Challenges in using cryptography - End-user and developer perspectives

Abstract

"Encryption is hard for everyone" is a prominent result of the security and privacy research to date. Email users struggle to encrypt their email, and institutions fail to roll out secure communication via email. Messaging users fail to understand through which most secure channel to send their most sensitive messages, and developers struggle with implementing cryptography securely.

To better understand how to support actors along the pipeline of developing, implementing, deploying, and using cryptography effectively, I leverage the human factor to understand their challenges and needs, as well as opportunities for support.

To support research in better understanding developers, I created a tool to remotely conduct developer studies, specifically with the goal of better understanding the implementation of cryptography. The tool was successfully used for several published developers studies.

To understand the institutional rollout of cryptography, I analyzed the email history of the past 27 years at Leibniz University Hannover and measured the usage of email encryption, finding that email encryption and signing is hardly used even in an institution with its own certificate authority. Furthermore, the usage of multiple email clients posed a significant challenge for users when using S/MIME and PGP.

To better understand and support end users, I conducted several studies with different text disclosures, icons, and animations to find out if users can be convinced to communicate via their secure messengers instead of switching to insecure alternatives. I found that users notice texts and animations, but their security perception did not change much between texts and visuals, as long as any information about encryption is shown.

In this dissertation, I investigated how to support researchers in conducting research with developers; I established that usability is one of the major factors in allowing developers to implement the functions of cryptographic libraries securely; I conducted the first large scale analysis of encrypted email, finding that, again, usability challenges can hamper adoption; finally, I established that the encryption of a channel can be effectively communicated to end users. In order to roll out secure use of cryptography to the masses, adoption needs to be usable on many levels. Developers need to be able to securely implement cryptography, and user communication needs to be either encrypted by default, and users need to be able to easily understand which communication' encryption protects them from whom. I hope that, with this dissertation, I show that, with supporting humans along the pipeline of cryptography, better security can be achieved for all.