Chapter-4-Shelley.md

Chapter 4: Shelley (Consensus)

‘If winter comes, can spring be far behind?’ ― Percy Bysshe Shelley

The Scalability Challenge

Distributed systems are made up of a group of servers¹ (nodes) that have agreed to execute a protocol, or a collection of protocols, to achieve a shared purpose. This objective might be as simple as distributing a file using the BitTorrent protocol,² or decentralized storage.³

As more nodes join the network, the most efficient protocols acquire resources. If several peers are simultaneously downloading a movie file provided by BitTorrent, for example, it may be downloaded substantially quicker on average. Because peers both contribute and use up resources, the speed increases. When someone says a distributed system scales, they’re usually referring to this feature.

The problem with most existing blockchain protocols is that they were not built to be scalable in the first place. Blockchains, for example, are often a linked list of blocks that can only be added to. A blockchain protocol’s security and availability are dependent on numerous nodes having a complete copy of the blockchain data. As a result, N nodes must copy a single bit of data. Additional nodes place a burden on resources required.

This is true for both transaction processing and gossip protocol⁴ across the system. Increasing the number of nodes in the consensus system does not increase transaction processing power. It simply indicates that more resources are required to do the same task. More network relaying means that more nodes must send the same messages to keep the whole network in sync with the most recent block.

Cryptocurrencies cannot grow to a worldwide network, comparable to older financial systems, due to their structure. Legacy infrastructure, on the other hand, is scalable and can handle orders of magnitude greater processing and storage power. Bitcoin is a relatively tiny network in comparison to its payment counterparts such as Visa, Mastercard, PayPal, etc and it is currently struggling to handle its present load.

The Ouroboros consensus protocol boosts scalability for Cardano. It allows for the decentralized election of a quorum of consensus nodes, which may then execute more conventional protocols to meet the demands of huge infrastructure providers like AWS, Google and Facebook.

For example, electing a quorum for an epoch implies there is a trustworthy group of nodes to keep the ledger up to date for a specified length of time. It is simple to elect many quorums at the same time and divide transactions among them.

What is a Consensus Protocol?

A consensus protocol is a collection of rules and parameters that regulate the behavior of distributed ledgers: a set of rules that each network member must follow. There is no one central authority in charge of public blockchains. Instead, a consensus mechanism is employed to enable dispersed network users to agree on the network’s history as recorded on the blockchain - to achieve agreement on what has occurred and proceed from a single source of truth.

A single record is provided by that one source of truth. This is why blockchains are frequently referred to as ‘trustless,’ since trust is built into the protocol rather than needing users to trust one another. Unknown actors may communicate and trade with one another without requiring the mediation of a third party or the sharing of personal data.

Consensus is the method through which everyone participating in the blockchain’s operation comes to an agreement. There must be unanimity on which blocks to generate, which chain to use, and how to establish the network’s single state. Individual nodes examine the current state of the ledger and form a consensus using the consensus protocol Ouroboros. It has three key responsibilities: doing a leader check and deciding whether a block should be created, handling chain selection, and verifying blocks that have been produced.

Blockchains achieve consensus by enabling users to group transactions submitted to the system into blocks and add them to their own chain (sequence of blocks). The objective of the various consensus protocols is to determine who is permitted to generate a block? When? …and what to do in the event of a disagreement? For example, what if two participants add different blocks at the same time? There are many different types of consensus protocols,⁵ the main two being proof of work and proof of stake. Ouroboros is a consensus mechanism based on proof of stake that has been shown to have the same level of security as proof of work.

Ouroboros

Ouroboros is a mythological creature represented as a snake, or dragon, devouring its own tail in a closed circle. The name ‘Ouroboros’ comes from Ancient Greek and literally means ‘tail eater’ or ‘tail devourer.’

Ouroboros is a symbol for infinity of time flowing back into itself in an endless circle, as if trapped in an unending loop. The first appearance of the Ouroboros was in Egypt in the 13th century BC. Alchemists later used Ouroboros in their mystical symbolism. Ouroboros has been understood and employed in a number of ways by many civilizations throughout history. One of the most prevalent interpretations of the symbol is that it signifies the Universe’s interconnection and infinity.

Charles Hoskinson named Cardano’s proof-of-stake consensus Ouroboros in 2017. In this sense, Ouroboros reflects the blockchain’s potential for endless growth and scalability. The core theme of Ouroboros is the provision of expanded possibilities for the planet, as well as its preservation through drastically decreased energy usage.⁶

Ouroboros was the first blockchain consensus system to be created via peer-reviewed research as a more energy efficient and sustainable alternative to proof of work, which is the foundation of previous cryptocurrencies such as Bitcoin and, until recently, Ethereum. Ouroboros and its various iterations give a new foundation for solving some of the world’s toughest issues safely and at scale.

Ouroboros ensures the security and sustainability of any blockchain that uses it by combining unique technology and mathematically validated methods (including behavioral psychology and economic philosophy ideas). Ouroboros has demonstrated to be secure and capable of facilitating the spread of global, permissionless networks with little energy consumption. Cardano is the first network of its kind. Ouroboros enables users - in this example, stake pools⁷ - to establish new blocks based on the amount of stake they own in the network, and thus enable the creation of a distributed, permissionless network.

The different implementations of Ouroboros

May 25, 2020. re: Ouroboros. CH:⁸

So if I had to succinctly say ..’well what is Ouroboros?’... I’d say it’s a first principles-based, new way of doing consensus that keeps the stuff we’ve come to know and love from Bitcoin, that’s very useful and great, but then does it in a much more sustainable energy-friendly way, that also allows you to layer on many more utilities than just mining to the system. It does so in a way that attracts lots of competition, lots of decentralization and lots of businesses. Mining tends to centralize to a small group of actors because of economy of scale

Ouroboros Classic

Ouroboros Classic was the first implementation released in 2017. It established the protocol as an energy-efficient alternative to proof of work, provided a mathematical framework for analyzing proof of stake, and presented a unique incentive mechanism for proof-of-stake users.

What set Ouroboros apart from previous blockchains and other proof-of-stake protocols was its capacity to provide unbiased randomness in the protocol’s leader selection method, as well as the security guarantees that came with that. Randomness prevents patterns from forming and is an important aspect of the protocol’s security. When a behavior can be expected, it may be manipulated — and although Ouroboros assures transparency, it prevents coercion. Ouroboros is notable for being the first blockchain technology to undergo such thorough security testing.

The research paper on Ouroboros has in depth explanations of its functionality. The blockchain is divided into slots and epochs by Ouroboros. Each slot in Cardano lasts 1 second, and each epoch (which is a collection of slots) comprises five days’ worth of slots.

The awareness that attacks are unavoidable lies at the heart of Ouroboros’ design. As a result, the protocol includes tolerance to prevent attackers from spreading other copies of the blockchain, and it assumes that an opponent may transmit arbitrary messages to any member at any moment. In reality, the protocol is guaranteed to be safe as long as honest players hold more than 51% of the stake.

Each slot has a slot leader who oversees adding blocks to the chain and passing them on to the next slot leader. To prevent hostile efforts to undermine the protocol, each new slot leader is obliged to treat the final few blocks of the incoming chain as transitory, with only the chain before the predetermined number of transient blocks being deemed resolved. This is also known as the settlement delay. This means, among other things, that a stakeholder may be offline and still be synchronized to the blockchain, as long as the latency isn’t longer than the settlement delay.

Each network node in the Ouroboros protocol keeps a copy of both the transaction mempool,⁹ where new transactions are inserted if they are consistent with current ones, and the blockchain. When a node becomes aware of a newer, more legitimate chain, the locally stored blockchain is replaced.

The disadvantages of Ouroboros Classic were that it was vulnerable to adaptive attackers — a real-world danger that was addressed in Ouroboros Praos – and that there was no safe means for a new member to bootstrap from the blockchain, which is addressed with Ouroboros Genesis.

Ouroboros BFT¹⁰

Following ‘Classic’ was the Ouroboros BFT paper in 2018 (deployed in May 2020). Cardano employed the Ouroboros BFT (Byzantine Fault Tolerance) protocol during the Byron reboot, which was the transfer of the old Cardano codebase to the new. Ouroboros BFT aided in the preparation of Cardano’s release of Shelley, with it, decentralization.

Rather than needing all nodes to be always available, Ouroboros BFT assumes a federated network of servers and synchronous communication between the servers, allowing for easier and more predictable ledger consensus.

Other advantages include immediate evidence of settlement, transaction settlement at network speed (i.e., the speed of your network connection to an OBFT node determines transaction settlement), and instant confirmation in a single round trip of communication. Each of these has a substantial impact on performance.

Ouroboros Praos¹¹ (current protocol at time of writing, Jan 2024)

The Ouroboros Praos paper was published in 2018 (deployed in August 2020) and is based on Ouroboros Classic, but with significant security and scalability enhancements. Ouroboros Praos, like Ouroboros Classic, divides transaction blocks into slots, which are then aggregated into epochs. Praos, on the other hand, is inspected in a semi-synchronous context and is safe against adaptive attackers, unlike Ouroboros Classic.

It presupposes two things: that adversaries may transmit arbitrary messages to any participant at any time, and those adversaries can delay honest participant communications for more than one slot. Praos assures that a powerful attacker cannot guess the next slot leader and conduct a targeted attack (like a DDoS attack) to pervert the protocol by using private-leader selection and forward-secure, key-evolving signatures. Praos can also tolerate adversarial controlled message delivery delays and gradual corruption of individual participants in an evolving stakeholder population, which is critical for maintaining network security in a global setting, as long as an honest majority of stakeholder population is maintained.

Ouroboros Genesis¹² (paper 2018, not deployed at time of press, but in development)

Ouroboros Genesis was the fourth version of the protocol and built on Ouroboros Praos by including a unique chain selection mechanism that allows parties to bootstrap from a genesis block — without the requirement for trusted checkpoints or assumptions about prior availability. Genesis also proves the protocol’s universal composability,¹³ demonstrating that it may be used with other protocols in real world configurations without weakening its security posture. This boosts its security and long-term viability, as well as that of the networks that use it. Genesis is in development and will allow the IOG relays to finally be turned off, which along with the peer-to-peer deployment, will complete the decentralization of the physical network.

Genesis is being implemented by Tweag (Tweag.io). Their Director of Engineering gave an explanation of Genesis at the Dubai Cardano Summit¹⁴

so it's effectively..if you're concerned about these ...what are called long range attacks. So attacks that would target the chain in the past, with the intent of having an impact on the future... now Ouroboros is is provably secure under certain conditions provided that the nodes are always online, but if a node joins the network, then those those guarantees don't apply, and so Ouroboros Genesis is about saying 'let's allow nodes to not join the network without needing to talk to a trusted relay'... so this is about decentralization

Ouroboros Peras will bring fast finality to Cardano. Block finality is about the point at which a block of transactions is considered permanent and cannot be changed or reversed. There is not much more detail about this at time of writing but we should hear more in 2024.

Ouroboros Hydra

Hydra is an off-chain scaling architecture that tackles three major scalability issues: large transaction output, low latency, and lightweight storage per node. The Hydra whitepaper proposes and details the inclusion of multi-party state channels,¹⁵ which provide parallel transaction processing to greatly boost Cardano’s transaction-per-second (TPS) output, as well as speedy confirmation of transactions. The paper refers to off-chain ledger siblings — state channels – as heads, reflecting the implementation’s namesake. This makes the ledger multi-headed.

