You are on page 1of 1

Name: b03901156 Date: February 26, 2017

Paper: R. Anderson, Why cryptosystems fail, in Proceedings of the 1st ACM Conference on
Computer and Communications Security, 1993.

Summary: The author analyzes the reason why a crypto product adhering to strict standards
set by crypto community fails repeatedly when it is deployed in real world. The author uses
ATM as a case throughout this paper to explain most attacks on the crypto system are allowed to
happen due not to weaknesses of crypto product but due to implementation and management errors:
human error, negligence, malignance, incomplete standard, inadequate quality control or feedback
loop. The author believes that the current view of computer security systems should follow the
airline models which enables information owners to control and trade their own property, instead
of trusting everything to a centralized administrative structure (railway system).

Strengths of the paper: The paper presents a strong viewpoint that when analyzing security
systems, we should take the environment where the systems are used into consideration. The author
gives the informative insights into the working of banking systems and includes many interesting
examples in real life to explain how frauds actually happens. In addition, the author illustrates the
analogy between secure systems and safety critical systems and points out that a false threat model
is being accepted due to the lack of feedback when it comes to systems security issues.

Weaknesses of the paper: This paper only outlines the problems with the security model briefly
but doesnt give the specific information about the model. The author doesnt provide statistics
and details when depicting the monetary loss of banks and customers due to security problems. In
addition, the author presents that the current computer security systems should follow the airline
paradigm to put more intelligence in the human operator; however, the viewpoint may be not
practical since it is difficult to train people well and do quality control of their work.

Reflection: According to the authors viewpoints, a false threat model is being accepted because
of the lack of feedback when it comes to systems security issues. I think it is quite true since
from the angle of an engineer of cryptographic systems, it is indeed quite difficult to obtain the
information on how the systems fail from the computer users; in addition, from the computer users
views, it is hard for users to learn lessons from the previous attacks or mistake, which is the main
reason why repeat problems keep happening just like the cases illustrated in this paper.
However, I do quite disagree with the authors conclusion that current computer security sys-
tems should be designed according to the aviation industrys paradigm to put more intelligence on
human operator, which indicates that a well-trained crew is the first line of defense. I think it is
not quite true since training people well is difficult, not to mention preventing people from making
mistakes when executing the security systems. I think the aviation industrys paradigm is much
better than railway industrys paradigm applied in security systems since it is quite flexible, inter-
active and more practical in real lifes cases, but the paradigm still requires adjustments to decrease
the errors due to humans implementations or managements and solutions used to improve human
imperfections.

You might also like