[Cryptography] A software for combining text files to obtain high quality pseudo-random sequences in practice

[mok-kong shen]

Am 11.07.2017 um 23:43 schrieb mok-kong shen:
> Am 10.07.2017 um 18:38 schrieb rcs at xmission.com:
>> In English, in normal Ascii, the vowels are odd.
>> With your encoding, vowels are even -- the low bit is 0.
>> Vowels are pretty common, around 50% of typical text.
>> This will spoil the statistics of your random bits.
>>
>> Rich Schroeppel
>> rcs at xmission.com
> Your point is certainly true, many thanks for calling attention to 
> that. That would imply
> though that one should be more careful in interpreting the test 
> statistic from Maurer's
> test. I am attempting  currently to find means to mitigate against 
> that bias and hope to
> be able to report soon.

[Addendum] I like to report that I have already finished the 
computational work and shall
release a new version 1.1 of my software within a few days.

What is of interest to you: A check of the least significant bits of the 
5 bits of the original
text files coded according to my scheme shows that the situation is not 
too bad after all.
The ratio of 0's to the total number of bits checked varies roughly 
between 0.52 and 0.57,
i.e. not as critical as your argument would seem to indicate. After the 
combination of 4 files,
that ratio is almost 0.5. So my suspicion of the possible 
non-sensitiveness of Maurer's
test is entirely wrong. My new version will include a function zeros() 
that can be used to get
the said ratios. So, if you feel not certain of my report, you could 
easily check that. (Of
course you can also very simply write a function to do the check 
yourself, if you don't like
to wait the release of my version 1.1.)

M. K. Shen
to my scheme bits from the