They don't want you to know there is a PoW out there that has an insane ceiling.
How kaspa:native retunes every 100 milliseconds
The job
Ten blocks a second only holds if difficulty moves with hashrate. If the target is stale, blocks bunch or stall. Kaspa retunes the target about every 100 milliseconds so the clock stays at 10 BPS.
Before Crescendo, the node inspected a full window of 2,641 blocks, about 44 minutes at 1 BPS, and did that work every block. At 10 BPS that window would be 26,410 blocks. Too heavy.
KIP-4 switched to a sampled window. Same 2,641 seconds of history. Blocks sampled about every 4 seconds by DAA score. About 661 samples. Same duration. Less work per block.
The miner writes bits and daaScore. bits is the compact target. daaScore is the network’s second-counter for monetary months and for the sample window. Nodes recompute both. They do not trust the miner.
Why it matters: Emission is per second. Difficulty is per second. Block rate can change without rewriting money. Crescendo multiplied blocks by 10 and divided per-block subsidy by 10. The DAA kept the same wall-clock window.
So, kaspa:native does not wait a week to notice hashrate.
It samples the last 2,641 seconds, every tenth of a second, and keeps 10 BPS as a rule. Consistency.