The Lightning Network
Move payments off the chain, settle on it only when necessary. It works, it is genuinely clever, and it has real limitations nobody should hide.
If every coffee purchase must be recorded on a ledger every full node stores forever, coffee purchases will be expensive. Lightning's answer is that most payments need not be recorded at all — only the opening and closing balances between two parties.
Payment channels
- Two parties lock funds into a shared on-chain address. That transaction is recorded normally and costs a normal fee.
- They then exchange signed statements updating how the locked amount is divided. These are valid transactions, held privately, not broadcast.
- They can update thousands of times, instantly and at no cost, because nothing touches the chain.
- When they close, the final state is broadcast and settled. Two on-chain transactions have covered any number of payments.
Broadcasting an old state — claiming a balance from earlier, when you held more — allows the counterparty to prove it during a dispute window and take the entire channel balance. Cheating is possible and losing everything is the price, which is why it is not attempted.
Routing
You need not have a channel with everyone you pay. If you have a channel with A and A has one with B, a payment can be routed through A. Cryptography ensures intermediaries cannot steal in transit: each hop is released only when the next produces proof of delivery, so either the whole path completes or none of it does.
The real limitations
- Capital must be locked up. Funds committed to a channel cannot be used elsewhere while it is open, so routing capacity has a genuine cost.
- Liquidity is directional. A channel can hold plenty on one side and nothing on the other, so a payment may fail for lack of a path even when both parties hold funds.
- Receiving requires inbound capacity. A new user can pay immediately but cannot receive until someone has committed funds toward them — an awkward first-use problem.
- Nodes must be reachable. Watching for fraudulent closes requires being online, or delegating that job to a watchtower service.
- Hub formation. Routing works better through well-connected, well-funded nodes, which pushes toward concentration — the tension the base layer was designed to avoid.
Lightning demonstrably moves payments instantly for negligible fees, and has been adopted in specific niches — remittances, tipping, some point-of-sale, and national-scale pilots. It also remains harder to operate than a payment app, and the liquidity management problem is intrinsic rather than a matter of polish. Both statements are true and advocates and critics tend to quote only one.
Why the design is worth understanding regardless
The general pattern — keep the slow, expensive, maximally-secure layer for settlement and build fast layers on top — is the dominant scaling approach across the whole field, including Ethereum's rollups. Lightning is the earliest large-scale example of it.