Testimony by
    Ronald L. Rivest
    (MIT Institute Professor, Cambridge, MA)

Venue:
    Hearing held by
        Committee on House Administration
        Chairperson: Rep. Zoe Lofgren (California)

Hearing Title:
    "Exploring  the Feasibility and Security of Technology to Conduct
    Remote Voting in the House"

Date:
    Friday, July 17th, 2020

Time:
    1:00pm EST


Testimony (both oral and written):
==================================

Chairperson Lofgren and members of the Committee: I thank you
for inviting me to testify regarding the feasibility of using
technology for conducting remote voting in the House.

My bottom line is that such remote voting is feasible and can be made
adequately secure.

By way of introduction, I am an Institute Professor at the
Massachusetts Institute of Technology; my background includes computer
science, information security, cryptography, and election security.

I am well known for the invention, with Adi Shamir and Len Adleman, of
the RSA public-key cryptosystem, the first (and still widely used)
implementation of public-key cryptography, enabling both secure
communications and digital signatures.

I have worked for over two decades on voting security.

I was a member of the Technical Guidelines Development Committee from
2004--2009, advisory to the Election Assistance Commission; I chaired
the subcommittee on Computer Security and Transparency.

I am a founding member of the CalTech/MIT Voting Technology Project.

And I am on the Board of Verified Voting, a non-profit promoting voting
system security, especially through the use of risk-limiting audits.

I speak here only about the security aspects of remote voting, not about
the appropriateness of remote voting for the House; that question is
beyond my pay grade!

I see that the House, under Resolution 965, is already using proxy
voting for remote voting.  That resolution also authorized the
examination of ways to vote remotely in a secure manner; hence today's
hearing.

As noted, I think the House is in a good position: there are indeed
suitable secure voting technologies available.

The important reason why that is true is that House votes are NOT
SECRET.  Voting in the House is not based on secret ballots.  

That makes all the difference, as manipulation or alteration of votes
can be detected and corrected.

For the record, I note that in the US, SECRET ballot voting was first
implemented in Massachusetts in 1888.  However, implementing secure 
secret ballot remote voting is still beyond the state of the art.

Designing a secure voting system requires, first of all, a clear
statement of the security objectives.  A system can't be said to be
secure if there is no specification of what security should mean for
that system.  What are the baseline voting security requirements?
Here are four:

    (1) Only eligible voters can vote, at most once each.

    (2) Votes are cast as intended.

    (3) Votes are collected as cast.

    (4) Votes are counted as collected.

Each property should not only be true, but be VERIFIABLY true.

Counting (tabulation) is not an issue, since non-secret ballots can be
posted publicly and the tally then verified by anyone.

One recommended principle for achieving voting system security is that
of SOFTWARE INDEPENDENCE, a notion developed by John Wack and myself.
This principle basically says that you never want to be in a position
where you have to say, "Well, the result must be right, because the
computer says so!"

In other words, the election outcomes must be AUDITABLE.

Here a sketch of a simple architectural approach for secure remote
non-secret voting, to illustrate:

    -- there is a public web site where all cast votes are posted

    -- each congressperson composes his/her vote, digitally signs it, 
       and sends the resulting digitally signed ballot for posting on 
       the public web site.

Many digital signature schemes are available; NIST has developed
digital signature standards.  Digital signatures are now implemented
in every browser.  One approach uses the RSA public-key cryptosystem.

A nice thing about digital signatures is that the signature on a
digitally signed document (such as a ballot) is verifiable by anyone.

Note that a digital signature is not just a cut-and-paste image of
a handwritten signature; it is a mathematical function of the message
being signed and secret information specific to the signer.

Digitally signed ballots can be authenticated using public
information, both as to origin (who the voter is) and as to content
(what the ballot says).

Vote manipulations are not possible, as forging digital signatures is
not feasible. 

The most an adversary can do is to delete or duplicate votes.

An adversary can conceivably delete or duplicate votes even now,
with proxy voting.  If a congressperson can't submit a ballot, they
can't vote.  Detection and correction mechanisms can work
for voting with digitally signed ballots as for proxy voting.

It is important to note that voters (in this case congresspeople) can
check, or audit, that their votes are correctly recorded on the public
web site.  Missing votes can be restored.

The approach sketched here bears many similarities to your current
"proxy voting" procedures; the public web site becomes the "proxy" for
those voting remotely.  Indeed, such a system should provide a smooth
and secure extension of your current proxy voting procedures, which
need not be abandoned.

This sketch is intended only to show that it is possible to use
technology to do remote non-secret voting in a secure manner; other
approaches are possible.

This concludes my testimony; I would be happy to answer any questions
you may have.