The longer read

Concepts and glossary

Every idea here arrives in the same order: first something you already know from daily life, then how the machinery works, then the small print about where it can go wrong.

L0 · Analogy

The tab and the notary

When you run a tab at a neighbourhood bar, nobody calls a notary after each drink. You trust the arrangement because it is cheap to check, because the till is right there, and because walking out has consequences. At closing time you settle, and if there is ever a quarrel, the notary and the law are still there as the last word.

The notary is slow, formal and expensive on purpose. That is exactly why you only visit when you have to. The tab is quick and informal, and it is safe only because the notary stands behind it.

Bitcoin is built the same way. The base chain is the notary: a public ledger that thousands of independent computers check, extended by proof of work, in blocks that arrive about every ten minutes on average. It is not designed to be a fast till. It is designed to be the place where the last word is hard to change.

Why scale in layers at all?

There are two ways to carry more traffic over a road. You can widen the road, or you can send fewer cars down it. Widening the base chain means every checker needs more bandwidth, storage and time, and that nudges ordinary people out of the checking business. When fewer people check, the system leans harder on trust, which is the thing it was invented to remove.

Layering is the other approach. Keep the road narrow and dependable, and let most journeys happen somewhere else, with the road acting as the place where journeys begin, end, and get argued about.

Three questions help you compare any layered design:

What do I lock up? Moving to a higher layer usually means committing funds first.
Who can stop me leaving? The strongest designs let you walk back to the base chain on your own.
Who do I have to believe? Every shortcut replaces some checking with some belief. Find out whose.

This is also why the site's depth gauge exists. It does not measure chain layers. It measures how deep the explanation has gone, so you can stop at the analogy or keep digging.

L1 · Mechanics

Payment channels

Two friends keep one notebook. A line only counts if both sign it. Each new page replaces the last, and if either friend pulls out an old page, the other can prove it is old.

To open a channel, two people create a transaction that sends funds into an output requiring both of their signatures, a 2-of-2 multisignature. Before that transaction goes anywhere, they also sign a refund path, so neither can be trapped by the other disappearing.

Each payment is a fresh pair of signed, unpublished transactions that re-split the locked funds. They are called commitment transactions. At any moment, each person holds a signed copy of the current split that they could publish on their own. Nothing is broadcast unless something goes wrong or it is time to leave.

Old states must become dangerous to publish. The usual trick is a revocation: when a new state is agreed, each side hands over the secret that would let the other punish an attempt to publish the old one. A stale state is then a bet the cheater loses. Honest parties, or watchtowers acting for them, need to be able to notice and respond in time.

Closing is easiest when both people cooperate: sign one final split, publish a single transaction, done. The channel therefore costs two on-chain transactions in the ordinary case, however many payments it carried. The simulator on this site shows exactly that, with the escape hatches left out for clarity.

Payments across more than one hop work by chaining channels with conditional payments, commonly built from hash time-locked contracts. Each hop releases funds only if a secret is revealed before a deadline, so either the whole path settles or none of it does. That leads directly to the main practical cost of channels: capacity must exist along the whole path, in the right direction.

Try a channel in the simulator

Sidechains

A second railway line with its own timetable and its own station staff, joined to the main station by a transfer desk. The trains may be quicker. The transfer desk is where you must trust somebody.

A sidechain is an independent chain that has its own rules and its own way of producing blocks, connected to the base chain by a bridge. To use it, you lock value on the base chain, and an equivalent amount becomes available on the sidechain. To leave, the process runs in reverse.

Because the sidechain does not inherit the base chain's checking, its safety depends on its own participants. The bridge is commonly controlled by a group of signers, sometimes called a federation, or by some other rule set. How much of that group you must trust is the central question for any sidechain.

What you gain is freedom. A sidechain can pick different block times, different features, even different privacy properties, without asking the base chain to change. What you accept is that the guarantees you rely on are now those of the sidechain and its bridge, not of the base chain alone.

Rollups

A courier collects a hundred parcels, and lodges one sealed receipt at the registry. The registry never opens the parcels. It keeps the receipt, which is enough to settle any argument about what was delivered.

A rollup executes transactions away from the base chain, then posts a compact record to it: the data needed to rebuild the result, a commitment to the new state, or both. The base chain becomes the referee and the archive rather than the executor.

