————————————————————
Now everyone is bragging about how L2 block times are getting shorter, I think it's time to talk about what’s really going on.
Specifically, let’s clarify what L2 blocks are? How do they differ from L1 blocks? And why we actually don’t need to care too much about L2 block times—although it is indeed an interesting engineering metric.
🔻What is a block?
Back when the term "blockchain" had some meaning (around 2009), the concept of blocks was introduced because we needed a unit to package a bunch of transactions and then hand it over to the consensus mechanism for processing.
For example, on Bitcoin, every miner (block producer) tries to order some transactions to meet the proof-of-work (PoW) requirements and then broadcasts this block to the network. Other nodes will check whether this block really meets the PoW standards.
On Ethereum, the current model uses proof-of-stake (PoS) and an account model. After executing each block, the block producer computes a hash value (state commitment) of the blockchain state, and validators recompute this value to facilitate verifying whether the block is reliable.
The overall process has always been like this:
1️⃣ Select some transactions, order them, and that’s called a block.
2️⃣ Calculate some specific things (like PoW puzzles or state commitments) and add them to the block.
3️⃣ Broadcast this block and additional data to other nodes on the network.
4️⃣ Other nodes recalculate this data and compare it with the data from the block producer. If they match, the block is accepted.
On L1, blocks are important; you need to check the chain's integrity through blocks, handle forks by blocks, and so on.
💡To summarize: a block is a core basic unit in the consensus mechanism.
🔻What is an L2 block?
L2 exists because the consensus process is too slow, and decentralization needs to support those slower computers and networks. L2’s approach is to offload transaction processing to the fastest machines and then send a verifiable execution summary to L1 for consensus.
To put it simply, most L2s are actually centralized systems pretending to be blockchains— but that's not a problem.
At this point, block time becomes a bit blurred. L2 continues to produce blocks mainly to be compatible with L1 software, but this is basically artificially created. When sending summaries to L1, L2 usually packages several blocks into one batch to reduce costs.
Although state commitments are occasionally needed for fraud proofs or validity proofs, not every block requires them. Therefore, L2 blocks are basically not very useful.
Some say their L2 blocks are fast, but they just adjusted a system configuration to shorten the block time. Although they still have to process a certain number of transactions in each block, that’s about it.
🔻So what is important?
As a user, the only thing you care about is: the time it takes for transactions to go back and forth. In other words, how long does it take for my transaction to reach the L2 sequencer, be executed, and then the result to appear on the RPC node I'm using? Let’s focus on the last step: how long does it take to transmit the transaction execution result to the RPC node?
Slower blockchains usually wait until a block is finished before sending it to other nodes. Solana pioneered the idea of 'streaming' blocks: as soon as transactions are processed, they are immediately sent to other validators.
Solana splits these into 'entries' (up to 64 transactions per batch) and then into 'shards' for transmission across the network. We have a deep article specifically discussing this; you can check it out if you’re interested. This data continuously flows from the leader node to other nodes, meaning you can know the execution status of transactions before the block ends.
Now L2 is also starting to adopt this mechanism; for example, Base has introduced Flashblocks, breaking the original 2-second block time into 200-millisecond mini-blocks.
MegaETH has the concept of 'mini-blocks', producing one every 15 milliseconds on the test network (most of the time). Eclipse directly adopted Solana's entry/shard system. This way, users don’t have to wait too long, transactions can be executed, and the experience is quite good!
But to be frank, the real highlight here is "shortening the interval of network communication." It has nothing to do with which blockchain is inherently better than others. We just cut the block into smaller pieces and transmit them in parallel while executing. You can call these small pieces blocks, mini-blocks, or shards; it doesn't matter. The ultimate goal is faster communication, not making the block itself better.