Instead of scaling vertically by adding more powerful hardware, Ouroboros Hydra allows Cardano to expand horizontally, enhancing performance by including extra nodes. Early testing indicates that each head is capable of 1,000 TPS. Hydra is being developed in collaboration with the Ouroboros protocol and the Cardano ledger, although it may be used with other systems as long as they have the same properties as Cardano. Hydra was later decoupled from Ouroboros and became an open-source project¹⁶ in its own right. More about Hydra later in Chapter 8.

Consensus Redux

The Consensus Redux paper¹⁷ was published August 2020. This paper discusses the concept of a self-healing ledger. How does the ledger recover when it's been attacked? How does a network recover when it's been attacked? The paper outlines solutions in both categories.

Ouroboros Crypsinous¹⁸

The Crypsinous paper was published in 2019 but has yet to be implemented. Crypsinous is the first privacy-preserving proof-of-stake protocol. It boasts increased security features to protect against adaptive attacks including a coin evolution technique based on SNARKS¹⁹ and key-private forward-secure encryption.

Ouroboros Chronos²⁰ (Paper 2021, not deployed at time of press)

Chronos will deliver greater security and network resilience to communication delays by providing more accurate global timekeeping. To maintain the robustness of any dispersed network, global time synchronization is required. Time synchronization is critical in smart contract implementation, from guaranteeing up-to-date information amongst all participants to maintaining correct transaction processing and block construction.

IOG discovered a means to synchronize clocks across a blockchain in conjunction with experts from the Universities of Edinburgh, Purdue, and Connecticut to create a more secure and tamper-proof global time source. This includes time synchronization from internet of things (IoT)²¹ devices, such as supply chain monitoring tools, and general distributed systems, especially if a central clock failure poses a security issue. The research is implemented as Ouroboros Chronos (Greek for ‘Time’).

Chronos is a cryptographically secure blockchain protocol that also offers an accurate source of time thanks to an innovative time synchronization technique that avoids the flaws of externally maintained clocks. This also enables blockchain to precisely time-stamp transactions, making the ledger more resistant to time-based attacks.

By syncing local time to a uniform network clock with no single point of failure, the new protocol may greatly improve the resilience of essential telecoms, transportation, trade systems, and infrastructures. This scientific accomplishment also represents a huge step toward building completely auditable and fraud-proof financial systems by providing exact time and hence full traceability of all transactions.

September 22, 2020, re: Chronos. CH:²²

Ouroboros Chronos was a special paper. Leslie Lamport²³ was one of the first guys to do major work in distributed systems and he wrote this beautiful paper on clocks and time in the distributed system from the 1960s or 1970s. It’s one of the most cited papers of all time in the systems world… ‘Time, clocks, and the ordering of events in a distributed system’²⁴ and Chronos was like unfinished business in that respect. It’s fun to write papers like that and it has real use. You can completely decouple a dependency on NTP²⁵ and these types of things, there’s a lot of theory there.

Timing is everything

Within computer systems and applications, the idea of time is critical. You wouldn’t be able to access any transport layer security (TLS)-based websites, exchange data, or use other cryptographic techniques without this notion.

Time tracking, on the other hand, is a challenging issue to tackle. Accurate time synchronization requires data transfer throughout the whole internet, which costs time as well. It’s also difficult to forecast how long a particular data transfer will take since the network status is continually changing and is influenced by variables like congestion and data size, among others. As a result, discrepancies are common, and it’s critical to supply the tools and solutions necessary for accurate timekeeping.

It’s easy to take timekeeping for granted with basic PCs. Behind the scenes, however, there is a strict protocol to follow. The Network Time Protocol (NTP),²⁶ for example, uses a worldwide hierarchy of servers to solve the timekeeping problem. This comprises up to 15 Stratums, each with its own routing designed to synchronize in the most efficient way possible. The development of a Bellman-Ford shortest-path spanning tree,²⁷ which reduces latency and transmission time inconsistencies, also helps.

Time synching on the Blockchain

For distributed ledger technology, the idea of timekeeping is different. The network cannot verify that a transaction being processed is genuine, and does not reverse the prior one, without a correct timestamp. Different timestamping algorithms are employed across a variety of blockchain ledgers; however, they aren’t always reliable. Bitcoin, for example, employs timestamps for consensus security but not for timekeeping.

Timekeeping is also necessary for smart contract execution. Decentralized finance (DeFi) smart contract attacks are vulnerable to inaccuracy. Vulnerabilities in smart contracts aren’t necessarily caused by bad code; time discrepancies should be fixed to prevent any potential attacks on the ledger. Some blockchains, such as Solana, have experienced critical ‘clock drift’ issues.²⁸

Time for reflection

Building complex trading platforms, such as Axo (axo.trade), is not trivial when calculating exact times on a decentralized blockchain. The topic has been debated in IOG’s blogs, in the docs²⁹ and more heatedly on twitter. Although Cardano's current measure of time appears satisfactory to cover many use cases, it is not perfect. To dive deeper into the technical minutiae, it’s best to take in the views of several experts. Axo chief Jarek Hirniak, an experienced developer in trading software, wrote Time on Cardano,³⁰ while Sebastien Guillemot has covered ‘timing’ comprehensively in several of his YouTube videos.³¹

December 12, 2020. Is Chronos implemented? CH:³²

We'd like to be reliant on nothing and no one, and as decentralized as possible. So, we were the first to address this in the very first paper of its kind, with something called Ouroboros Chronos, but it would be an unnecessary diversion to just go and chase 3 months of implementation to pull Chronos in, when we don't need it as a system right now. […]

The other thing is that Chronos should be implemented with Consensus Redux, should be implemented with Genesis, and should be implemented with fast finality. and high throughput enhancements to the system, so Ouroboros 2 (subsequently named Ouroboros Omega) is a rollup of all the research that we've done which will make it considerably better than anything in market. So, the name of deployment execution is saying ‘what is a problem today? What's a theoretical but unlikely issue? Versus what is a practical and certainly problematic issue to resolve?’ …and understand how to balance things and create a roadmap accordingly.

Ouroboros Leios

The Leios paper³³ was published in November 2022 and is surprisingly readable. It was previewed in this video³⁴ by Professor Aggelos Kiayias. The focus of Leios is Input Endorsers which leverage the aforementioned foundational Ouroboros functionality as well as Mithril (See Chapter 8). As input endorsers are quite technical and based on concepts not discussed yet, they are covered in the last section of Chapter 8, Basho (Scalability). From section 6.2 of the paper:

Just as Ouroboros Praos+Genesis is a valid combination, Ouroboros Leios+Genesis makes perfect sense too and provides the same benefits. Thus in some sense Leios and Genesis are independent features. In practice Genesis will be deployed by the time that Leios is available and so it would be a regression if the Genesis feature were not included.

Ouroboros Omega³⁵ will be the capstone of all the proof-of-stake research IOG have accomplished since 2016. There should be a paper soon on the ‘convergence of all the ideas.’

February 9, 2021. Can you talk about Ouroboros Omega? CH:³⁶

Omega is the culmination of all of our research of the last 6 years for Ouroboros: no reliance on external clock, self-healing so it can gradually recover 51% attacks, the ability to bootstrap from Genesis, semi-synchrony, adaptive security, instant finality… all kinds of stuff…multi-validation per block… there's so much stuff …. Things like the consensus Redux paper, the Ledger Combiners paper³⁷ … the Chronos paper, the Genesis paper, the Praos paper and then all the theory and then some of the engineering acumen, including an improvement to about a thousand TPS (transactions per second).

How the Consensus Layer Works

Abstraction³⁸ is a key feature of the consensus layer.

The Cardano consensus layer is responsible for two major duties:

1. It uses the blockchain consensus mechanism to keep track of transactions. Consensus, or ‘majority of opinion,’ in the context of a blockchain, implies that everyone participating agrees on the one true chain. This means that the consensus layer is responsible for accepting blocks, selecting between rival chains if any exist, and determining when to create its own blocks.

2. It is in charge of preserving all of the information needed to make these judgments. The protocol must verify a block in relation to the state of the ledger before deciding whether to accept it. It must preserve enough history to be able to rebuild the ledger state on a different chain (a different point of a fork in the chain) if it chooses to move to a different chain. It must maintain a mempool³⁹ of transactions to be placed into those blocks to be able to create blocks.

Figure 4.1: Cardano Node component parts, drawing by @ktorz

The Cardano node is made up of several interconnected layers:

• The networking layer is based on a P2P (peer-to-peer) networking stack tailored to Ouroboros consensus protocol. This includes a framework which supports pipelining, multiplexing and various protections against adversarial peers.

• The consensus layer is an implementation of the Ouroboros family of protocols.

• The settlement layer is a multi-era ledger implementation derived from formal specifications. This is where the core Cardano entities are defined as well as the rules for using them. This is the bedrock on top of which consensus and networking are built upon. The settlement layer is completely stateless⁴⁰ and only contains pure functions. As a result, it isn’t necessary for the consensus layer to understand the precise nature of the ledger state, or even the contents of the blocks (apart from some header fields required to run the consensus protocol).

• The scripting layer, Haskell code on the Plutus Platform (subject of a later chapter) is a scripting language embedded in the Cardano ledger to provide smart-contract capabilities to the network.

Consensus Roles

The consensus protocol has three major roles:

1. Leader check (who should produce a block?)
2. Chain selection
3. Block verification

The protocol is designed to be independent of a specific block or ledger, allowing a single protocol to be used with a variety of blocks and/or ledgers. As a result, each of these three roles establishes its own ‘picture’ of the data.

Every slot has a leader check that determines whether the node should create a block. In practice, the leader check may need the extraction of certain information from the ledger state. The chance of a node being elected as leader (authorized to generate a block) in the Ouroboros Praos consensus protocol, for example, is dependent on its stake, the node’s stake.

The process of selecting between two competing chains is known as chain selection. The chain length⁴¹ is the most important selection criteria here, however other protocols may have extra requirements. Blocks, for example, are usually signed using a ‘hot’ key that lives on the server and created by a ‘cold’ key that never appears on any networked device. If the hot key is compromised, the node operator may create a new one using the cold key and ‘delegate’ to it. A consensus protocol will favor the newer hot key over two chains of similar length, ie. Each chain includes a tip⁴² signed by the same cold key but a different hot key.

Block validation is primarily a ledger problem; checks such as ensuring all transaction inputs are accessible to prevent double spending are specified in the ledger layer. What’s within the blocks is mainly unknown to the consensus layer; in fact, it may not even be a cryptocurrency, but a distinct use of blockchain technology. IOG used the example of a Pokémon ledger⁴³ on Ouroboros previously. However, block headers include a few items that are expressly designed to aid the consensus layer.

Node configuration

To execute, each protocol may need certain static data, such as keys to sign blocks, a unique id for the leader check, and so on. This is referred to as the ‘node configuration.’

The ledger state

The consensus layer not only handles the ledger state but also operates the consensus protocol. It is unconcerned with the look of the ledger state; instead, it assumes that some form of ledger state is connected with a block type.

If a block is invalid, you may get an error while applying it to the ledger state. The ledger layer defines the precise kind of ledger errors, which are ledger specific. The Shelley ledger, for example, will have staking errors, but the Byron ledger would not since it does not enable staking; and ledgers that aren’t crypto ledgers will have quite different errors.

