Cryptography Reference
In-Depth Information
Material Objects
In visual cryptography, assumptions regarding the relative difficulty of
number-theoretic problems are traded for assumptions regarding the capa-
bilities of the human visual system. This exchange leads to a series of
schemes categorized according to trade-offs between protocol security and
other design constraints. Pursuing an intriguing path, Moran and Naor
have argued that it is also possible to further extend the “standard model”
by including the capabilities of some of the traditional paper-and-ink
objects that have provided security in the physical world
—
for example,
sealed envelopes and scratch-cards (such as those used for instant lotteries).
They model such objects as “tamper-evident seals,” defined as “a crypto-
graphic primitive that captures the properties of a sealed envelope: while
the envelope is sealed, it is impossible to tell what's inside, but if the seal
is broken the envelope cannot be resealed (so any tampering is evident).”
54
Not only are the protocols designed to be implemented using actual physi-
cal envelopes, but Moran and Naor place explicit upper bounds on the
number of rounds and envelopes that can be used so that the protocols
remain practical for humans, rather than mere theoretical constructs. Just
like the visual authentication schemes, the security of the resulting proto-
cols does not rely on any computational assumptions but rather on the
(physical) tamper-evident properties of the envelopes.
55
These tamper-evident seals are introduced in the context of an informa-
tion security problem with important implications: survey-based method-
ologies encounter elevated bias when investigating socially desirable or
undesirable behaviors (e.g., voting preferences or immigration status), as
survey respondents tend to lie. One solution is to use a “randomized
response” technique, first proposed by Warner in 1965.
56
In its simplest
version, the respondent privately flips a coin before answering the ques-
tion. Heads, he answers truthfully, tails, he lies. Given that only half the
population surveyed will have answered truthfully, the researcher merely
doubles the observed response to obtain the true proportion. Yet individual
respondents are shielded by the uncertainty generated by the coin toss, as
each is equally likely to have lied
—
that is, the protocol provides them with
plausible deniability.
As befits the work of cryptographers, Moran and Naor's protocols are
designed to address the issue whereby a respondent maliciously deviates
