[split] Verbose cipher?

[Koen G]

The question of h2 vs h1 is very complex. Both don't necessarily correspond in a "linear" way. What I mean is this:

You can think of what I've been doing as a branching path. I replace something first, based on desired h2 and h1 behavior. Then I replace the next thing based on what performs best next, and so on.

But maybe the best path is one that first sacrifices a bunch of h1. And then suddenly after x transformations, *poof* suddenly your values look like Voynichese.

Basically with a genetic algorithm approach, you lock yourself into certain initial conditions. Maybe if you just brute-force all possible combinations and pick the one which has h1 and h2 values closest to Voynichese, you might get better results. Maybe not, but it's possible.

The problem of word length and spaces remains the same though.