Description and analysis of the RIES Internet voting system

RIES is an evolving family of systems (RIES-2004, RIES-KOA, RIES-2008) for electronic elections via the Internet. It has been used in practice for medium scale elections for the Dutch District Water Control Boards and for expatriates in national parliament elections. We describe and analyze the security of RIES, based on the documentation made available to us. The aim is to make RIES easier to understand for all parties involved: policy makers, scientists, election officials, implementors, and the general public. This document makes explicit what the assumptions in RIES are, what kind of restrictions apply, what level of security is achieved, etc., focussing on the security aspects, both technical and organizational / procedural. RIES provides integration of Internet voting and voting by regular mail, and has been developed in that specific context. This has set the framework of requirements for the design of RIES to comparison with postal voting systems. Hence certain reasonable goals for elections (like vote freedom) have been out of scope from the start. Consequently, the general voting requirements formulated by the Korthals Altes Committee are not all satisfied: not only vote freedom but also vote integrity and confidentiality are not structurally guaranteed in the RIES design. RIES is built on certain cryptographic primitives, like one-time signatures. Keys for individual voters are generated centrally. There are no anonymous channels. The structural protection and safeguards offered by cryptography are therefore rather limited. Many of the guarantees in RIES thus rely on organizational controls, notably with respect to (voter) key generation, production of postal packages, insider attacks (especially at the server), integrity and authenticity of the software, and helpdesk procedures. RIES-2008 is designed and built in an open spirit. Its source code and documentation will shortly be available openly for inspection and analysis. Additionally, the designers and organizers have put considerable effort in publicly explaining and discussing their system. The technical and organizational set-up seems carefully designed. There are however pragmatic elements in the system, such as the use of replacement packages, that are open to manipulation and abuse, notably by insiders. The RIES Internet election system also offers potentially dangerous ways for manipulation of elections, in principle applicable on a large scale and different from attacks on postal elections. One of the distinguishing aspects of RIES is that it allows independent recounts of the final outcome and individual checks to see if own votes have been included. This interesting and useful feature does however not compensate for the structural weaknesses. In a larger context we see RIES (esp. RIES-2008) as a project that yields valuable hands-on experience and expertise on how to organize and run electronic elections. We do not think RIES-2008 is a suitable system for use outside a context of postal elections, and in particular not for ‘general’ elections (like for national/European parliaments or local/regional councils). We do encourage further research, development and experiments to gain more experience in this area.