What a Blockchain Actually Is

The word "blockchain" has been stretched to mean money, ideology, fraud, and the future all at once, which is another way of saying it has been stripped of meaning. Underneath the noise is a specific engineering result: a way for parties who do not trust each other, and who answer to no shared authority, to agree on one ordered history of events and keep agreeing as new events arrive. That is the entire problem the technology exists to solve. Everything else is a consequence of how it solves it.

The problem: agreement without an authority

Any shared record of ownership needs a way to resolve conflicts. The conventional answer is an authority: a bank, a registrar, a company that runs the database and whose word is final. That works precisely to the degree you trust the authority not to be compromised, censored, coerced, or simply switched off.

Remove that authority and a hard question appears. How do a set of independent participants, none of whom trust the others, converge on a single shared history, and stay converged as new entries are added? The difficulty is not storing the data. It is getting everyone to agree on the same version of it, in the same order, with no one in charge of deciding.

The sharpest instance is the double-spend problem. A digital token is data, and data can be copied. Nothing intrinsic stops you from transferring the same unit to two counterparties at once and letting each believe they received it. Physical cash avoids this by being scarce in your hand; digital value has no such property. For decades the only known fix was a central operator holding the authoritative ledger. A blockchain was the first widely adopted system to resolve it without one.

Blocks, hashes, and the chain

A blockchain records transactions in batches called blocks. Each block is an ordered set of transactions plus metadata. The structural question is what binds these blocks into one tamper-evident sequence, and the answer is cryptographic hashing.

A hash function takes an input of any size and returns a fixed-length output with two properties that matter here: the same input always yields the same output, and any change to the input, however small, yields a completely different and unpredictable output. The output is effectively a fingerprint of the data. Anyone can recompute it and check that it matches; no one can feasibly work backward from the fingerprint to forge matching data.

Each block stores the hash of the block before it. That single design choice is what the name refers to. Because block N commits to the exact contents of block N−1, altering any past transaction changes that block's hash, which invalidates the reference stored in the next block, which invalidates the next, and so on to the present. Rewriting one entry means rewriting every block after it. The chain of hashes converts an unordered pile of records into a single ordered history where tampering is not merely detectable but structurally self-evident.

Consensus: deciding who writes next

Hash-chaining makes tampering visible. It does not decide who is allowed to append the next block, or what happens when two participants propose competing next blocks simultaneously. That is the job of a consensus mechanism: a rule by which thousands of independent machines, called nodes, agree on one canonical chain without coordination from a center.

The governing idea is to make appending a block costly in a way that is difficult to fake but trivial to verify, so that the honest majority's version is always the cheapest to extend and the most expensive to overturn. Two designs dominate.

Proof of Work

Under Proof of Work, used by Bitcoin, nodes called miners compete to find an input that, hashed together with the block, produces an output below a target threshold. There is no shortcut; the only method is to try enormous numbers of candidates. The first to succeed appends the block and collects a reward. Because each attempt costs real electricity and hardware, rewriting historical blocks would require out-computing the entire honest network, which is prohibitively expensive. Security is purchased directly with energy, which is simultaneously the model's strength and its most cited liability.

Proof of Stake

Under Proof of Stake, used by Ethereum since 2022 and by most newer chains, the right to append is not bought with electricity but with capital at risk. Participants called validators lock up a quantity of the network's own asset as a bond. The protocol selects validators to propose and attest to blocks roughly in proportion to the amount staked. A validator that signs fraudulent or conflicting blocks has part of its stake destroyed, a penalty called slashing. The incentive shifts from "attacking costs more than it earns" to "attacking forfeits your own capital." Proof of Stake consumes a fraction of the energy, at the cost of greater protocol complexity and a persistent debate over whether staking concentrates influence among those who already hold the most.

A third design deserves mention here, because it is routinely misdescribed. Proof of History, associated with Solana, is not a standalone consensus mechanism. It is closer to a second layer sitting on top of Solana's underlying Proof of Stake: a cryptographic clock that stamps transactions into a verifiable order before consensus runs, so validators can agree on sequence without the constant back-and-forth messaging other chains depend on. That ordering is what lets the network execute so fast, and the tradeoff shows up as a markedly higher rate of dropped and failed transactions, especially under load.

The two networks also treat staking itself very differently. Ethereum sets a fixed threshold: to participate in validation a validator must post exactly 32 ETH, no more and no less. Solana imposes no such minimum. It instead assigns leader slots (the turns to produce blocks) in proportion to stake, so the more you stake the more often you lead, and a validator only needs enough stake to cover its role while voting or leading. One design gates participation behind a fixed bond; the other weights it continuously by size. Both get their own treatment further up the ladder.

"Decentralized" and "immutable," precisely

Decentralized means no single party controls the ledger or can unilaterally rewrite it, because many independent nodes hold copies and enforce the same rules. There is no single server to seize or shut down. But decentralization is a measured quantity, not a binary. When a small number of mining pools, staking providers, or cloud hosts operate most of the network, the system is decentralized in design and concentrated in fact. The correct question is always how decentralized, verified against who actually runs the infrastructure.

Immutable means that once a transaction is buried under enough subsequent blocks, reversing it is computationally or economically infeasible. It is protected by cost, not by physical law, and the guarantee runs in both directions. The same property that prevents a government from censoring your transaction also means a transaction sent to the wrong address, or extracted through a contract bug, has no support desk and no reversal. Immutability is powerful precisely because it is unforgiving.

What blockchains are good and bad at

A firm that investigates these systems earns trust by stating the tradeoffs plainly rather than selling the technology as universal. Here is the grounded version.

Blockchains are exceptional at a narrow set of things. They settle questions of ownership and ordering without a trusted intermediary, which is the core competency and the reason they exist. They provide censorship resistance, because no single entity can block or reverse a valid transaction. They provide auditability, because the history is public and independently verifiable by anyone. And they support programmable agreements (smart contracts) that execute exactly as written with no human in the loop.

They are bad at, or simply the wrong tool for, a longer list. They are typically slow and expensive relative to a centralized database, because global agreement on every entry is inherently costly; a well-run central system will almost always be faster and cheaper. They offer little privacy, since most public chains expose every transaction permanently to anyone who looks. They cannot fix mistakes, because the immutability that protects you also strands you: no undo, no chargeback, no override. They are poor storage, being expensive ledgers rather than hard drives. And they do not eliminate trust; they relocate it. Remove trust in a central operator and you must now trust the code, the consensus design, the key custody, and the operators running the nodes. Trust does not disappear. It moves, and it tends to move to places that are harder to inspect.

Where a trustworthy, accountable authority already exists and everyone is content to rely on it, a blockchain is usually the wrong answer. The technology earns its cost only where that authority is missing, unreliable, or dangerous to depend on.

Where to go next

You now have the spine of the field: a blockchain is a replicated, append-only ledger, bound into one order by hash-chaining, kept honest by a consensus mechanism, and decentralized and immutable within real and measurable limits. Wallets, tokens, smart contracts, MEV, and the failures we hunt all build on this.

From here the ladder continues:

• Hashes, Keys, and Signatures:the cryptography that proves identity without revealing the secret behind it.
• Wallets and Private Keys:what you actually control, and how people lose it.
• Smart Contracts, Explained:code that custodies money, and why that is both the point and the danger.
• Proof of History and Modern Consensus:the mechanism behind Solana's throughput, in full.