sbp

This paper ignores some very basic features of the VMS. The Ghibelline merlons on f86v, for example, show that the manuscript was produced with knowledge of northern Italian castles. Tucker and Talbert entirely fail to address this.

http://www.jasondavies.com/voynich/#Ros/0.767/0.174/6.00

Another strong refutation is that Tucker and Talbert believe the text of the manuscript to be a simple transliteration of Nahuatl. This has already been effectively ruled out by Kevin Knight, who shows for example that bigram predictability far exceeds most natural languages.

http://www.isi.edu/natural-language/people/voynich-11.pdf

(Section 4.2).

It is important to note in conjunction with this that he also found that the VMS has extremely weak word order, and that the line unit has significant features that distinguish it from paragraph features. (Sections and 5.1 and 6.2) These features combined make it very unlikely that the VMS is as simple as Tucker and Talbert think.

Ultimately, the VMS has remained a very tough nut to crack, even after repeated attempts at smashing through the outer layers by several generations of professional and amateur cryptanalysts. The longer this situation continues, the less likely there is to be any simple solution such as we had for the Soyga tables.

http://www.dtc.umn.edu/~reedsj/soyga.pdf

Yeah, I saw the explanation. It should be possible to calculate the average rate of submission, and therefore the ID allocation rate, and then increment by that plus some safety margin when restoring.

It's just ironic that the first time in nearly 1500 days of lurking on HN I actually decide to post, the site itself eats my submission.