The Voynich Ninja

It should be easy to think that the creator(s) of the ms were not the only individual(s) to have seen it when now lost context concerning use and contents remained within living memory.

Not having seen any, what are your imagined probability distributions for:

1) the number of people who could decipher it or knew definitively that it is not readable and why (not necessarily the same as number of creators)
  -max at any one time
  -in total

2) the number of people who had seen at least part of the ms (or a copy) in person, in conjunction with someone from 1); in other words, number of people who had seen at least some part of the ms "demonstrated" or "used" by someone who could decipher it or knew definitively that it is not readable and why.
  -max at any one time
  -in total

3) the number of people who were aware of the existence of the ms, but had not seen it (or a copy) in person, while at least one person from 1) was alive
  -max at any one time
  -in total

4) the maximum number of complete copies of the ms at any given time while at least one person from 1) was alive

5) the maximum number of partial copies of the ms at any given time while at least one person from 1) was alive

Rather than give more complete distributions, because that would take longer, I'll come up with boundaries and a guess at the median. Conceivable upper bounds can be much higher under improbable scenarios. The assumption for all of these is that the ms text contains enciphered information. 

1a) 1 to 1000. Median, 10 or fewer.
1b) Within a scale factor of 2 of 1a.
Why: These things could be on a need to know basis. Under many arrangements only a subset of those who access the information would need to know. 

2a) 1 to 10,000. Median, 100 or fewer.
2b) Extrapolate from 1b.
Why: Why go to all the trouble to protect the information if it's for a wide audience?

3a) 0 to 40,000. Median, 200 or fewer.
3b) Extrapolate from 1b.
Why: This should not be much higher than 2 unless the audience is bad at keeping secrets (probability unknown, but reasonable). 

4) 1 to 10. Median, <2. Full ciphertext copy only qualifies.
Why: Secret, difficult, expensive to copy. 

5) 0 to 10,000. Median, dozens. Ciphertext or plaintext partial copies qualify, even if the latter are generic, as long as they come from the ms.
Why: Depending on what qualifies as a partial copy, it's easy to imagine either very many or very few.