Why ten blocks a second is hard
The problem chapter, told as a story over one recorded window of real blocks. Every figure in it comes from that chapter's own sourced material: ten blocks a second, a 250 millisecond delay, and the roughly 71 percent a single chain would throw away at that delay.
Recording: Nov 4, 2026 · 247 real blocks, recorded from api.kaspa.org.
The first forty blocks of the recording are already in the view. Scroll on: each step changes one thing, and together they tell the collision story.
Step 1 · The speed
Ten blocks a second is the goal. This block, near the start, points back at three blocks that came before it. At this speed, new blocks are always being found while the last ones are still spreading.
Step 2 · The collision
A block takes about 250 milliseconds to reach everyone. In that gap, two or three more blocks are found, and none of their makers has seen the others yet. That is the collision.
Step 3 · The old answer
A single chain would keep one of those blocks and throw the rest away. At a 250 millisecond delay it would throw away about 71 percent of all blocks found. Speed, wasted.
Step 4 · The kept web
Kaspa keeps them instead. The darker line is the selected chain, one path through the web. Every other block hangs off it, counted in the order, none of it wasted.
Step 5 · The whole window
The rest of the recording arrives: 247 real blocks in thirty-six seconds, about ten a second, each one kept, each one placed in the same order on every computer that checks.
Step 6 · The answer
The newest block points back at eleven blocks at once. That is the answer to the collision: never choose between blocks that arrive together. Build on all of them.
This story is the short form of The problem, which carries the collision math, the delay slider, and the sources behind every figure used here.