2,562 characters resolve to 196 documented parts, in a sequence where each part's real weight in the writing system, not an alphabetical guess, decides what a learner sees next.
Radical-like components that attach to a base character and recur across families of meaning: 扌 (hand) gates 142 characters including 打, 找, 指 and 接. 氵 (water) gates 135, including 汽, 酒 and 湯. These are the pieces a reader learns to spot on sight.
Whole characters that are also common enough, and structurally central enough, to function as building blocks for other characters. 口 (mouth) is the single highest-leverage part in the whole set, gating 174 other characters.
The full character set is sequenced in 103 iterations, each one adding only characters that the parts already introduced make readable. By the final iteration the sequence covers 7,187 words at 100 percent of the working vocabulary it targets.
Most character lists are ordered by frequency, by grade level, or not ordered at all. This one is ordered by leverage: a part goes in when it unlocks the largest available block of new, readable characters, and a character goes in only once every part it needs has already been taught. The first block of 25 characters this way already reaches 38 words. By character 120, the count is at 82 new words for that block alone, a leverage ratio of 3.28, meaning every new character taught was worth more than three words a reader could now recognize.
That is what "engineered, not memorized" means in practice. It is not a claim about the script being simple. It is a claim about the order you learn it in being solvable, and this set is the solved order.