RE: Mark Knowles' theories of pharma and other labels
ReneZ > 10-04-2020, 07:42 PM
The work of Timm and Schinner shows a great amount of statistics which are interesting and useful. Some of these are related to auto-copying but others not (e.g. the existence of a network of similar words).
However, none of the statistics actually tell us whether this is the result of an intentional autocopying, or whether this is a side effect of some other mechanism that was used to generate the words. (I have already shown this with two examples, in one of the threads dedicated to this paper).
The conclusion that the text is meaningless is not warranted from the data presented.
When Gordong Rugg presented his work, I challenged him to actually create a table and a grille that would be able to regenerate one page in the MS.
He did not do that.
Now, I know that it is actually trivially easy to do that. The Rugg method has this as a very clear 'advantage' over the method of Timm. Every single page in the MS has lots of words that are not a small edit distance away from previous words, so it can't be done for the Timm method.
Of course, this is based entirely on reverse engineering, so in the end it does not mean much.