The {8am} strategy

[A.Wilmarth]

I wasn't to ask a couple of questions about your results if that is okay. Please don't take these questions as criticisms; I definitely do not have the authority to criticize anything, but rather I am coming from a place of asking myself similar questions.

I read the 3 related blog posts but I'm still not clear on RFD calculations. Is it:

Code:

abs(Freq in VMS- Freq - OTHER) for each term / number of terms * 100?

Related to this, what is the cut-off for maximum absolute difference allowed between VMS character frequency and other document letter frequency and how did you decide this? Maybe another way of asking this question is, judging from your reported RFDs and presuming you are multiplying by 100 to get percent, you are starting at raw absolute differences of .0005 or less, so how do you decide, hypothetically, that VMS c maps to N at a .0004 difference vs an S which was at say .0006?  Are the frequency differences much higher than I am thinking so they generally lead to only one choice, or are you always choosing the smallest?

Stats are a weakness of mine so I could be wrong here but because you are picking letters based on related frequency, shouldn't each anagram have a near perfect positive relationship at the get go? Is there another factor I am missing or is there a specific value your correlation coefficient covers which isn't already covered by the RFD?

Why are you (usually) choosing to count just the frequency in the herbal section vs the entire document and how do you choose which documents to "map" to?


Thank you!