The Cardano consensus layer was created with the Cardano blockchain in mind, which initially ran Byron and was updated to run Shelley. It’s fair to ask why didn’t IOG developers create for that particular blockchain first? Then generalize when utilizing the consensus layer for other blockchains? However, there would have been significant drawbacks in doing so:

• It would obstruct IOG’s capacity to conduct tests. They wouldn’t be able to choose which nodes create blocks when, they wouldn’t be able to use a dual ledger, and so on

• It would entangle things that should be logically separate. The Shelley ledger, in the abstract method, is made up of three parts: a Byron chain, a Shelley chain, and a hard fork combinator that connects the two. Without abstractions, such separation of concerns would be more difficult to establish, resulting in code that was more complex to comprehend and maintain

• Abstract code is less likely to include flaws. For example, since the dual ledger combinator is polymorphic⁴⁴ in the two ledgers it combines, and they are of different types, IOG couldn’t construct type correct code that attempts to apply the main block to the auxiliary ledger

• When the time inevitably comes to instantiate the consensus layer for a new blockchain, writing it in an abstract manner from the start forces IOG to think carefully about the design and avoid coupling things that shouldn’t be coupled, or making assumptions that may or may not be true in general. It might be difficult to fix such issues once the design has been implemented.

To achieve all of this, a programming language with exceptional abstraction capabilities is required, and Haskell is well equipped in this regard.

January 10, 2021. Please explain Cardano being 100x more decentralized than BTC or others, how is this calculated? CH:⁴⁵

It’s calculated by those who produce blocks, so if you have a hundred times the unique entities producing blocks, then we say it’s a hundred times more decentralized, but there are many measurements of decentralization. You can look at network propagation, so how many full nodes you have. You can look at unique development entities responsible for it …you can look at the funding sources and see how many unique funding sources are there. You can see the totality of the user count…

You can see it by the total applications on the system but usually when we say decentralization, we strictly mean the amount of nodes participating in the consensus process and how many unique people are making blocks. In the case of bitcoin, 3 to 5 usually make the same blocks again and again… and we consistently have 300 to 500 stake pools consistently making blocks that are unique. There’s over 1200 registered (currently over 3,000), so I feel very comfortable in the 100x statement, but reasonable people can have different definitions, there’s no consensus in the industry of what decentralization is.

Hard Forking Business

In the context of the blockchain, a ‘hard fork’ refers to a significant change in the chain, such as switching from one protocol to another. A hard fork in most blockchains denotes block modifications or a change in their interpretation. Typically, when a hard fork is performed, the existing protocol is turned off, new rules and modifications are introduced, and the chain is restarted. A hard-forked chain will be distinct from the prior version, and that the pre-forked blockchain’s history will no longer be accessible.

The Cardano blockchain has hard forked from a federated Byron model to a decentralized Shelley model. This hard fork, on the other hand, was one-of-a-kind. IOG guaranteed a seamless transition to a new protocol while maintaining the history of earlier blocks, rather than making major modifications, the chain did not alter drastically. Instead, it included Byron blocks before adding Shelley blocks after a transition time. There was no ‘turning it on and off again’. The entire history was retained. There has been no downtime or restarts with Cardano, which is not always the case with other chains.⁴⁶

Hard Fork Combinator

In Cardano docs and blogs, you may have seen a hard fork referred to as an ‘HFC’ or a ‘HFC event’. A combinator is a technical term that refers to the joining of two or more processes. In the context of Cardano, a hard fork combinator combines protocols, allowing the Byron-to-Shelley transition to be completed without the need for a system restart. It guaranteed that the ledgers of Byron and Shelley display as a single ledger. It was not necessary for all nodes to update at the same time when switching from Ouroboros BFT to Ouroboros Praos. Instead, nodes updated in stages; some running Byron blocks, others running Shelley blocks.

The hard fork combinator is intended to allow the integration of many protocols without requiring major changes. Byron and Shelley blocks are now combined in the Cardano chain. For future ‘hard fork events’, Basho and Voltaire blocks will be combined as well - all as a single property. By simplifying the prior Byron-to-Shelley evolution, this (HFC) hard fork combinator also made the transfer from Shelley to Goguen seamless with gradual, iterative upgrades rolled out as the Allegra, Mary and Alonzo updates.

HFC history to date

The Cardano platform entered a rapid development phase with the introduction of the Incentivized Testnet in 2019, which marked the beginning of the Shelley era. The Ouroboros Classic consensus system previously supported Byron and ada, and subsequently migrated to Ouroboros Praos. As Cardano decentralized, this was the version of proof-of-stake (PoS) protocol that first powered Shelley. It incorporated monetary rewards into the staking procedure for ada holders and stake pool owners.

In February 2020, IOG upgraded Cardano with a hard fork that moved the mainnet from Ouroboros Classic to Ouroboros BFT, an enhanced version of the original consensus mechanism. This BFT hard fork kicked off a transition phase under Ouroboros BFT, a pared-down version of the protocol aimed to ease the transfer to Praos while still avoiding malicious behavior. Users were unlikely to have noticed. It meant a routine software update for Daedalus wallet users. Exchanges were required to update manually, but they had many weeks to do it and IOG were available to assist.

The ‘Byron reboot’⁴⁷ followed that in March 2020. Many Cardano components received completely new code, including a new node to handle delegation and decentralization, as well as future Shelley features. The new code base was redesigned to be modular, which meant that many components could be updated without impacting the others.

In turn, the BFT served as a springboard for the Shelley hard fork, which happened after the Haskell testnet was complete. For exchanges, ada holders, and wallet users, this second hard fork was similar to the first, a non-event in terms of user disruption. However, although everything seems to be in order on the surface, there was a lot of activity going on behind the scenes. IOG’s developers were hard at work making it a seamless, benign experience for the end user. IOG chief architect Duncan Coutts⁴⁸ explained at the time:

IOG’s blockchain engineers believe in smooth code updates. Instead of trying to do the jump from Ouroboros Classic to Praos in a single update – which would be an incredibly complex task – it’s been a two-stage approach using Ouroboros BFT as an intermediary. The BFT code is compatible with both the Byron-era federated nodes and the Shelley-style nodes released in the Byron reboot. It’s like a relay race: one runner (in our case, running one protocol) enters the handover box where the other runner is waiting; they synchronize their speeds (so they’re perfectly compatible with each other) and then hand over the baton (operating the mainnet), and then the new runner with the baton continues from the handover box for the next lap.

IOG were able to swiftly design and test a new wallet using Daedalus Flight, and once everyone was using it on the mainnet, and IOG finished changing over the core nodes, the old code was obsolete. Flight is IOGs version of Chrome Canary for Daedalus.

In summary, the only real hard fork for Cardano was the switch from Ouroboros Classic to BFT (February 2020). The Byron mainnet was relaunched to run the BFT protocol, allowing for a smoother transition to Ouroboros Praos with fewer chain disruptions. The BFT protocol was meticulously built to preserve blockchain history and make the blockchain look as a single entity.

As IOG chief executive Charles Hoskinson discussed in his whiteboard video about the hard fork,⁴⁹ the goal was to have a ‘graceful entry into Shelley.’ The hard fork combinator was crucial in accomplishing this transition.

Goguen Era Updates

Allegra (Token Locking)

Token locking was a feature introduced to the Shelley protocol to allow a variety of smart contract use cases, such as generating and transacting with multi-asset tokens and adding support for the Voltaire voting mechanism. Token locking is the act of reserving a certain number of assets and agreeing not to sell them for a given length of time. This functionality was enabled in the Allegra⁵⁰ upgrade in December 2020. Allegra was named after Allegra Byron, Lord Byron’s daughter and sister to Ada Lovelace.

Token locking allowed for more complicated deal settlement and fund accounting. It’s used in the following scenarios:

• Contractual agreement - when someone engages into a contractual agreement, such as to sell an asset like a painting, it is essential to guarantee that the painting will not be sold to anyone else save the person who pays the money. The token may represent both the painting and the ‘promise’, the actual token locking in this example. The contract becomes invalid if the painting is sold to a different third party

• Vote registry - Token locking allows users to lock a set quantity of their tokens to reflect their voting rights inside the Voltaire voting system. Holders of ada tokens who participate in the voting process must ‘lock’ their tokens. This will reflect their voting rights in proportion to their investment and will prevent the hazards of double-counting votes, awarding more votes than is feasible, contradicting votes, or vote duplication

• Multi-asset tokens - In addition to ada, Cardano allows for multi-asset tokens, with the ledger supporting the creation and usage of several bespoke token types. Token locking enables ada tokens to be ‘locked’ to create a bespoke asset of equal value.

Mary (multi-asset support)

This then paved the way for the Mary (multi-asset support) upgrade in March 2021. It was named after Mary Shelley (the author and wife of Shelley). MAS (multi-asset support) means you can record that a particular token is being used for a specified purpose. The token may represent any object that is recorded on (native to) the blockchain ledger, such as ada, and other custom tokens. Mary enables users to generate (custom) tokens with unique characteristics and conduct transactions with them directly on the blockchain.

The ledger’s accounting architecture now handles not just ada transactions, but also transactions that hold several asset types at the same time. Developers benefit from native support since they don’t have to write smart contracts to handle bespoke token minting or transactions. Instead, the accounting ledger keeps track of asset ownership and transfers, minimizing unnecessary complexity and the risk of human error while also ensuring considerable cost savings.

To fulfill commercial or business goals, developers, enterprises, and apps can build general purpose (fungible) or specialized (non-fungible) tokens. Custom payment tokens or rewards for decentralized apps, stablecoins pegged to other currencies, and unique assets that represent intellectual property are just a few examples. All of these assets may then be traded, swapped, or used to purchase goods and services.

Alonzo (smart contracts)

Still part of the Goguen era, Alonzo was the next protocol update in Sept 2021. To enable functional smart contracts, Alonzo built on top of transaction metadata, token locking, and native asset functionality. By allowing the development of smart contracts and decentralized apps (dApps) for DeFi (decentralized finance), this update created a diverse platform that opened up options for enterprises and developers.

This functionality is available and enabled by the tools and infrastructure that make up the Plutus Platform. Alonzo enhanced Shelley’s basic multi-signature (multisig) scripting language with a rigorous methodology based on formal methods and verification. For more sophisticated and secure scripting capabilities, Multisig was updated into the Plutus Core language.⁵¹ Alonzo enables this through extended unspent transaction output (eUTXO) accounting. More on this later.

Alonzo was named after Alonzo Church (1903-95). Church was a logician and mathematician who worked on logic and the foundations of theoretical computer science in the US. He is also recognized for establishing lambda calculus,⁵² a formal system that may be used to argue that the Entscheidungsproblem⁵³ is unsolvable. Later, when working with Alan Turing, they realized that the lambda calculus and the Turing machine⁵⁴ had equivalent capabilities, displaying numerous mechanical computing processes. Plutus Core (Cardano’s smart contract language) is a variant of lambda calculus, which is one of the reasons for naming the smart contract upgrade after Church.

Vasil

Named after a Bulgarian mathematician Vasil Dabov, who was also a Cardano community member, the Vasil hard fork in Sept 2022 brought five key features to boost performance. These were mainly focused on Plutus V2 after CIPs (Cardano Improvement Proposal) were submitted by the community. CIP-31 (Reference Inputs), CIP-32 (Inline Datums), CIP-33 (Reference Scripts), CIP-40 (Collateral Outputs), and diffusion pipelining are discussed later.

The terminology and the use of the same terms in different contexts can get quite confusing. Fear not, as there is page the documentation⁵⁵ explaining the difference between an ‘era’ and a ‘phase’, and how a ‘hard fork’ differs from an ‘intra-era hard fork’. Starting with Alonzo, the ‘ledger eras’ are named after mathematicians and computer scientists, kind of like the hurricane naming system but in alphabetical ordering. The release dates are named in honor of Cardano community members.

Figure 4.2: Cardano Phases, eras, and intra-era hard forks

The SECP update in Februrary 2023 was executed as an ‘intra-era hard fork’. This brought new cryptographic primitives to Cardano. More about this update in chapter 6.

