrapped tokens. Information 2021, 13, 6.
160. Šimunić, S.; Bernaca, D.; Lenac, K. Verifiable computing applications in blockchain. IEEE access 2021, 9, 156729–156745.
161. Eskandari, S.; Salehi, M.; Gu, W.C.; Clark, J. Sok: Oracles from the ground truth to market manipulation. In Proceedings of the Proceedings of the 3rd ACM Conference on Advances in Financial Technologies, 2021, pp. 127–141.
162. Ezzat, S.K.; Saleh, Y.N.; Abdel-Hamid, A.A. Blockchain oracles: State-of-the-art and research directions. IEEE Access 2022, 10, 67551–67572.
163. API3. API3 DAO Tracker - on-chain analytics: Members, staking rewards, API3 token circulating supply. https://tracker.api3.org/.
164. Band. Band Protocol - Cross-Chain Data Oracle. https://bandprotocol.com.
165. Blockchains could breathe new life into prediction markets. The Economist 2018.
166. Chainlink. Linkedin. https://www.linkedin.com/company/chainlink-labs.
167. Chainlink. A Global Team of Smart Contract Experts. https://chain.link/team.
168. Chainlink. Chainlink Developer Docs. https://docs.chain.link/.
169. Chainlink. Mixicles: Smart Contract Privacy for DeFi on Public Blockchains. https://blog.chain.link/breaking-down-mixicles-and-its-potential-to-unlock-enterprise-demand-for-defi-applications-on-public-blockchains/, 2019.
170. Pyth-Network. Governance and Staking for Pyth Tokens. https://github.com/pyth-network/governance, 2022.
171. Ranasinghe, R.; Ruane, A.; Vautard, R.; Arnell, N.; Coppola, E.; Cruz, F.; Dessai, S.; Islam, A.; Rahimi, M.; Ruiz Carrascal, D.; et al., Climate Change Information for Regional Impact and for Risk Assessment. In Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change; Masson-Delmotte, V.; Zhai, P.; Pirani, A.; Connors, S.; Péan, C.; Berger, S.; Caud, N.; Chen, Y.; Goldfarb, L.; Gomis, M.; et al., Eds.; Cambridge University Press: Cambridge, United Kingdom and New York, NY, USA, 2021; p. 1767–1926. [CrossRef]
172. Schletz, M. ReFi Ecosystem Litepaper. https://www.openearth.org//blog/current-state-of-refi-a-litepaper-exploring-how-to-create-interoperability-in-the-ecosystem, 2022.
173. Core, T. Tellor launches to Mainnet! | tellor. https://tellor.io/tellor-launches-to-mainnet/, 2019.
174. U.S. Bureau of Labor Statistics. Consumer Price Index. https://www.bls.gov/cpi/.
175. Witnet. Witnet Foundation | LinkedIn | Team. https://es.linkedin.com/company/witnet. Retrieved August 26, 2023.
176. Caldarelli, G. Overview of blockchain oracle research. Future Internet 2022, 14, 175.
177. Juels, A.; Breidenbach, L.; Coventry, A.; Nazarov, S.; Ellis, S.; Magauran, B. Mixicles: Simple private decentralized finance, 2019.
178. Team, M. The Maker Protocol: MakerDAO’s Mulfi-Collateral Dai (MCD) System. White paper 2020, pp. 1–25.
179. Mammadzada, K.; Iqbal, M.; Milani, F.; García-Bañuelos, L.; Matulevičius, R. Blockchain oracles: A framework for blockchain-based applications. In Proceedings of the Business Process Management: Blockchain and Robotic Process Automation Forum: BPM 2020 Blockchain and RPA Forum, Seville, Spain, September 13–18, 2020, Proceedings 18. Springer, 2020, pp. 19–34.
180. Marbouh, D.; Abbasi, T.; Maasmi, F.; Omar, I.A.; Debe, M.S.; Salah, K.; Jayaraman, R.; Ellahham, S. Blockchain for COVID-19: review, opportunities, and a trusted tracking system. Arabian journal for science and engineering 2020, 45, 9895–9911.
181. Pasdar, A.; Lee, Y.C.; Dong, Z. Connect api with blockchain: A survey on blockchain oracle implementation. ACM Computing Surveys 2023, 55, 1–39.
182. PwC. Blockchain: The $5 Billion Opportunity for Reinsurers 2016.
183. Al Sadawi, A.; Hassan, M.S.; Ndiaye, M. On the integration of blockchain with IoT and the role of oracle in the combined system: The full picture. IEEE Access 2022, 10, 92532–92558.
184. De Pedro, A.S.; Levi, D.; Cuende, L.I. Witnet: A decentralized oracle network protocol. arXiv preprint arXiv:1711.09756 2017.
185. White, L.J.; et al. Credit-rating agencies and the financial crisis: Less regulation of CRAs is a better response. Journal of international banking law 2010, 25, 170.
186. Xu, X.; Pautasso, C.; Zhu, L.; Gramoli, V.; Ponomarev, A.; Tran, A.B.; Chen, S. The blockchain as a software connector. In Proceedings of the 2016 13th Working IEEE/IFIP Conference on Software Architecture (WICSA). IEEE, 2016, pp. 182–191.