CS Colloquium (BMAC)



Computer Science Department and the Electrical and Computer Engineering Department Colloquim
Gestalt vs. Constructivism: Designing Blind Cloud Storage in the Real Second

Speaker: Moti Yung, Security and Privacy Scientist, Snap Inc. and Adjunct Research Faculty, Columbia University

9:30 ~ 10:30AM, September 26, 2017

Where: Stadium 1204

Contact: Indrajit Ray (

Abstract: Design of a basic primitives involves specification, and then designing and implementing it. The foundations of systems approach, typically implies that the specification and design have to capture the primitive in its entirety (analogous to the abstract concept of Gestalt applied in art and science). Therefore the designer needs to capture all functions (correctness properties) and safe operation (security and privacy properties) based on the complete picture. In actuality, often one needs to retrofit a primitive into a live and existing system; this aspect captures the fact that systems are to be kept alive, the process of securing systems is ongoing, and engineering constraints exist in the real world while new functions need to be supported. As a result, in reality one has to design security primitives and understand them based on relations among existing and new components (analogous to the concept of Constructivism). In the classical body of knowledge of security the Gestalt approach and designing from the start dominate, and is strongly advocated, but in the reality of system building the opposite is true. The speaker will demonstrate the specification and the design of Snapchat's cloud memory which was introduced into the system with the main privacy goal of hiding the content from the server itself. This is an example of design in practice vs. design foundations. The existing component, the approach, and the novelty will be covered, and explained.

(Work done with Snap Inc. Security Group)

Bio:Dr. Moti Yung is a computer scientist whose main interests are in cryptography, security, and privacy. He is currently with Snap Inc. Yung earned his Ph.D. from Columbia University in 1988. He worked at IBM Research, was a vice president and chief scientist at CertCo, and was director of Advanced Authentication Research at RSA Laboratories, and a research scientist with Google. He holds an adjunct faculty appointment at Columbia University where he co-advised several Ph.D. students over the years. He has served as consultant to leading companies and to open projects with various governments as well. Yung’s major interests are building strong useful foundations for the field based on practical needs and reality of systems, and, on the other hand, transforming theoretical ideas into practice. He has contributed extensively to numerous new cryptographic ideas, foundations, techniques, protocols, and systems, and to new central notions like encryption, signature, and protocols (one example is innovating the notion of public key cryptosystems secure against chosen-ciphertext attack -- currently a major requirement from public-key encryption operating on the Internet). He has also contributed to innovative security and privacy industrial constructions embedded in actual large scale systems and networks. Examples of the latter are IBM’s SNA network authentication, the Greek National Lottery system, the security and privacy aspects of Google’s global systems such as the Ad Exchange (ADX), Google private Ads attribution based on off line user activities, and Snapchat's cloud storage security). He also predicted innovations that were later adopted by business (like "social based authentication for account recovery"). Further, his interest in the subtlety of the notion of "trusted systems" has led to predicting what is known as ransomware, and predicting the crypto subversion attack vector: a deniable algorithm substitution attacks on cryptosystems (which was allegedly mounted within the deployed American Federal Information Processing Standard detailing the Dual_EC_DRBG). He is a fellow of the ACM, IEEE, IACR, and the European Assoc. for Theoretical Computer Science (EATCS), and a recipient of ACM's SIGSAC Outstanding Innovation Award.