June 5, 2020. What is the best code in Cardano? CH:⁵⁶

The hard fork combinator code is the second most elegant code we have; the most elegant code belongs to the network code. We have lots of crazy stuff there, very elegant, very academic scary code …but beautiful.

Path to Full Decentralization

Cardano hit a milestone at the end of March 2021 when d, the parameter that determines what proportion of transactions are processed by the genesis nodes, reached zero. The duty for block creation was totally decentralized at this time. Cardano’s network of 1,800 (March 2021) community pools were fully responsible for the generation of blocks.

The day d reached 0 was a watershed point for Cardano. When IOG released the Shelley update in July 2020, they set d to 1.0, which meant that every block was generated by IOG’s federated nodes. Of course, this was the opposite of decentralization, but it was a prudent (secure) strategy in the short term until the stake pool operator (SPO) network was set up and mature.

IOG steadily lowered d at a rate of 0.02 each epoch over time, an increase of two percentage points in community block production every 5 days. When d dropped, the community created more blocks, and more stake pools were able to generate blocks. The network’s diversity and geographic dispersion grew as d decreased.

d hit zero on March 31, 2021, at the end of epoch 257. That day was memorable because, although d is a modest number, its importance is enormous. That zero was the most crucial external signal of decentralization, a parameterized symbol confirming a major principle of Cardano’s ideology of pushing power to the edges.

Trickier parameter adjustments

Setting stable parameter settings, while still being flexible for the future, is critical to Cardano’s continued development and decentralization. While the transitioning the d parameter was smooth and predictable, changing other parameters requires a more nuanced approach. IOG consulted with the community before setting initial values. Around 20 parameters⁵⁷ regulate network behavior, and settings had to be specified for all of them before the mainnet could be launched. The majority of these parameters are technical in nature, so although they must be set appropriately to ensure system safety and improve performance, their specific values have little impact on user experience. You can review the full list of current parameter settings on ADApools.org.⁵⁸

June 4, 2019. Re: Ouroboros parameters. CH:⁵⁹

What is the protection mechanism against too many stake pools? So, you have to delegate to the stake pools …so there’s only a finite amount of ada and the financial incentives are basically set up in a way where you will have a ceiling of stake pools, and if you have more than that, you actually make less money. So, I’d highly encourage you to read our paper, we wrote out of Oxford with Alexander Russel.⁶⁰ The paper covers how you parameterize that model and that’s why we can believe that there’ll be a stable thousand stake pools after a while.

K = number of stake pools

A crucial element is the desired number of stake pools, or the k parameter. k is the reward scheme parameter that determines the soft limit on the pool size. Cardano incentives have been intended to foster an equilibrium with k completely saturated pools, which means that when all stake is delegated equally to the k most appealing pools, rewards will be optimum for everyone. At equilibrium, assuming rational participants and no external influences, the mechanism is constructed such that the stakeholders’ optimal response behavior converges to k pools of equal size, each delivering the same amount of rewards per unit of stake to their delegates.

The greater the value of k, the more decentralized the system. More k, on the other hand, results in a less efficient system (higher expenses, greater energy consumption) and fewer returns for both delegators and stake pool owners. IOG knew that the community would be motivated to build up pools based on what IOG learnt from both the Incentivized Testnet (ITN) and the Haskell Shelley testnet.

Decentralization occurred fast but decentralization alone is insufficient. Cardano requires long-term commitment from its operators, and operators must be appropriately rewarded for continuing to sustain the system. The initial proposal was k=150 for the mainnet deployment of Shelley, to be progressively raised to create a balance between decentralization and rewards for stake pool managers. This was a little improvement over the parameter choice in the incentive testnet (ITN), which was k=100. This was seen to be a cautious approach at the time, aimed to enable a seamless transition of the ITN environment to the mainnet. This guaranteed that the system was stable and efficient at first and could progressively evolve over time to become more decentralized, and secure, in the future.

Cardano is an order of magnitude more decentralized than any other blockchain due to k stake pools of roughly similar size. And this was only the start. Tiny, progressive increases in the k-parameter aren’t possible (unlike, for example, the decentralization d parameter, which lent itself to a gradual reduction). Each rise in k necessitates action on the part of pools and delegators. For pool operators, this entails fine-tuning their settings, particularly their margin; for delegators, it entails selecting new pools to delegate their ada to, particularly if their existing selection becomes oversaturated.

As a result, the optimal method for increasing k is to do it in bigger, less frequent increments – and as quickly as realistic network dynamics and economics will allow. The optimum option is one that causes the least amount of disturbance to successful pools and their delegators while increasing the chance for medium and smaller pools to mint blocks and attract additional stake. It’s also critical to have a long-term strategic aim in mind as a community: to spread decentralization as broadly as possible.

The k = 500 change

At epoch 234, an adjustment to k=500 was made on December 6, 2020. Small-to-medium-sized pools that were having trouble attracting delegation benefited from the change to k=500. It also restricted pool size to 64 million ada, meaning that more than 100 of the biggest pools would overfill.

It’s always best practice for delegators to monitor their preferred pools. You should redelegate if you see your chosen pool becoming oversaturated. It’s crucial to remember that rewards will still be paid out from relatively saturated pools, but they’ll decrease when the pool’s saturation rises. No one who delegated to an over-saturated pool ever lost any ada. It’s only that if one remains in a saturated pool, the return on investment will be diminished.

Monetary Policy

Transaction fees and monetary expansion are used to fund staking rewards for both delegators and stake pool operators. Every epoch, all transaction fees from all transactions from all blocks created during that epoch are collected and placed in a virtual ‘pot.’ A predetermined proportion of the leftover ada reserves, ρ, is also contributed to the pot. Then a portion of the pot, τ, is paid to the Treasury, and the remainder is used as epoch rewards. ρ (Rho) and τ (tau) are letters of the Greek alphabet…the chief scientist at IOG, Prof Aggelos Kiayias, is Greek which might explain the naming policy.

This approach guarantees that the share of rewards taken from the reserves is large in the beginning, when the number of transactions is still relatively modest since users are only starting to create their businesses on Cardano. Early adopters will be enticed to act swiftly to reap the benefits of the high initial payouts. The higher costs compensate for declining reserves over time as transaction volume grows.

This technique also guarantees that the rewards provided are predictable and evolve over time. There will be no ‘jumps’ like the bitcoin halving⁶¹ occurrences that occur every four years. Instead, a steady exponential fall is guaranteed by a predetermined proportion deducted from residual reserves every epoch.

So, what kind of value should ρ take? What percentage of the budget should go to the Treasury? This is also another trade-off: larger values of ρ indicate bigger rewards for everyone at first, as well as a faster-filling treasury. However, greater levels of ρ imply a quicker depletion of reserves. Paying big incentives and incentivizing early adopters is crucial, however, it is equally critical to present all stakeholders with a long-term view.

Cardano will never run out of reserves; instead, it will be an exponential decay. Calculate the ‘reserve half-life,’ or the time it takes for half of the reserve to be used up, to obtain a sense of the effect of a certain value of ρ. After careful consideration, 0.22% was the proposal for ρ. When the numbers are crunched, the ‘reserve half-life’ comes out to be roughly 4 to 5 years. In other words, half of the remaining reserve will be consumed every four to five years. Because this is near to the four-year ‘bitcoin half-life,’ Cardano holdings will diminish at a similar pace to bitcoin reserves.

It’s worth mentioning that Bitcoin took almost eight years to attain its highest level of popularity and pricing. As a result, it is reasonable to anticipate Cardano transaction volume and exchange rate to rise substantially over the coming years to more than compensate for the reduction in monetary expansion.

Funding the Treasury

IOG also recommended a starting figure of 5% for τ, the proportion of rewards that are automatically sent to the Treasury at the end of each epoch. Over the following five years, at least 380m ada will be transferred from the reserves to the Treasury.

The actual amount coming to the Treasury will be much larger. To begin with, it’s unrealistic to anticipate that all ada will be delegated based on learnings from the Incentivized Testnet, but also anticipating future usage of ada. Some of it will be stored on exchanges, traded, and used in smart contracts. Unclaimed rewards will result from the ada that is not delegated. The treasury will salvage these unclaimed rewards.

It’s not expected that pools’ pledges will be overly large, just high enough to make launching a Sybil attack undesirable. The difference between prospective pool rewards with a very high pledge and pools with a more realistic pledge level, ends up in the treasury as well. For the foreseeable future, the total of all ada going to the Treasury indicates that there will be enough funds to pay for new features and upgrades.

Pledge Influence Factor (a0)

The ada pledged by pool owners protects against Sybil attacks by ensuring that delegated stake is not overly drawn to pools whose owners attempt to abuse the system by forming multiple pools without owning a substantial amount of stake themselves. There was a lengthy debate, at the time, on this episode of the Cardano Effect.⁶² The debate has continued for some time about why, how and by whom, the parameters should be adjusted. Umed Saidov (who wrote this analysis⁶³ in 2020) and ‘Big Pey’ laid out their case for change in this passionate discussion⁶⁴ at the Rare Bloom event.

The pledge influence factor has a direct impact on a pool’s rewards: the higher the influence factor, the greater the impact of a bigger pledge on rewards. A greater influence factor improves Sybil protection and makes the system safer and more secure, but it also offers stake pool owners who can afford a bigger pledge an advantage.

The initial value of 0.3 was intended to strike a compromise between the amount of Sybil protection and the pledge needed. However, there is no minimum pledge. The pledge may be made as low or as high as the pool operator decides. Their choices impact the rewards, but there is no ‘hard’ rule requiring them to pledge a certain amount. This means that pool pledges will eventually go as high as pool owners are willing to go, and it will be up to the community to strike a balance between security, economic concerns, and the desire for justice and equal opportunity.

Pledge deferred update

SPOs that concentrate their pledge into a limited number of pools are rewarded with the a0 parameter. This has worked well in encouraging pools with high levels of ada pledge to combine into big private pools (like IOG do), giving smaller pools a better chance to recruit delegators. However, IOG felt the existing method could be improved, therefore debated and modeled ideas to make pledge more successful at resolving pool splitting for lower pledge levels.

While there has been much internal debate, IOG came to the conclusion that any change in k should be only made after revising the formula for a0 to get the desired outcomes (especially encouraging stake to flow to smaller single pools rather than split pools). Because this was a complete formula change rather than just an epoch boundary change, it had to be issued as part of a hard fork. IOG planned to make this adjustment for some time but given their product pipelines and their team’s focus on the Goguen rollout, the update has not occurred yet and pledge influence factor a0 remains at 0.3.

As things stand (Jan 2024)

Different parameters are used to as lever to control Cardano’s protocol behavior. Some parameters are predictable and go unnoticed and largely left alone. Others are more prominent as they directly impact on-chain activities and need regular tuning as the protocol moves out of testnets and hard forks change the playing field conditions for stake pool operators and delegators.

Two variable parameters continue to divide opinion among the SPOs community as they relate to the core principle of Cardano, decentralization, and they impact their ability to earn a living. These are the k value (the number of stake pools who receive rewards), and the minimum pool fee, which is a set amount paid to a pool before rewards are split with delegators.

k (currently 500) impacts the max pool size, based on the equation: = (45bn - reserves) / k = ( 35bn / 500 ) = 70m

IOG conducted a community survey⁶⁵ to ‘read the room’ but there was not consensus on what changes to make, or how to implement them. IOG previously proposed raising k to 1000, but this was postponed as the survey was not clear cut, IOG were busy with the Vasil hard fork and it was decided more planning and discussion was required.

The MinPool fee has also been an emotive topic. SPOs are the engine room for Cardano operationally. They produce new blocks and ensure a decentralized network grows. They are paid for their work in two ways:

• a fixed fee, a set amount taken if blocks are produced, and
• a variable fee, a percentage of the total rewards paid to a pool.

