I never said happy accident. Like most algorithms (the reallocation size of a vector when it reaches capacity, the widths of slots in a timer wheel, the hierarchical sizes of a generational gc, etc.) the "correct" answer is determined experimentally. My point is that the choice of encoding small lengths is just a decision, like every other decision and is no more or less elegant than encoding a payload determining marking bit, and shouldn't be celebrated independently. Calling me "bafflingly arrogant" though is a bit perplexing. Can we disagree without resorting to petty jabs at character?