DOS and Windows XP are both binary (unlike genetics). Anyone who knows anything about software can easily see a pathway, in fact, multiple pathways between the two.
DOS and Windows XP can both be considered as messages.
Anyone familiar with cryptography can tell you that to extract one message from another requires a Key and an Algorithm (which could be as simple as a bitflip, but more than likely would be much more complex).
The math in this thread effectively covers the mathematical probability/improbability of such a Key self-forming on its own at random (i.e. without intelligent intervention).
Essentially (though the actual number can vary slightly based upon definitions/pre-conditions used), the Key would have to be less than 96 bytes long for even a remote chance at forming "naturally."
Can you show a 96 byte Key that can extract the Windows XP "message" from DOS?
Not on your life.