Staking pools can chose to set any variable fee they like, but the ledger rules has a minimum fixed fee in place. This was there originally as defence against Sybil attacks (an adversary might hijack the network by creating a lot of zero-fee pools). Many feel this protection is no longer needed, including IOG who admit this risk may be a more theoretical than practical at this stage as delegation is so decentralized.

A big reason to reduce, or remove it, is it is an obstacle for small new pools to compete with the bigger, more established stake pools. This can be a double whammy as the minimum fee can also incentivize larger SPOs to create new pools since there is a guaranteed profit from doing so.

To get a sense of how emotive these issues are, you should listen to the Twitter Space hosted by Bullish Dumpling⁶⁶ and read Adam Dean’s twitter thread.⁶⁷ Army of Spies dedicated a video⁶⁸ to it and IOG have written a blog post⁶⁹ mapping out potential courses of action. We haven’t even mentioned CIP 50 (Liesenfelt Shelleys Voltaire Decentralization Update) which is worthy of a book in itself. I suspect these decisions are deliberately being left to test-drive the new on-chain governance mechanisms. These new levers of power are explored in Chapter 9. Charles Hoskinson all but confirmed this on a recent AMA:⁷⁰

It's not about IOG changing parameters, the system is stable at the moment, it would be nice to get all the parameters and tune them on a regular basis, but that's the entire point of CIP 1694. You guys should be doing that. If you want to sign up and have a custodian sit down and think about how to tune the system every month, so that it promotes some outcome, I think that's not the point of Cardano, the point of Cardano is the Cardano community self-regulates Cardano. So what we did is we got the system to a stable state, with that stable state it's able to survive and thrive, the next step is to get the community to take that stable state and create a government on top of it. So if you want to change the system parameters, then go ahead and get something like CIP 1694 in. Obviously, there'll be some fine tuning throughout this year, there's some final things to do as things change, as peer-to-peer gets fully implemented and Genesis gets fully implemented, but it defeats the entire purpose of decentralization to have a tuning committee that's federated

In May 2023, the Cardano Foundation conducted a SPO on-chain poll,⁷¹ a new mechanism (defined in CIP-94)⁷² to allow stake pool operators to vote on crucial issues based on the stake delegated to their pool. Shortly after the poll, ‘minPoolCost’ was subsequently cut to 170 ada. It was decided it would be risky to change the k parameter at the same time. More about the governance around protocol updates in Chapter 9.

Ouroboros lays the foundation for the Basho Era

Considering all the features and functionality introduced by the hard forks we just talked about, it's not a surprise that network activity would pick up, but Cardano’s design will enable it to bend and adapt as needed.

Network bandwidth

All Cardano activities are built on top of networking. The Cardano network distributes transactions and blocks among nodes all around the globe that build and validate the blockchain. This is known as data diffusion, and it is necessary for the consensus protocol to make judgments by providing the necessary information to nodes. These choices move the chain forward, since node consensus guarantees that all transactions are confirmed and approved, allowing them to be included in a new block in a transparent manner.

The speed with which the system as a whole operates is influenced by network performance. This covers things like:

• throughput (amount of data transferred)
• timeliness (time for block adoption)

These two needs are in direct opposition to one another. When the created blocks are used as effectively as possible, throughput can be increased. This, in turn, requires adequate buffering to offset latency, reducing the negative effects of a globally distributed system. When the system is saturated, more buffering may frequently suggest greater block (and network) usage, but it comes at the expense of higher latency (time to adoption in the chain).

A block’s budget

To comprehend how quickly Cardano transactions and scripts may be completed, the concept of the block budget is key. A block’s total size strikes a balance between maximizing network use and reducing transaction latencies. A single block may include a variety of transactions, including smart contracts written in Plutus, native tokens, metadata, and straightforward ada transactions (payments).

Another attribute is the block time budget, which is a set amount of time allowed to complete all of the transactions in a single block. This is split between the amount of time available for Plutus script execution and the amount of time available for other transactions. This attribute guarantees that transactions containing Plutus scripts do not consume the available time budget, and that basic payments may always be processed in the same block as Plutus scripts.

The entire time budget for creating each block (including networking overhead) is set to 1 second, with a Plutus script execution budget of around 50 milliseconds. This is a reasonable provision; IOG’s testing has demonstrated that many scripts on a testing environment will run in 1 millisecond or less.

The block time budget was set initially at 1 second. Due to security concerns, the Praos consensus process picks just a small proportion (1 in 20) of the blocks that might possibly be added to the chain. Obviously, the size and effective payloads of various transactions will change. A single transaction, for example, may close off an entire Catalyst voting round, moving millions of dollars in value.

As previously stated, each block contains a number of transactions submitted by end users through wallets, the command-line interface (CLI), and other means. These transactions are stored in the mempool, a temporary in-memory storage space, until they are ready to be processed and included in a block. As a block is minted, pending transactions are removed from the mempool, allowing new transactions to be added. The potential of nodes being overwhelmed during high-demand times was eliminated by employing a fixed-size mempool, although this meant that a wallet or application may have needed to re-submit transactions. The mempool size was determined using queueing models.

Stress testing the network

Ouroboros is built to manage massive amounts of data, as well as transactions and scripts of various sizes and complexity. With the settings in place the Cardano network was still only using around 25% of its capacity on average in October 2021. Of course, the most efficient option is for Cardano to operate at or near 100% capacity (network saturation). While many networking systems would suffer in such circumstances, Ouroboros and the Cardano network stack have been engineered to be fair and very durable even in the face of extreme saturation.

Benchmarking data reveals that even at 200% saturation, overall performance remains stable, and no network failures occur. Even with stress tested at 44x, there were no problems in the entire network capacity (though some transactions may be slightly delayed). Backpressure⁷³ is used to regulate the overall system load, which is how the network is meant to function.

While certain users participating in a big NFT drop⁷⁴ may suffer lengthier wait times for their transactions or may need to resubmit the rare transaction from a large batch (or spread the drops over a longer time period), this does not indicate that the network is not coping. It really signifies that the network is working properly. It’s referred to as ‘graceful degradation,’ and you can understand in more detail by reading the network design paper.⁷⁵ See ‘Cardano Upcoming NFT drops.’⁷⁶

Wallet types

End-user wallets submit payments and other transactions to the blockchain on their behalf, as well as monitor the blockchain’s progress. One of the most important functions of a wallet is to submit transactions on behalf of the user, validate that they have been approved into the blockchain, and retry them if the first attempt fails. That is, when the network saturates, the wallet should consider the implications of backpressure as well as other network effects (temporary disconnection, possible chain forks, etc). Wallets may be one of two types:

• Full-node wallets (like Daedalus), which operate a node that connects directly to the Cardano network using local compute and network resources

• Light wallets, on the other hand, take advantage of pooled processing and networking resources to service a large number of users.

Both kinds of wallets may need to retry transactions during times of strong demand (e.g., an NFT drop). Light wallets may need to temporarily scale available compute and network resources to meet user demand since they share resources across numerous users. Full node wallets, on the other hand, may be unaffected. Transactions may be somewhat delayed, but each wallet will have the dedicated resources, including its own network connections, to attempt the submission. dApp providers should follow similar principles: if particular network endpoints are offered, system resources should be adjusted to match demand.

Chain v transaction confirmation

‘What is Cardano’s TPS (transactions per second)?’ is one of the most common questions on Crypto Twitter. How many network confirmations does Cardano require before a transaction goes through? The solutions to these concerns necessitate a more in-depth examination of the principles of chain confirmation and transaction confirmation, as well as their relationship to the protocol.

Chain confirmation

This is the threshold at which the protocol guarantees that the chain will not change any further due to randomness or random occurrences. After enough future k blocks have been issued, chain confirmation happens at some point in the future. The stability window is the period between now and when chain confirmation for a certain transaction happens (that is, the number of slots required for a block to become stable, where stability is defined as a block that cannot be rolled back). The formula for calculating this window is:

        3k/f 

• where k is the parameter that restricts a pool’s growth by lowering its rewards yield beyond a particular threshold,
• f is the parameter that determines a pool’s maximum size

Transaction confirmation

When a transaction is accepted into the chain, it becomes immutable at this point. The terms ‘block depth’ and ‘settlement window’ are used here. If the block containing the transaction is deep enough in the chain, it is deemed confirmed. Deep enough is a relative term: the depth of a block shows how many further blocks have been added to the chain since that block was introduced. Because blocks have depth, the transactions included within them have depth as well.

The transaction is deemed verified when the depth of a specific block exceeds a certain threshold, and the assets in that transaction can be used ‘safely’ (i.e., the protocol can guarantee the transaction is immutable, so the assets can be traded, exchanged, etc).

The settlement window is the amount of time between when a transaction is confirmed and when the transaction’s assets may be utilized to swap with other assets. A transaction becomes immutable as soon as its depth is greater than 3k/f slots (that is, 129600 slots on current mainnet, or 36 hours).

The chance of immutability

Another factor to examine when deciding whether or not a transaction is verified is its possibility of immutability. The likelihood of a transaction being immutable is proportional to the number of blocks added to the chain since the transaction was approved. The bigger the number of blocks added, the more likely the transaction will become immutable. When a transaction’s depth exceeds 3k/f slots, it becomes immutable. The Ouroboros Praos protocol guarantees this.

Sebastien Guillemot covers transaction finality, amongst other in things, in this informative deep dive on Cardano & Algorand: Leader Selection Explained⁷⁷

P2P (peer-to-peer)

The networking layer of Cardano is a physical infrastructure that unifies nodes and their interactions into a single system. The network disseminates transaction and block generation information to all active nodes. The system validates and adds blocks to the chain in this manner, as well as verifying transactions. As a result, a distributed network of nodes must have minimal communication latency and be robust enough to deal with outages, capacity restrictions, and hackers.

If there is to be complete decentralization in terms of block generation, it’s also critical that there is decentralization of network connectivity as well. Cardano achieves this with peer-to-peer (P2P) networking.

During the Byron era the P2P ran as a federated system, with nodes connected by a static configuration defined in a topology file. Since Shelley, the system then moved to a hybrid mode. During this transition, IOG added a manually constructed peer-to-peer (P2P) network of stake pool operator (SPO) relay nodes. SPO core nodes could now connect to both federated relay nodes and to other SPO-run relay nodes. This connectivity was not automated but still enabled the exchange of block and transaction information without relying on federated nodes. The goal was to switch off all federated relay nodes so that SPOs can connect to each other’s relays.

How it works

Cardano’s network is a cutting edge, self-organizing network. It includes block producing nodes, relay nodes connecting the network and supporting nodes like Daedalus full node wallets, exchange nodes, etc. The network binds all these nodes together. The SPOs run the block producing nodes which isolate themselves from the network using relays, which talk directly with the rest of the network. The relays take care of the overall network connectivity.

A deeper dive into the intricacies around the techniques and mini-protocols involved is in the documentation.⁷⁸ Some of the innovative tools used by IOG include IOsim,⁷⁹ an open-source Haskell package which allows IOG to stress test the network with heavy network traffic to help them run simulations of rare large-scale events.

The network is also parametrically polymorphic,⁸⁰ meaning there is a degree of separation of the network layer from the rest of the cardano-node code. This allows the ledger and consensus layers to be developed independently of the node code, so critical updates to the network and node don’t need to happen at the same time.

All connections are multiplexed⁸¹ which means combining a series of mini protocols into a single Transmission Control Protocol (TCP) connection channel. This has three benefits: bidirectional peer communication (any peer may commence communication with no constraints since both parties have read and write rights inside the same channel), and improved node-to-node communication without compromising performance. It also makes troubleshooting easier as you have less connections.

