A modest proposal for electronic
voting machines in the US
December 15, 2006
Anyone following the controversies in recent US elections is aware of the
negative role that has been played by electronic voting machines. Yet there
is no reason for this. Electronic voting machines should not be expensive,
they should not have hidden software and they should not be usable for fraud.
So what went wrong? If I felt paranoid, I’d say they are just what they
were designed for, instruments for fraud, for secrecy, for confusion and for
making a few people rich. But even if I were not paranoid, I’d have to say
that they are overpriced, untested and do not encourage public trust in the
process. What went wrong is a lack of debate on the speciﬁcations.
My suggestion for such speciﬁcations is that the computer should simply
help the voter print a legal vote (i.e. one that is possible). When the printout
is all right, it becomes the ballot and is dropped in an old fashioned ballot
Since that might seem to be too vague, I detail below a proposal for
setting up a set of machines for the next elections that are not only very
inexpensive and foolproof, but also a huge improvement over anything that
has been implemented in the US!
2 Details of the proposal
First, the machines will be standard Windows computers, equipped with
a functioning black and white Laser printer. Since these machines already
exist in front of most workers in America as well as in most homes, the
investment on election day shall just consist in transporting a bunch of PCs
to the election hall, a minute fraction of the exorbitant price charged by the
current system providers.
Second, the PCs will all be running the same public domain software
(to be developed by volunteer programmers under the the Free Software
Foundation license) with the following speciﬁcations:
In its setup phase, it will accept the jobs that candidates are running for,
together with the applicants and their party in a standard ASCII format.
For instance, it could look like this:
NJ, district 125
President of the US, George W. Bush (R), John Kerry (D)
Governor of NJ, John Doe (R), Jane Smith (D), Joe Blow (I)
Senator of NJ, ...
If the input ﬁle does not have the right format, the program would reject
it, or ask for clariﬁcation. Because everyone could have a copy of the software
on their PC months in advance of the election, it would be surprising for a
ﬁle to be rejected on election day.
In its operating phase, the software would simply guide the voter till a
legal choice has been made for each position:
For president, half the time, the machine would say:
“Please vote for John Kerry (Democrat), for George W. Bush (Republi-
can) or abstain.”
And half the time it would instead say:
“Please vote for George W. Bush (Republican), for John Kerry (Demo-
crat) or abstain.”
When all the voter’s choices have been made, the machine would display
them on the screen and ask the voter to conﬁrm his choices. If he does, the
screen would be printed on standard stock and the paper would be compared
to the screen. It should be very rare for the two to be diﬀerent and of course
cause an investigation. When the voter is happy with the printed version of
his vote, he’d tell the PC that he has voted, then take it to the ballot box
and drop it in. The machine would count his vote for research purposes but
only the ballot box results would matter.
Every hour, the machine would send a tally of its votes so far to a central
PC on the election hall director’s PC. As indicated in the previous para-
graph, this would only be for research purposes and would have no legal
Finally, in its ending phase, the PC would print a total of the votes cast
for each candidate, again only for research purposes.
Because the computer in the proposed scheme is only used as a help in the
voting process, it is very hard to imagine how it could be used to cheat.
Therefore, I believe that it would not be used for cheating. What else could
go wrong? The only thing I can see is that the printer might fail, or be out
of paper. But that is hardly a show stopper. A new printer could replace
the old one on the spot.
Of course, people might tamper with the ballot box, but this has little to
do with this project. To prevent fraud, many people will have to be present
and keep eyes on the ballot box and all the other procedures which might be
illegal. No one ever said that democracy comes cheap!
4 Request for feedback and help
If one is willing to accept a certain level of complexity and if an additional
requirement is to be able to detect fraud, it is well worth considering Ronald
Rivest’s article “The ThreeBallot Voting System,” which can be found at
Please e-mail me at firstname.lastname@example.org if you have feedback or if you
would be willing to work on the actual software.