|
Electronic Voting Machines
|
VERIFIED VOTING NEW MEXICO |
Is fraud the basic danger to an accurate vote?
Lack of public confidence would constitute a more serious issue than
fraud. In securing a reliable voting system on which voters can rely,
it is important to avoid any situation where one is placing unnecessary
trust in a person, business, or machine. The maker of Australia's touch
screen system has said as much: "Why should you have to trust me?" A
properly designed voting process can remove the need for blind faith.
Fortunately, such a process is known and is available.
What are the two basic problems with implementing such a process?
Many say that the problems with voting machines are not serious.
Are they right?
No. Current analyses by the Library of Congress, General Accounting
Office, Maryland's SIAC commission, and qualified technologists have
found the Federal Election Commission and NASED certification and
testing process to be outdated, and that touch screens bring new and
unresolved vulnerabilities to the election process. Current
certification and testing is not yet sufficient to eliminate design
errors in complex systems. NIST has not revised its voting standards,
and current machines don't qualify even for the lesser standards used
in health care or military hardware applications. California, Louisiana
and NY State Assembly, for example, have requested
replacement/retrofits of current Touch Screens to permit voter
verifiable balloting and other improved standards.
How is a vote recounted in the event of computer malfunction?
When a voting machine malfunctions, do we trust its printout replicas
of the cast ballots and trust those for hand-counting? By New Mexico
law, the ballots are stored redundantly in the machine, if these two
memory banks disagree, which should be believed? If the malfunction of
the machine is too severe to permit a memory printout without
intervention by technical repair persons, might this compromise the
voting tally? Addressing this situation, the Republican party filed a
lawsuit in November of 2003 in Fairfax County concerning unsupervised
repairs during an election. Furthermore, ballots printed out from a
machine at day's end reflect what the machine stored and was not
reviewed for voter intent. Providing a voter-verified ballot as the
only legally recountable source would resolve these problems.
Is it expensive to hand-count paper ballots?
No, it is not expensive, because they can be manually run through an
optical scanner rather than handcounted. In any event, federal law
specifies that ballots must have a recountable manual audit trail.
Are optical scanners and paper ballots more error prone than
electronic systems?
Optical scanners are the most accurate method of counting votes ever
devised. An MIT-Caltech report concluded, from a study of several years
of records, that errors occur in DRE systems twice as frequently as in
optical scanners. In fact, DREs tie with punch cards as the least
accurate among all voting systems.
Optical scanning of ballots produced by touch screen systems can be
made virtually 100% accurate. They can be clearly printed without
indistinct pen marks, and use auxiliary bar codes with redundant error
checking and anonymous serial numbering to reduce their error rate to
nearly zero, far better than current bar code scanners in reliable use
everywhere. Rare errors tend to be observable rather than completely
undetectable, as is the case in DREs. All electronic touch screens
render mistakes permanent, with no way to recount.
Could paper ballots be more easily forged, stuffed or lost than
computer memory cards?
Computer generation of optical scan ballots can be made much safer than
traditional, hand-marked, indistinguishable ballot forms. The addition
of crypto-graphic signatures can render them virtually unforgeable and
unstuffable. Use of checksums and error correcting schema can make
ballots very difficult to discard maliciously.
What is open source?
Computers run on source code, sets of instructions derived from
readable text. These codes need to be examined for correctness. Source
codes that a company keeps secret are called proprietary, whereas those
made available for general examination are called open. A company which
permits examination doesn't thereby lose its patents and copyrights.
If we have voter verified ballots, why do we need open source too?
Neither code review nor voter verification is
foolproof. Touch screens occasionally record a choice other than the
button pressed by the voter. This might happen if the screen display
and button sensors are misaligned, or if there is a source code bug. A
sharp-eyed voter might catch this mistake on his voter verified ballot,
but might not. Will every voter check? Open source review allows a
better likelihood that bugs will be found; with closed source, we are
entirely dependent on voter verification.
Isn't closed source more secure against hackers?
No. Hackers routinely discover security holes in Microsoft's closed
source software. There can be competitive business reasons to keep
codes secret, but security is not one of them.
Won't all bugs be found by independent testing lab reviews?
No. The authors of the SIAC report on Maryland's software problems
conceded it is 99.9% certain that all security holes will not be caught
in any source review - especially if deliberately placed there. And
even if you did manage to eradicate them there is no way to be sure you
did. Because touch screen systems are of necessity too complex to
permit the sort of exhaustive inspecting simpler systems permit, the
many-eyes approach of open sourcing, although it isn't perfect, is
considered superior in eliminating bugs.
How can any voter or poll worker be sure that the source code is
what is actually running on the computer?
Certification of source code must be accompanied by means of verifying.
If a computer is connected to a network or modem at any point, it is in
communication with a remote computer outside the poll workers' ability
to monitor. This is true even if the connection is only for data
transmission as almost any direct electronic communication is
potentially bi-directional. This capacity to send and to receive would
compromise the integrity of the certified machine unless all
communications are made to pass through a non-erasable, inactive
medium, i.e., a CD-Rom which can be secured, duplicated, tagged, sealed
and otherwise controlled by poll workers as warranted.