The P2P governor⁸² keeps track of connections and which peers are active and doing well. It optimizes network speed and reliability on an hourly basis. Peers are divided into three kinds:

• cold peers: peers that are known of, but where there is no established connection
• warm peers: where a connection is established but it is used only for network measurements and not for any application-level consensus protocols
• hot peers: peers where the connection is actively used for the application-level consensus protocols.

IOG’s Technology Manager Kevin Hammond gave a clever analogy during his ScotFest talk:⁸³

By hot connections, what I mean is me, and the people in this room are hot connections. Warm connections are the people watching me on YouTube, you're not active in this room at this point in time, but you could be very shortly, if you come to Edinburgh… Cold connections, those who are not yet connected, but we’re aware of, people who aren't watching the YouTube video live but will watch it later. The cold connections may become (promoted) to warm connections when they watch the YouTube video, and then perhaps (promoted) to hot connections at some point in the future…this is the fundamental concept that we are using with the p2p governor system

P2P roadmap

To get to a self-organizing, fully automated, decentralized P2P network, the network needs to pass through the following four phases. This strategy was not drawn up on the back of a beer mat but based on deep technical research⁸⁴ beyond the scope of this beginners’ book.

Figure 4.3: ‘Hybrid’ Mode, from Kevin Hammond ScotFest talk

Hybrid Mode is the current state of the network at time of writing. In this setup, the black nodes are block producing nodes, which connect via the red circles (relays). The community has produced a manual topology that allows block producing relays to connect with each other. The backbone, the outer ring, is the set of relays that is currently operated by IOG, which the exchanges, Daedalus wallets and light wallets connect with.

Figure 4.4: Dynamic p2p phase, from Kevin Hammond ScotFest talk

Dynamic P2P was rolled out⁸⁵ in 2023. This phase introduces a dynamic self-discovery layer between the SPOs. Rather than having a manual configuration, there is now an automatic self-configuring, reconfiguring layer. SPOs don’t have to go and download the latest list of relays, the relay nodes themselves will automatically reconfigure the network keeping it optimized and secure.

As always, it’s best to check IOG’s weekly dev updates⁸⁶ for the very latest. For those more technical, there are regular updates from engineering teams in the archive.⁸⁷

Figure 4.5: ‘Genesis’ phase, from Kevin Hammond ScotFest talk

The third stage is Ouroboros Genesis, which is in design and development. Genesis will enable secure node bootstrapping by removing the need for the IOG relays. The outer layer will be gone, leaving a completely peer-supported network.

Figure 4.6: ‘Peer sharing’ phase, from Kevin Hammond ScotFest talk

The final stage 4 is ‘peer sharing’ when the orange nodes above will be introduced. These will be well connected nodes, run by the community, which will maintain and establish the network. So, then you will no longer have to go to a trusted sources to download the latest set of known relay peers, individual nodes will automatically reconfigure, and the network will be self-organizing.

Cardano Foundation role in network resilience

The Cardano Foundation also plays an active role in ensuring the network is resilient. They guide SPOs in best practices and provide tooling. For example, at the Dubai Cardano Summit, Markus Gufler held a Masterclass⁸⁸ to demo a network health monitoring tool which gives SPOs a way to understand the scope and limits of pool operations, as well as the global network, by combining telemetry data together.

Cardano Entropy (Randomness)

It shouldn’t be possible to anticipate or influence future blockchain states if the system is genuinely decentralized. All future on-chain events must be fully random. Cardano makes this possible by including an extra entropy mechanism that may be applied to assure the system’s randomness. Any complicated cryptographic system relies on true randomness. A source of pure, unexpected randomness must exist for a cryptographic environment, like the Cardano blockchain, to operate and be accepted by the community as truly decentralized and fair. This assures that the chain’s future path cannot be altered by anybody with knowledge of the past.

The entropy parameter specifies Cardano’s randomization source. Outside of the Cardano ecosystem, this signifies something unexpected, ensuring that no one can ‘hijack’ the blockchain’s randomness. This parameter’s value was computed based on occurrences that could not have been predicted in advance or about which anybody might have had insider knowledge.

Entropy addition mechanism

To appreciate the entropy addition mechanism, one must first comprehend completely decentralized block generation and how the transition nonce⁸⁹ will impact this process.

The Ouroboros protocol uses an evolving sequence of leadership nonces to decide which pools are chosen as block producers (cryptographic seeds used to generate a sequence of values using a repeatable random number generation algorithm). These nonces determine the block production schedule. This timetable is determined by each leader nonce for a whole 5-day epoch, during which the nonce controls the stake pools used to guide the development of each block. To achieve the basic ledger features needed, the leadership nonces and stake distributions develop in lockstep.

Nonce Value

IOG added a transition nonce to the running leadership nonce just after March 31, 2022. When the stake distribution for the April 10 epoch was decided, the transition nonce then had to depend on random values (which are introduced by on-chain transactions) that can’t be anticipated. This emphasizes transactions that emerge on the blockchain between the 12-hour mark — when the stake distribution is settled — and the 42-hour mark, when the hash value is removed.

The transition nonce is a reflection of entropy generated by a multitude of external, unpredictable sources determine the transition nonce, and thus contribute directly to Ouroboros’ perpetual cycle of randomness generation.

A user gives a nonce for a transaction when using the Cardano command line interface (cardano-CLI), however the nonce retrieved when viewing the details of the transaction is different. You can review more details and the relevant command line options in Cardano docs here.⁹⁰

February 2, 2021… How can you explain verifiable randomness to a layman? CH:

That’s easy so imagine your gambling against the server, you know, like for example you’re playing Blackjack. So, when you play Blackjack in human life, you have the dealer and the dealer playing against you and he’s giving you cards, you want to believe that the deck is randomly shuffled but it doesn’t have to be. What if the dealer had pre-set the cards up so he always gets good cards, and you get bad cards? So, what does a dealer do to give you some faith in the process? He shuffles the cards in front of you. If you believe that shuffling is a randomization process then you will be satisfied that you’re playing a fair game, but the problem is what happens if you can’t see them shuffle the cards?

This is the problem when you gamble with central servers, so you’re playing blackjack against blackjackstarz.com or whatever the hell these sites are. Well, how do you know that that’s a fair deck? Well, they have a random number generator, but how do you know it’s not biased so that they win more than you win? Well, you trust that maybe a regulator stepped in and did that. What if you had a cryptographic protocol, you could verify the person’s running and you trust the protocol to produce true randomness, then that protocol would guarantee that they’re shuffling correctly.

That has to be done for a proof-of-stake system. Why? Because we are replacing traditional proof of work, where it’s a meritocratic thing where your computer is chipping away at computations, with a synthetic lottery proportional to your amount of stake you have in the system. So, if you have 25% of the total stake in the system, you should win on average 25% of the time. But what if randomness is biased? Then what can happen is even though you have 25% (of the stake), maybe you only win 8% of the time, because I biased the numbers in a way to reduce your chance of winning. So, you have to have faith that the randomness in the system is secure, it’s true.

Fees on Cardano

Cardano has a transaction fee scheme that covers the cost of transaction processing and long-term storage.

Fees are not paid directly to the block maker in the Cardano ecosystem, which is unusual in the blockchain world. They are instead pooled and given to all pools that produced blocks during an epoch. There are no costs for the memory cost of keeping track of the accumulated chain state, specifically UTXO.

Halting economic attacks

The Shelley hard fork changed the Cardano blockchain from federated to entirely decentralized, thus increasing the incentive for bad actors to carry out economic attacks.

When the running costs of the operators of a system are not compensated by fees imposed on the system’s users, an economic attack could occur. These conditions enable users to impose charges on operators without incurring the whole cost themselves, possibly resulting in a significant decline in operator involvement and, eventually, the system’s collapse.

Cardano’s fee structure is straightforward.

The structure of fees is based on two constants (a and b).

• a and b are protocol parameters, and the method for determining minimum fees for a transaction (tx) is a * size(tx) + b
• size(tx) is the size of the transaction in bytes.

Protocol parameters

Cardano’s update system may adjust protocol settings to respond and adapt to changes in transaction volume, hardware pricing, and ada valuation. Changing these settings is considered a hard fork since it affects which transactions the system accepts.

Parameter a

The transaction cost is dependent on the transaction size, as indicated by parameter a. Larger transactions require more resources to store and complete a transaction.

Parameter b

Regardless of the magnitude of the transaction, the value of b is the fee that must be paid. The purpose of this option is to avoid Distributed Denial-of-Service (DDoS) attacks. b makes such attacks extremely costly and removes the chance of an attacker flooding the system with millions of small transactions.

From Twitter Space, April 18, 2022, ‘Sunday Chat with Charles’ Re: Rewards going down, compensated by Transaction fees.⁹¹

Well, there’s a formal curve in the specification, so if you look at the inflation, it’s over a 140-year period, I think, so it goes down. But it’s bounded and there’s precise formulas for how that works. It’s unlike Bitcoin, which has a step function which reduces every four years by half. This is a continuous emission decline, so it’s a nice gradual monotonically decreasing curve that’s quite smooth.

Now Cardano is unique, unlike Bitcoin, where sidechains are going to really change things. So what people don’t seem to understand is that when a sidechain comes out, the sidechain is going to do its quorum sampling from stake pool operators. And when you mine that sidechain, you have to pay a block reward to get those stake pool operators interested in it, but then that gets paid to the operators and to the delegators, just like ada rewards are.

So as a sidechain ecosystem starts building up, and we have dozens and eventually hundreds of sidechains, that means when you hold ada, in addition to getting ada rewards, you’ll also get tokens from all the sidechains that are supported. So that plus transaction fees is really going to change the calculus. The other thing is that ada is a very volatile asset. So yeah, block rewards are worth something, but you know, ada was $3 less than seven months ago. It could be $5 in seven months. It could be $0.05 in seven months. So, percentages certainly matter, but the real value does as well. And there’s some elasticity there. So, I think the introduction of sidechains increases the transaction volume, and then also the volatility of the price of the underlying asset is going to really be a game changer for the return for holding in that respect, and this is what makes it fun.

You know, the other thing is multi-resource consensus is going to come at some point. It’s one of our proposals for the long-term road map, and there, you’ll have different ways of creating resources for consensus, and those could either be directly monetized with ada or they could be monetized with a different asset. So, there’s a lot there. That said, ada is a deflationary asset. Bitcoin Maxis (maximalists) seem to hate us, but we actually have the same monetary policy in that respect.

Fixed supply 45 billion, we’re gradually earning our way up to it and it’s monotonically decreasing inflation. So it’s a deflationary asset, at its core, and that’s one of the reasons why a lot of people like ada. But there’s going to be a whole ecosystem of tokens, a constellation of them, and lots of utility and things to do. DeFi also changes the equation as well, because you can use your ada and DeFi at the same time, so you can also augment your yields to offset the decline in yields. So, it just basically comes down to what your risk profile is from that respect.

The 3 different sides of full Decentralization

Decentralization is the transfer of power from a central authority to smaller entities. However, in the context of cryptocurrencies and blockchain, this definition merely scratches the surface. Cardano’s technological road to complete decentralization gradually unfolded, with stake pool operators (SPOs) producing varying degrees of blocks, peer-to-peer (P2P) network discovery, and ‘gossip’ with peers trading information among themselves. It entails the implementation of sophisticated community-led governance and decision-making mechanisms, with completely decentralized software and protocol changes as a result.

Pushing power to the edges

The balance of power has moved from the people to businesses like Facebook and Google, resulting in a virtual information monopoly. Centralized authority has data hegemony over their customers due to their unassailable market positions.

