What are the characteristics of Labelese?

[RenegadeHealer]

In conclusion any encryption theory of the Voynich must take a realistic account of what would have been mentally practical for the author.

I second this point. The output of low tech, analog, human randomization methods is harder to simulate with a computer model than one might intuitively think. This is because the factors that go into a random drawing of the old-fashioned kind are too many, subtle, and varied to be completely quantified. But complete quantification is kind of what a computer does. I'm not arguing it's inappropriate to use a piece of software to model a proposed analog random output generator. After all, unless Voynich studies has an eccentric billionaire patron, no one is going to pay human research subjects to spend hours a day for many days in a lab drawing lots and writing down the resulting vords on parchment by hand. But what I am saying is, the builders and users of random word generators should never assume a priori that their tool models the analog generation process they propose nearly exactly. Voynich-bot builders need to ask themselves: What aspects of the 15th century analog method does my machine not model well? And, do these discrepancies likely account for the statistical differences between my machine's output, and the real VMS's text, and how do I know? These questions deserve a thorough discussion in the "possible sources of error" and "further avenues for research" that conclude any good laboratory report.