Lecture 013

Guessing Pin

$\frac{\Sigma \vdash e_1 : l_1 \quad \Sigma \vdash e_2 : l_2}{\Sigma \vdash \text{ match }(e_1, e_2) : l_1 \sqcup l_2} \text{ match } F$

Match is kinda safe because the number of guesses to determine password is large.

$\text{guess} := 0; \text{ while } \lnot \text{ match }(guess, password) \text{ do } guess := guess + 1$

Declassification

$\frac{}{\Sigma \vdash \text{ declassify }_l e : l} \text{ declassify } F$

Note that the safety of declassification depends on the context.

$\text{ declassify }$ mean grade for entire class is safe.
$\text{ declassify }$ mean grade for a class consists of 2 student is unsafe.
$\text{ match }$ is safe for password.
$\text{ compare }$ is unsafe for password.

If $\Sigma \vdash \alpha \text{ secure }$ and $\alpha$ constains exactly one instance of $\text{ declassify }(e)$ , and $(\forall x \in \text{ use } e)(x \not\in \text{ maydef } \alpha) \implies \Sigma \models \alpha \text{ secure }$ .

Table of Content