Decentralization is the antidote to power concentration and the dangers it entails. Decentralization allows individuals to make choices and decisions, restores personal information ownership to the individual, pushes authority to the margins, and gives every network member (or ada holder) a stake. Cardano’s decentralization is built on three principles: block production, networking, and governance. These are inextricably intertwined and work together to produce a cohesive result: complete decentralization, which lies at their intersection.

1. Decentralized block production

To develop and prosper, every blockchain depends on the creation of new blocks. Core nodes – maintained by IOG, Emurgo, and the Cardano Foundation – were solely responsible for building blocks and maintaining the network throughout the Byron era deployment. Shelley and the Incentivized Testnet, which launched in 2019, served as a testbed for decentralized block production. The Incentivized Testnet experiment demonstrated that Cardano could be supported by a network of community-run stake pools.

The number of stake pools continues to grow and grow, with a good number of them forming blocks and rewarding delegates. Exchanges control some, while single-pool community operators handle others. Everyone contributes to the network’s worth. The former because of their capacity to attract new ada holders into the ecosystem, and the latter because of their contribution to decentralization and grassroots involvement. IOG is dedicated to promoting decentralization, and while they tweaked parameters in the past, the reigns will be handed over to the community gradually with the Age of Voltaire.

2. Decentralized Network

The introduction of peer-to-peer (P2P) networking is the second pillar of Cardano’s decentralization. The goal is to connect geographically dispersed pools to create a safe and reliable blockchain platform.

This feature will leverage a collection of mini protocols⁹² and a categorization of cold, warm, and hot peers on mainnet to allow a node to make the best connection choice possible. In terms of networking, there was a hybrid era where SPOs had to use manual methods to maintain network connections. As SPOs took over block production at d=0, all core nodes were decommissioned. IOG continued to maintain relays, but the SPO network are gradually taking over this duty. To learn more about this, watch March 2021 Cardano360 episode,⁹³ in which IOG’s principal architect Duncan Coutts outlined the P2P future.

3. Decentralized Governance

Cardano has already got transaction metadata and native tokens thanks to the Goguen rollout. Since the Shelley launch, this has perhaps been the most visible evidence of Cardano’s development and success.

At the same time, something even more powerful has emerged with Project Catalyst: an active community of builders, innovators, and entrepreneurs. The Catalyst community has a worldwide membership of entrepreneurs, professionals, and specialists from a variety of fields that make up this pool of decentralized talent, which offers a broad reservoir of innovation to guarantee the greatest and brightest ideas receive the financing they deserve.

Cardano’s ultimate goal is to create a blockchain where a community of stakeholders make practical choices regarding the chain’s protocol and growth, which is supported by a layer of robust governance. Catalyst was the forerunner of Voltaire, the development era that will usher in the third and ultimate degree of decentralization by integrating governance and on-chain decision-making/voting. Chapter 9 is all about the Age of Voltaire and how the governance mechanisms will be rolled out.

September 8, 2021. Re: Catalyst. CH:⁹⁴

The cryptocurrency space, as a whole, is not aware of this yet, they're still thinking that it's just all centralized and top down, and there's a few people floating around who make all these calls. They don't really understand that we've silently built one of the world's largest decentralized organizations and decentralized decision-making machines, and it's in its infancy.

We have constructed this meta brain, that runs this meta machine in the sky, and all these people whether in the south of France, or an island, or right here in Colorado with a lobster on the mic... we somehow just come together, with different values, different languages, different perceptions of reality and we converge to a point where we actually can vote on, and move forward, and make change and get funding behind people... more and more of the community is going to be involved in deciding the direction and governance of Cardano in general.

What's so cool about this is it's completely bottom up. There's not a lot of top down here, the top down stuff is just ensuring that there's best practices and good infrastructure ...and asking questions about what tools we need to construct and how do we improve things? ...but the solutions are coming bottom up, and it's growing rapidly, and more and more people come in every time we do a fund, it just massively increases [..]...we don't even have the mobile interfaces in yet, or the Daedalus integration in yet.

Half a million, million people are going to be in this club next year, if not more. So this is a great opportunity for us to kind of explain where this is going, and why this is the heart of Cardano and what's going to make Cardano frankly not only here to stay, but the dominant protocol of all the protocols.

Edinburgh Decentralization Index

‘Decentralized’ has become a meaningless buzzword for many crypto projects. How does one judge claims with so much misinformation, willful ignorance, FUD (fear, uncertainty and doubt) and good old-fashioned greed and corruption prevalent? The politest description of many claims is that they are subjective and warrant scrutiny. Blockchain and crypto are complex and nuanced for most casual investors, who have little time in their busy lives to do any deep research into projects.

With a regulatory vacuum and no requirement for transparency, it’s not hard to make a digital asset look superficially attractive. With no agreed definition of what decentralization is, who is to say something is not decentralized? Who decides on the criteria of what we are all supposed to be measuring? Who has any remit in a permissionless and open industry? With many blockchains categorized based on high-level criteria, they are often very different once you look closer.

Consider just some of the findings from a well-researched article⁹⁵ by the Cardanians (cardanians.io) published in October 2022:

• Cardano has one entity (Binance) with more than 10% share in the network. Ethereum has numerous such entities with largest having 30%+ share
• Cardano’s minimum attack vector (MAV)⁹⁶ is approx 24, Ethereum and Bitcoinn both have a MAV of 3.
• Anyone can stake their ada on Cardano, with no need to entrust your ada to a third party. You can spend or withdraw your ada anytime. Liquid staking means your funds are never locked. With Ethereum, there is no protocol-level stake delegation. If you have less than 32 ETH, you are obliged to trust a validator with your coins and signing keys.
• It is easy and inexpensive to run a Cardano stake pool. On Ethereum, you are required to lock 32 ETH to run a validator (stake pool equivalent).

While the above study is compelling and the numbers speak for themselves, such a study is complex and nuanced as configurations change all the time. Definitions of the different terms vary based on who you ask. Clearly, a more permanent and objective formal analysis is required. An academic institute may be the best candidate to host such a research agenda. Enter the Edinburgh Decentralization Index (EDI), a project launched by Dr. Daniel Woods and Prof Aggelos Kiayias at ScotFest.⁹⁷ The goal is to propose a structured approach to quantifying decentralization in blockchain systems.

The EDI will be a bellwether for the industry, a north star for retail investors, regulatory bodies, and governments. The research at Edinburgh is a collaborative, interdisciplinary initiative spearheaded by one of IOG’s oldest academic partners, the UoE’s Blockchain Technology Laboratory (BTL).

The EDI will be based on a unified framework for assessing different aspects of decentralization. The researchers will start with Bitcoin and eventually include other chains. The index measures the following criteria:

• Hardware
• Software
• Network
• Consensus mechanism
• Tokenomics
• API
• Governance
• Validators’ geographic distribution

To dig deeper into the EDI, study the research paper⁹⁸ which outlines the framework.

Footnotes

1. A server is a computer program or device that provides a service to another computer program and its user, also known as the client. ↩

2. BitTorrent is a communication protocol for peer-to-peer (P2P) file sharing which is used to distribute data and electronic files over the Internet. BitTorrent is one of the most common protocols for transferring large files, such as digital files containing movies or music. ↩

3. Comparing 4 decentralized data storage offerings, techtarget.com/searchstorage/tip/Comparing-4-decentralized-data-storage-offerings ↩

4. A gossip protocol is a procedure or process of computer peer-to-peer communication that is based on the way epidemics spread. Some distributed systems use peer-to-peer gossip to ensure that data is routed to all members of an ad-hoc network. Some ad-hoc networks have no central registry and the only way to spread common data is to rely on each member to pass it along to their neighbors. ↩

5. 6 types of blockchain consensus mechanisms, essentialcardano.io/article/6-types-of-blockchain-consensus-mechanisms ↩

6. Fernando Sanchez, ‘Why They’re Calling Cardano the Green Blockchain’, iohk.io/en/blog/posts/2021/08/17/why-they-re-calling-cardano-the-green-blockchain/ ↩

7. About stake pools, docs.cardano.org/getting-started/operating-a-stake-pool/about-stake-pools ↩

8. Lark Davis interview, youtu.be/BptZkkNN3tw?t=575 ↩

9. Mempool role in multiple transactions, docs.cardano.org/core-concepts/multiple-transactions ↩

10. Kiayias, Russell (2018) 'Ouroboros-BFT: A Simple Byzantine Fault Tolerant Consensus Protocol', eprint.iacr.org/2018/1049.pdf ↩

11. David, Gazi, Kiayias, Russell (2017) 'Ouroboros Praos: An adaptively-secure, semi-synchronous proof-of-stake blockchain', eprint.iacr.org/2017/573.pdf ↩

12. Badertscher, Gazi, Kiayias, Russell, Zikas (2019), 'Ouroboros Genesis: Composable Proof-of-Stake Blockchains with Dynamic Availability', eprint.iacr.org/2018/378.pdf ↩

13. The framework of universal composability is a general-purpose model for the analysis of cryptographic protocols. It guarantees very strong security properties. Protocols remain secure even if arbitrarily composed with other instances of the same or other protocols. Security is defined in the sense of protocol emulation. Intuitively, a protocol is said to emulate another one, if no environment (observer) can distinguish the executions. Literally, the protocol may simulate the other protocol (without having access to the code). The notion of security is derived by implication. ↩

14. Cardano Summit 2023 - LIVE at the IOG booth, youtube.com/live/0JRtWUR4doM?feature=shared&t=2640 ↩

15. State channels refer to the process in which users transact with one another directly outside of the blockchain, or 'off-chain,' and greatly minimize their use of 'on-chain' operations. ↩

16. Hydra Head GitHub, github.com/orgs/input-output-hk/projects/21 ↩

17. Badertscher, Gazi, Kiayias, Russell, Zikas (2020), ‘Consensus Redux: Distributed Ledgers in the Face of Adversarial Supremacy’, eprint.iacr.org/2020/1021.pdf ↩

18. Kerber, Kohlweiss, Kiayias, Zikas (2019), ‘Ouroboros Crypsinous Privacy-preserving proof-of-stake’, eprint.iacr.org/2018/1132.pdf ↩

19. SNARK stands for ‘Succinct Non-Interactive Argument of Knowledge.’ A (zero knowledge) zk-SNARK is a cryptographic proof that allows one party to prove it possesses certain information without revealing that information. This proof is made possible using a secret key created before the transaction takes place. ↩

20. Badertscher, Gazi, Kiayias, Russell, Zikas (2019),‘Ouroboros Chronos: Permissionless Clock Synchronization via Proof-of-Stake’, eprint.iacr.org/2019/838.pdf ↩

21. The Internet of Things (IoT) is a system of interrelated computing devices, mechanical and digital machines, objects, animals or people that are provided with unique identifiers (UIDs) and the ability to transfer data over a network without requiring human-to-human or human-to-computer interaction. ↩

22. Slot Leader Episode 1, youtu.be/YT0PXYBEnuE?t=2218 ↩

23. Leslie Lamport, everipedia.org/wiki/lang_en/Leslie_Lamport ↩

24. Leslie Lamport, ‘Time, clocks, and the ordering of events in a distributed system’, dl.acm.org/doi/10.1145/359545.359563 ↩

25. Network Time Protocol (NTP) is a networking protocol for clock synchronization between computer systems over packet-switched, variable-latency data networks. In operation since before 1985, NTP is one of the oldest Internet protocols in current use. ↩

