What is Kaspa's sparse window for difficulty adjustment?

Kaspa's sparse window is a technique for computing mining difficulty by sampling a well-chosen subset of past blocks instead of examining every block in the recent history. Rather than scanning all blocks in a time window, the sparse window picks a smaller set that meets four properties: it is deterministic (every node calculates the same result), uniformly distributed (spread evenly across the full window), incremental (future blocks can inherit it without recalculating from scratch), and secure (a miner cannot cheaply manipulate which blocks get sampled). The window is described by two parameters — `length` in seconds and `size` in blocks — so Kaspa can express a target like "sample 1000 blocks from the last 500 minutes" regardless of how many blocks per second the network produces. For a beginner, this matters because it keeps the network's difficulty calculation honest and tuneable without opening the door to manipulation.