Two broad families exist. In optimistic designs, a posted batch is assumed correct unless someone publishes a fraud proof during a challenge window. In validity-proof designs, a cryptographic proof that the batch followed the rules is posted with it. Either way, the important questions are who orders transactions (a sequencer), whether the data is truly available to everyone, and whether you can always exit.

On networks with expressive smart contracts, rollups are a mature idea. On Bitcoin, the base layer is deliberately minimal, and how to verify rollup-style claims without adding new assumptions is a subject of ongoing discussion and proposals. This site treats rollups as a general concept and flags that Bitcoin-specific designs are the most uncertain part of the picture.

L2 · Fine print

What can go wrong

A layer two does not remove risk. It moves it somewhere else, and sometimes makes it easier to overlook. A few categories to keep in mind, with a question you can put to any design.

Liveness

Some protections only work if someone is watching at the right time. Ask: what happens if I am offline for a while?

Exit

You are only as free as your worst-case way out. Ask: can I get back to the base chain without anyone's permission, and how long might it take?

Custody and bridges

If a small group can move the pooled funds, then you are trusting that group. Ask: who holds the keys, and what stops them?

Data availability

If the data needed to verify the state is withheld, nobody can check anything. Ask: where is the data, and who can see it?

Complexity

More moving parts means more places for a mistake to hide. Ask: how many things must work correctly for my funds to be safe?

Changing rules

Layers can be upgraded, sometimes by a small group. Ask: who can change the rules, and with how much notice?

This page explains ideas. It is not a recommendation to use any particular product, and it cannot tell you whether a specific system is safe. Read the documentation of whatever you use, start small, and test your exit before you need it.

Check yourself

Three quick questions. Your answers stay on this page, and a wrong answer simply shows you the reasoning.

Glossary

Plain definitions, in alphabetical order. Type to filter by word or by meaning.

Search the glossary

Showing all terms.

Base layer core: The main chain that everything else ultimately relies on for final settlement. Also called layer one.
Block core: A batch of transactions added to the chain. Bitcoin targets a new block about every ten minutes on average.
Bridge sidechain: The mechanism that moves value between a base chain and another chain. It is often where most of the trust and most of the risk are concentrated.
Capacity channel: The total funds locked in a channel. Payments change how that total is split, never the total itself.
Commitment transaction channel: A signed but unpublished transaction recording the current split of a channel. Either side can publish their copy to leave.
Cooperative close channel: Both people sign a final split and publish one settlement transaction. Usually the quickest and cheapest way to leave.
Data availability rollup: Whether the data needed to verify or rebuild a layer's state has actually been published and can be obtained by anyone who wants to check.
Federation sidechain: A group of signers who jointly control a bridge or a sidechain. Users trust that enough of them behave honestly.
Finality core: The point after which a transaction is, for practical purposes, not going to be reversed. Different layers reach it in different ways.
Fraud proof rollup: In optimistic designs, evidence that a posted batch was computed incorrectly, submitted during a challenge window.
HTLC channel: A hash time-locked contract: a conditional payment released if a secret is revealed before a deadline, otherwise returned. It lets payments hop safely across several channels.
Inbound and outbound liquidity channel: How much you can receive or send through a channel, given how its funds are currently split between the two sides.
Multisignature core: A spending rule that needs several signatures. A channel's funding output typically needs both people, known as 2-of-2.
Proof of work core: The way Bitcoin orders blocks: producers show they spent real computation, using SHA-256 hashing, which makes rewriting history costly.
Revocation channel: The mechanism that makes an old channel state dangerous to publish, because the other side can punish the attempt.
Rollup rollup: A layer that runs transactions elsewhere and posts compact data, commitments or proofs to the base chain so it can referee the results.
Sequencer rollup: The party or system that orders transactions on a layer two before they are posted. Who it is affects censorship and ordering risks.
Settlement core: Writing a final outcome to the base chain, where it can be relied on without trusting the layer that produced it.
Sidechain sidechain: An independent chain with its own rules and consensus, connected to a base chain through a bridge.
Timelock core: A condition that stops an output being spent until a stated time or height. Channels use them to create safe waiting periods.
Unilateral close channel: One person publishes their latest signed state without the other's help, usually followed by a waiting period before they can spend.
UTXO core: Unspent transaction output. Bitcoin tracks coins as outputs that transactions consume and create, rather than as account balances.
Validity proof rollup: A cryptographic proof that a batch of changes followed the rules, posted so that the base layer need not re-run the batch.
Watchtower channel: A helper that watches the base chain for outdated channel states on your behalf, so you do not have to stay online.