kentucky wrote:

Given:
p is the true and unknown plaintext - (assume ASCII).
k is the true key.
c is the related cipher text generated by c = E(p, k)

What follows are decryptions of c using trial keys K1 and K2 to
give possible plaintext P1 and P2.

1. P1 = D(c, K1)
2. P2 = D(c, K2)

Questions:
What is the truth of the following statements:
(1) When Ki approximates the true key with greater than
50 % of its bits matching the true key,

There is no such thing as an "approximation" to a key.
A key is either correct or it isn't.

then the probability that Pi has the statistics of ASCII plain text
is greater than when Ki has little resemblance to the true key k.

This is nonsensical. There is no such thing as "resemblance to the
true key'. One either has the correct key or one does not.

(2) If you are armed with a sufficient collection of
Pi, which I now assume to approximate the true p,
one can use "superposition" of Pi to get even better
approximation, P', of the true plaintext, p.

By "approximation to the true p", I assume you mean that
it differs from p in some small number of bits?

In either case, your assumption is nonsense. Getter a
'better approximation" in the sense that you mean is impossible
for any correctly designed cipher. If it is possible for some cipher,
then that cipher is weak and should not be used.

(3) The bit errors in each byte of P' can be corrected
by noting that in each byte of P' the number of bit errors
will be 0, 1, or 2 mostly, and that the bit corrections shall
result in meaningful ASCII text.

This is more Nonsense.

May I suggest that you get a textbook on the subject and read it
before making any more meaningless speculations?

.

• References:

## Relevant Pages

• How do you explain?
... I assumed that these were candidate keys that gave ASCII looking results. ... When Ki approximates the true key with greater than ... There is no such thing as an "approximation" to a key. ... for any correctly designed cipher. ...
(sci.crypt)