Lean blocks
Sequentia makes blocks ten times as often as Bitcoin and a tenth of the size. A full node costs the same to run as a Bitcoin node, and nobody can make the chain run faster.
Why assets need more on-chain room than money
Bitcoin can keep its blocks slow because money is the most liquid thing there is. Most bitcoin payments can move to second layers such as Lightning, with the blockchain used to open, rebalance and settle.
Other assets are different. A share in a small company or a regional bond will never have enough liquidity for a payment-channel network. For those, an on-chain transfer remains the normal case. A chain built for assets has to offer on-chain confirmation faster than ten minutes, and has to care about fitting more transfers into less space.
The three numbers
All three are consensus rules. A block that breaks one is invalid for every node.
| Rule | Value | Compared with Bitcoin |
|---|---|---|
| Minimum time between blocks | 60 seconds | 10 minutes on average |
| Maximum block weight | 400,000 units | 4,000,000 |
| Slot unit | 10 seconds | (no equivalent) |
Same disk, same bandwidth
400,000 weight units every 60 seconds is exactly 4,000,000 every 600 seconds. A Sequentia chain running completely full grows on disk at the same rate as a Bitcoin chain running completely full.
The limit counts everything in the block, including the producer’s proof and the committee’s certificate, so what is held equal to Bitcoin is total disk growth and not just user payments.
A pace that is a rule, not a habit
The first edition of the whitepaper suggested a target time per block and left it to the software. That turned out not to be enough.
Producers are paid in fees. On a busy chain full of paying transactions, they would have every reason to produce blocks faster. But a faster chain means more disk, more bandwidth and more validation for every node, forever. So the pace was made a consensus rule: a block’s timestamp must be at least sixty seconds after its parent’s.
The chain can run slower, when a producer is absent or the network is escaping a stall. It cannot run faster.
Whitepaper 10.4Can a producer cheat by lying about the time?
The rules compare timestamps written in blocks, since a validator cannot know when a block was really made. As in Bitcoin, no block may be stamped more than two hours ahead of the validating node’s clock.
A producer trying to run faster could only do it by stamping blocks ahead of real time. After a one-off burst of at most 120 blocks, two hours’ worth, the chain could produce nothing until real time caught up.
The two-hour check uses the network-adjusted time inherited from Bitcoin Core: the node’s own clock, corrected by the median of what its peers report, with no central time server. The check is deliberately soft: a block too far ahead is set aside, not marked invalid, and accepted once the clock catches up.
Local clocks otherwise matter only for production. They govern how quickly the chain moves, never which blocks are valid.
Who goes next when the winner is absent
The lottery gives every staker a number. It also sorts them into ranks: rank 0 for the best draws, rank 1 for the next band, and so on.
A staker in rank r may not propose until r × 10 seconds after the parent block. So if the best-ranked producer is absent, the next may step in ten seconds later, and the one after that ten seconds later again.
This fixes an orderly queue of fallbacks without slowing the chain: since blocks cannot come sooner than sixty seconds anyway, ranks 0 to 6 all fit before the floor.
Whitepaper 10.3Why ten seconds and not sixty
The rank of a round’s winner is a matter of chance. In about 63% of rounds the best draw falls in rank 0, in about 23% in rank 1, in about 9% in rank 2, and in the remaining few per cent further down.
If the slot unit were sixty seconds, a winner in rank 2 could not propose before 120 seconds. About 14% of rounds are won in rank 2 or below, so the average block would arrive after about 73 seconds, even with every staker online.
With a ten-second unit, only a winner in rank 7 or below is delayed at all: about one round in a thousand. As a share of the chain’s capacity lost to the gate, that is about 18% with a sixty-second unit, about 4% with thirty seconds, and about 0.02% with ten.
Ten was chosen over smaller values because the committee moves from one candidate to the next only once per round, and a round is about nine and a half seconds at a full committee. A fallback producer that became eligible sooner would still have to wait for the round to turn. A smaller unit would also make more candidates eligible at once, each of whose proposals every node must receive, forward and verify.
Without the sixty-second floor, the average wait imposed by the gate alone would be under six seconds, and nothing in consensus would stop producers running the chain at that pace. It is the floor that limits the speed.
What it adds up to
- A payment confirms in about a minute and is final ten seconds after its block is proposed.
- A full node is affordable. It grows no faster than a Bitcoin node, and checking a full committee’s signatures takes about two seconds even on a machine limited to a quarter of one processor core.
- An empty certified block is under 700 bytes. The committee’s proof is about 300 bytes at any size.
- Efficiency is a priority. Techniques that fit more into a block, such as batching payments in several assets into one transaction, matter more here than on a chain where most activity moves off-chain.
Read further
- Final in seconds
- Parameters at a glance
- Whitepaper, Principle 10