Below we discuss a statistical method for discriminating between signal and noise. We might be able to tell whether a transmission from deep space is a message of some kind simply by doing scatter plots of presumed symbol pairs.
This leads to a cryptological point: the use of dummies is not neceesarily a protection against frequency analysis. When a dummy is paired with a symbol, the scatter plot may well show low correlation (no football shape). The code-makers must make sure in advance that every dummy pairs with every symbol in a likely ratio.
Otherwise, code crackers can check correlations and weed out all the dummies. They are then left with the simple task of doing a routine frequency analysis on the remaining symbols.
Yet I can't help thinking that for dummies to be useful that they will either correlate poorly with symbols, or that, if they are designed to correlate strongly, the correlation will not be characteristic of letter/number correlations.
Computer codes don't ordinarily rely on dummies. But, secret messages are sent via all sorts of means. Low tech messages might easily rely on dummies. If you use such a system, beware.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment