The effectiveness of the patch randomization algorithm is a direct result of the monolithic synth engine design. It was inspired by nature.
Because there is “just” one synth engine that is responsible for all timbres (from percussion to basses and leads), it means that every patch on the Woovebox shares the same “reference genome”.
Randomizing a patch then selects two (or more) specimen (presets) and sees them “procreate” to have a baby, which then has traits of the parents.
It ensures the offspring patches are (usually) sonically coherent and useful. By procreating exclusively within the “tribe” of sound category (lead, bass, kick drum, etc.), the resultant patches also fit within that sonic category.
The alternative “lucky” randomization mode does away with the sound category “tribes” and expands the pool of possible parents to the entire preset pool, allowing - for example - a kick drum to procreate with a lead. This can produce interesting results, but is obviously less predictable in terms of usability.
This entire thing was planned from the start, and is one of the major reasons why I did not want to go the - almost ubiquitous amongst other grooveboxes - dumbed-down “mini engines for different timbres” route.
TL,DR; presets are having sexy time every time you press the randomize button