How does Kaspa choose which blocks to use for difficulty adjustment?

Kaspa samples a subset of blocks from its DAG rather than using every historical block to calculate how hard mining should be. The sampling method is designed around four guarantees: it is deterministic (every node reaches the same answer independently), uniformly distributed (all blocks have a similar chance of being selected), incremental (selection status can be passed forward without recalculating from scratch), and secure (a miner cannot cheaply manipulate whether their own block gets picked — doing so would cost as much as throwing away a valid block entirely). This matters for beginners because a well-designed sampling mechanism keeps mining difficulty honest and manipulation-resistant, which in turn protects the whole network's security as Kaspa's blockDAG grows.