26. Network time protocol, ntp.org/ ↩

27. The Bellman–Ford algorithm computes shortest paths from a single source vertex to all of the other vertices in a weighted digraph. It is slower than Dijkstra’s algorithm for the same problem, but more versatile, as it is capable of handling graphs in which some of the edge weights are negative numbers. The algorithm was first proposed by Alfonso Shimbel (1955), but is instead named after Richard Bellman and Lester Ford Jr., who published it in 1958 and 1956, respectively. ↩

28. Solana Downtime Series Continues, Network Faces Serious Issues Again, u.today/solana-downtime-series-continues-network-faces-serious-issues-again ↩

29. Time handling on Cardano, docs.cardano.org/explore-cardano/time ↩

30. Jarek Hirniak on Cardano’s timing issue, twitter.com/ravanave/status/1589241299667800065?s=20&t=tCzT7srtCJ362-aWyi-jYA ↩

31. MinSwap & Axo: sandwich attacks and timing in blockchain, youtube.com/watch?v=mGLm0QdVoNc ↩

32. Surprise AMA! 12/12/2020, youtu.be/GlVU8ZiVUL0?t=3061 ↩

33. Coutts, Panagiotakos, Fitzi, (2022) ‘Ouroboros Leios: design goals and concepts’, iohk.io/en/research/library/papers/ouroboros-leios-design-goals-and-concepts/ ↩

34. Advances in Ouroboros: Scaling for Future Growth, youtube.com/watch?v=xKv94MwSNBw ↩

35. Ouroboros Omega, twitter.com/iohk_charles/status/1357364560504709120 ↩

36. Surprise AMA 02/09/2021, youtu.be/K3l3_SAGXEE?t=1350 ↩

37. Gazi, Kiayias, Russell, Zikas (2020) ‘Ledger Combiners for fast settlement’, eprint.iacr.org/2020/675.pdf ↩

38. Abstraction is used to make models that can be used and reused without having to re-write all the program code for each new application on every different type of computer. Abstraction is usually achieved by writing source code in some particular computer language which can be translated into machine code for different types of computers to execute. Abstraction allows program designers to separate a framework from specific instances which implement details. ↩

39. The mempool (memory pool) is a smaller database of unconfirmed or pending transactions which every node keeps. When a transaction is confirmed by being included in a block, it is removed from the mempool. You can think of a mempool as being like a ‘waiting room’ where a transaction sits before it is added to a block. ↩

40. Stateful services keep track of sessions or transactions and react differently to the same inputs based on that history. Stateless services rely on clients to maintain sessions and center around operations that manipulate resources, rather than the state. ↩

41. Block height represents the number of blocks that were validated and confirmed in the entire history of a particular blockchain network, from the genesis block (or block zero) until the most recent one. Unlike the genesis block, all other blocks contain a reference, or hash, to the block that came immediately before it, and the block height is the number of each block in that sequence. So the block height of the genesis block is #0, and the block height of the first block is #1. ↩

42. cardano-cli (command line interface) query command contains subcommand, one of which is tip: tip: gets the node's current tip (slot number, hash, and block number) ↩

43. Anthony Quinn, ‘Combinator makes easy work of Shelley hard fork’, iohk.io/en/blog/posts/2020/05/07/combinator-makes-easy-work-of-shelley-hard-fork ↩

44. Polymorphic: occurring in several different forms. In computing (feature of a programming language) allowing routines to use variables of different types at different times. ↩

45. Surprise AMA 01/10/2021, youtu.be/iLq6mRk2dyg?t=3926 ↩

46. Solana Sputters Back to Life Following Downtime, Network Restart, decrypt.co/81004/solana-back-online-following-downtime-network-restart ↩

47. Tim Harrison, ‘What the Byron Reboot means for Cardano’, iohk.io/en/blog/posts/2020/03/30/what-the-byron-reboot-means-for-cardano/ ↩

48. Duncan Coutts, iohk.io/en/team/duncan-coutts ↩

49. How we will launch Shelley, youtube.com/watch?v=g7uySEgt06c ↩

50. Cardano Allegra and Mary hard fork changes explained, youtube.com/watch?v=9mjvXjxTks8 ↩

51. Plutus Core is the programming language in which scripts on the Cardano blockchain are written. Plutus Core is a small functional programming language — a formal specification is available. Plutus Core is not read or written by humans; it is a compilation target for other languages. ↩

52. Lambda calculus (λ-calculus) is a formal system in math logic for expressing computation based on function abstraction and application using variable binding and substitution. It is a universal model of computation that can be used to simulate any Turing machine. It was introduced by the mathematician Alonzo Church in the 1930s as part of his research into the foundations of mathematics. ↩

53. In math and computer science, the Entscheidungsproblem (pronounced German for ‘decision problem’) is a challenge posed by David Hilbert and Wilhelm Ackermann in 1928. The problem asks for an algorithm that takes as input a statement of a first-order logic and answers ‘Yes’ or ‘No’ according to whether the statement is universally valid, i.e., valid in every structure satisfying the axioms. By the completeness theorem of first-order logic, a statement is universally valid if and only if it can be deduced from the axioms, so the Entscheidungsproblem can also be viewed as asking for an algorithm to decide whether a given statement is provable from the axioms using the rules of logic. ↩

54. A Turing machine is a mathematical model of computation that defines an abstract machine, which manipulates symbols on a strip of tape according to a table of rules. Despite the model’s simplicity, given any computer algorithm, a Turing machine capable of simulating that algorithm’s logic can be constructed. ↩

55. Development phases and eras on Cardano, docs.cardano.org/explore-cardano/eras-and-phases ↩

56. Surprise AMA 06/05/2020, youtu.be/6pQzQbVgX7c?t=2449 ↩

57. Protocol Parameters, cips.cardano.org/cip/CIP-0009 ↩

58. Current Cardano Parameter Settings, adapools.org/protocol-parameters ↩

59. Special 4th of July Surprise AMA, youtu.be/1qoeLinJ3rg?t=3278 ↩

60. Alexander Russel, iohk.io/en/research/library/authors/alexander-russel/ ↩

61. The Bitcoin halving is when the reward for Bitcoin mining is halved. Halving takes place every 4 years. The halving policy was written into Bitcoin's mining algorithm to counteract inflation by maintaining scarcity ↩

62. Cardano Pledge, Rewards, and Network Security with Kevin, Lars, and Duncan | TCE 90, youtube.com/watch?v=X-ziLksiPOE ↩

63. Cardano Mainnet: Pledge Influence Factor Analysis, reddit.com/r/cardano/comments/gfed1l/cardano_mainnet_pledge_influence_factor_analysis ↩

64. Rare Bloom Day Two Cardano Livestream, youtu.be/eBH34WkU_2w?t=16180 ↩

65. Apr SPO Call - Discord Stage follow-up, input-output.typeform.com/report/Row2tnqQ/HSCUwpltfMyJ7yp7 ↩

66. Oct 25 2022: Cardano MinPool Fee Discussion, youtu.be/Gz8J5vX1Rnk ↩

67. @adamKdean, twitter.com/adamKDean/status/1612118333603860480?t=mM5focPvmY-GKax-lOkAPA ↩

68. Cardano (ADA) & Silver Bullets | Cardano Rumor Rundown #514, youtube.com/watch?v=yWRwncQd4Vs&t=28s ↩

69. Staking parameters and network optimization – where next for ‘k’ and ‘min fee’?, essentialcardano.io/article/staking-parameters-and-network-optimization-where-next-for-k-and-min-fee ↩

70. Surprise AMA 01/07/2023, youtube.com/watch?v=djhKk-3rYhU&t=328s ↩

71. Entering Voltaire: on-chain poll for SPOs, cardanofoundation.org/en/news/entering-voltaire-on-chain-poll-for-spos/ ↩

72. CIP-0094? | SPO On-chain Polls #496, github.com/cardano-foundation/CIPs/pull/496 ↩

73. Backpressure: Cardano is designed to automatically deal with heavy traffic. Ouroboros and the network stack function even when saturated. If the network is saturated, Cardano can use the admission control method to regulate and restore normalcy. This is the term ‘backpressure’ mentioned in blogs and documentation, it is basically a strategy for network load management. ↩

74. An NFT drop is the release of a non-fungible token project. A drop refers to the exact date, time, and generally the minting price of the NFT. Many NFT drops have purchase limits that apply to the number of NFTs you are able to mint in one transaction. ↩

75. Coutts, Davies, Szamotulski, Thompson (2020) 'Introduction to the design of the Data Diffusion and Networking for Cardano Shelley', hydra.iohk.io/build/7249613/download/1/network-design.pdf ↩

76. Cardano Upcoming NFT Drops, nftcalendar.io/b/cardano/ ↩

77. Cardano & Algorand: Leader Selection Explained, youtu.be/3k3ls7pzlKg ↩

78. P2P networking docs, docs.cardano.org/explore-cardano/cardano-network/p2p-networking ↩

79. io-sim, github.com/input-output-hk/io-sim ↩

80. Polymorphism is the provision of a single interface to entities of different types . ↩

81. Multiplexing, docs.cardano.org/explore-cardano/cardano-network/networking-protocol/#multiplexin ↩

82. Ouroboros.Network.PeerSelection.Governor, input-output-hk.github.io/ouroboros-network/ouroboros-network/Ouroboros-Network-PeerSelection-Governor.html ↩

83. Kevin Hammond, P2P Network talk @ 2hr 30min, twitter.com/InputOutputHK/status/1593884394715897858?s=20&t=0mPqLNPBdJ7vFHxTGFW6ew ↩

84. Kolyvas, Voulgaris (2022), 'CougaR: Fast and Eclipse-Resilient Dissemination for Blockchain Networks' iohk.io/en/research/library/papers/cougar-fast-and-eclipse-resilient-dissemination-for-blockchain-networks/ ↩

85. Dynamic P2P is available on mainnet, iohk.io/en/blog/posts/2023/03/16/dynamic-p2p-is-coming-to-cardano/ ↩

86. Cardano weekly dev updates, essentialcardano.io/development-update ↩

87. IOG Engineering updates, engineering.iog.io/archive/ ↩

88. SPO Masterclass: Cardano Network Health Monitoring Q&A, youtube.com/watch?v=hahYBjNrhEs&t=2s ↩

89. A nonce is an arbitrary number that can be used just once in a cryptographic communication. It is similar in spirit to a nonce word, hence the name. It is often a random or pseudo-random number issued in an authentication protocol to ensure that old communications cannot be reused in replay attacks. ↩

90. Explanation of the nonce value, docs.cardano.org/explore-cardano/explain-nonce ↩

91. Twitter Space ‘Sunday Chat with Charles’, twitter.com/IOHK_Charles/status/1515872352395055109?s=20&t=ivZvNmtLVKrKk_ZEg-1e_A ↩

92. Marcin Szamotulski, ‘Cardano’s Path to Decentralization’, iohk.io/en/blog/posts/2020/07/09/cardanos-path-to-decentralization-by-marcin-szamotulski/ ↩

93. Decentralization unpacked with Colin Edwards, Duncan Coutts, Lars Brunjes & Shawn McMurdo , youtu.be/mXYIQDUitYI ↩

94. Charles Hoskinson Explains Cardano's Secret Weapon (Project Catalyst), youtube.com/watch?v=vQOvX-HAQDQ ↩

95. Comparing the decentralization of Cardano and Ethereum, cexplorer.io/article/comparing-the-decentralization-of-cardano-and-ethereum ↩

96. The minimum attack vector (MAV) is the minimum number of participants required to hijack control in order to attack, or manipulate, the network. ↩

97. The first of its kind, introducing the Edinburgh Decentralization Index, youtube.com/watch?v=8GJQDsH854Q ↩

98. Karakostas, Kiayias, Ovezik (2022), 'SoK: A Stratified Approach to Blockchain Decentralization', arxiv.org/abs/2211.01291 ↩