The State of New Jersey passed a law in July 2005
requiring voter-verified paper records for direct recording electronic (DRE) voting systems by January
of 2008. Most of the systems in use in the state at the time the bill
was signed into law had no available printer with which the counties
could retrofit. However, when the market demands, vendors typically
will find a way to comply.
Last week New Jersey reported on tests of three
printers submitted for consideration to meet the states requirements.
Two were from Sequoia -- one for the Advantage electronic voting
machines used in most of the State's jurisdictions, and one for the
newer Edge e-voting machines used in Salem County . The third printer tested was for Avante. The Sequoia Advantage printer does "cut-and-drop" of the paper record into a secure receptacle, as does the Avante. The Edge printer does not cut the records into separate sheets. The Advantage printers are said to come at a price tag of some $2000 apiece
-- this may or may not take into consideration the cost of additional system
retrofits to allow the printers to be attached.
The State found a number of exceptions -- areas in which the printers,
in their determination, did not comply fully with the requirements as
stated. There were several which are of concern, and at least one which should render the system unusable.
The biggest problem could be one that may not look like much, at first
glance. The requirement is for the DRE system to stop recording votes
if the paper malfunctions, jams, the connection between DRE and printer
is lost, etc. In such an instance, the voter is allowed to vote on a
different system or on an emergency paper ballot. The testers indicated
exceptions for at least one system, as follows:
"If a mechanical error or malfunction occurs (such as a
paper jam, running out of paper, paper torn in half, or paper inserted
improperly), the DRE displays an error message on the screen to the
voter, but no warning signals are sent to notify the election officials
at the polling place. The DRE does not suspend voting operations. The
voter has the opportunity to continue voting and cast the vote but the
paper record is not printed. The vote is, however, electronically
recorded."
The same is true for situations where paper is low or runs out.
"Although the DRE can detect problems that may prevent
paper records from being correctly displayed, printed, or stored (like
paper jams or low paper), no warning signals are sent to notify the
election officials at the polling place."
If the failure to alert a poll-worker results in missing, jammed or
damaged voter-verified paper records, and voters can continue to cast
ballots without realizing the problem, that means election officials
will be unable to confirm the accuracy of the electronic vote count,
either through an audit or a recount. (New Jersey is currently working
on legislation to require random manual audits.)
Given that voter-verified paper records are new to many voters and to
poll workers, and notification to voters about the importance of
checking the paper record may be uneven or insufficient, it is crucial
that the basic safeguard of an alarm to let an election judge or poll
worker know about problems with the printer be fully functional.
According to these tests, that safeguard is not in place.
Another exception common to two of the systems tested was particularly interesting. In many states, voters get up to three
chances to mark their ballot correctly. You can spoil a ballot, try
again, spoil it a second time, but the third try is your last one,
typically. This is true in New Jersey as well.
Here's how it would work on an e-voting systems like the ones tested: When you are just about ready to cast your vote
after marking your choices on the electronic voting machine, the
system prints a voter-verifiable paper record -- so that you can
review it, and affirm that your choices were accurately recorded. If
you felt it was incorrect, you can "cancel" that record and go back to
any of the contests on the ballot using the electronic machine, and
re-select. Ready again to cast your ballot? Checking the paper record,
you see it still doesn't reflect what you want. You cancel a second time.
Now you're on your last try, so you mark your choices carefully.
You're ready to cast, but the voter-verified paper record prints --
and then quickly drops into the locked receptacle, too quickly for you
to have the opportunity to review it.
Now, the regulations require that if the voter-verified paper record
does not record accurately, the machine must immediately be taken out
of service, and that voter gets to vote on a machine that works. If
your check and re-check of the paper record resulted in you cancelling
it because it did not seem to record your choices accurately (rather
than because you missed something), you would get to vote on a
different machine --one that was recording your choices faithfully-- and that malfunctioning one would
be taken out of service.
So the State's testers correctly noted this hasty
"print-and-disappear-from-view" record as an exception in their report.
The voter must be able to affirm, whether it is the third try or the
first, that his or her choices were accurately recorded. That's the
whole point of the "voter-verifiable" paper record. You get to verify.
That's the document that is used in recounts or in audits to check the
voting system for accuracy, so if you don't get to see it, the audit
cannot be legitimate. It's that simple.
Now for the chilling part: Sequoia's New Jersey law firm came back with a
posted response to the various exceptions, and about this one they state:
"Since no further amendment of the vote is possible under the
Regulations, an opportunity to confirm the third choice is irrelevant
and superfluous."
Thus the one OFFICIAL
record of your vote to be used in recounts and audits may or may not be printed correctly, but you don't get to check.
It's of deep concern that this vendor's representation seems not to
grasp the significance of the voter-verified paper record. It's not
(just) to make voters feel better about the voting system, it's there
to provide a way to check the tally for accuracy. Failure to make it
available to the voter for review eliminates that failsafe.
When using a paper ballot, you, the voter, get to decide when you're
ready to cast it -- the equipment doesn't snatch it away before you get
to review it.
These concerns, coupled with the cost of the printers, bring to mind a
serious issue. New Jersey spent, as best we can calculate, somewhere
between $70 and $90 million for these systems, when you add in the cost
of the printers. They spent this money on a system which does not
provide dual switch input for voters with disabilities relating to
mobility and manual dexterity. They spent it on a system which does not
provide audio readback from the paper ballot for voters with visual
disabilities who want to be able to review their ballot for
accuracy.1
They spent it on a system known to have higher than usual residual vote
rates.2
They could have had paper optical scan ballots and ballot-marking
devices in every polling place for 1/2 to 1/3 of that cost, around
$34.5 million. Don't throw good money after bad, New Jersey. It's still not too late.
1. The NJIT reports indicated that the systems
must meet Federal and State standards for accessibility, which shall
include, but are not limited
to, an audio component that shall accurately relay the information
printed on the paper
ballot to the voter. However, the methodology for testing to this
standard is merely to run a mock election to check the accuracy of the
information printed in that particular mock election. It does not seek
to ensure that the audio stream relay the information from the paper
itself. Thus a voter who is unable to visually review the paper record
in an actual election is only able to audibly review the electronic
record and has no way to know if the paper was printed correctly. 2. The residual vote rate on bottom-of-ticket
items was a stunning five times higher on NJ’s full-face DREs than on
paging DREs in a recent study. Even a highly publicized top of ticket
Senate race showed a full percentage point worse performance on the
full-face DREs.
|