Reviewed and updated: October 19, 2026.
Cardano is easier to understand by following a transaction than by imagining two independent blockchains called settlement and computation. Its ledger model, validation rules, block-production protocol and scaling systems work together. The older CSL/CCL diagram was too literal to explain those relationships reliably.
eUTXO makes the inputs explicit
Cardano's extended UTXO explanation describes transactions consuming existing unspent outputs and creating new ones. Script-locked outputs can carry data, and the spending transaction supplies information used by the validator. The script evaluates whether the proposed spend satisfies its conditions.
In a fictional ADA-only transfer, Alice spends an output containing 12 ADA, sends 5 ADA to Ben and creates 6.8 ADA of change, assuming a 0.2 ADA fee. The original output is consumed; the two new outputs total 11.8 ADA. These invented figures illustrate accounting, not a current fee quote or a complete transaction specification.
Predictability does not remove input contention
A transaction can be checked against known inputs before submission, but another transaction may consume an expected input first. Consider a ticket contract with one state output holding the remaining inventory. Two buyers construct transactions against that same output. Both cannot consume it independently. The application needs a design for ordering, batching or distributing state.
Independent outputs can support parallel work. A single shared state output can become a coordination point. This is an application-design issue, not evidence that eUTXO prevents all concurrency or automatically provides unlimited concurrency.
Ouroboros is a family of consensus protocols
The official overview describes stake-based block production and several Ouroboros variants. A research paper, an implementation and an activated network upgrade are distinct milestones. This guide does not label a historical variant or a proposed throughput feature as the live version without deployment evidence.
Formal analysis proves specified properties under specified assumptions. It does not certify every wallet, contract, bridge, operator or economic incentive built around the protocol. When someone cites a security proof, ask which component it covers and what assumptions the application adds.
Hydra moves particular activity into a shared off-chain ledger
Cardano's Hydra documentation describes a Hydra Head as an off-chain mini-ledger shared by a group of participants through a state-channel design. It is not a blanket multiplication of base-chain throughput for every user and application.
For a hypothetical game session, participants might conduct many updates inside one head and later settle according to its protocol. That use case still needs rules for participation, committed assets, availability, closing and disputes. A benchmark for that session should state those conditions rather than present its update rate as global Layer 1 throughput.
Block limits are parameters, not automatic promises
The protocol-parameter guide identifies settings such as maximum block-body size and transaction limits and explains that adjustable parameters change through governance processes. The earlier claim that block size simply adapts to load, coupled with stale 72 KB and 80 KB figures, has been removed.
When comparing performance, record the actual network and protocol version, parameter snapshot, transaction complexity and confirmation criterion. Increasing a limit can affect propagation, storage and validation costs; it is not a free scaling gain.
A useful architecture review follows four questions: which outputs are consumed, which scripts authorize the spend, how the result enters the accepted chain, and which additional assumptions a scaling layer introduces.
Sources and scope
- Cardano Docs: extended UTXO
- Cardano Docs: Ouroboros overview
- Cardano Docs: Hydra
- Cardano Docs: protocol parameters
Continue reading: Bitcoin UTXOs · Layer 1 and Layer 2 · Proof of stake
Editorial update: Replaced the misleading independent-layer model and stale block-size claims; qualified formal-security and roadmap statements; added eUTXO accounting and contention examples.