And Other Critiques
By Paul R. Lehto, Attorney at law
The following is why I think the Holt Bill (HR 550) does much more
harm than good. Regardless of whether or not you are a "paper ballot"
person, I think my arguments below apply with equal force. Election
systems act like kaleidoscopes, and amendments turn the kaleidoscope
and force patterns to rearrange. These turns can be very damaging and
unexpected.
Please Understand: I'm not trying to rain on anybody's lobbying
parade and I support election protection completely: But in any case,
I always recommend that we lobby for our VALUES not for specific bill
language subjected to future word-smithing and changes anyway, and may
or may not deliver the result intended....
bill accomplishes what it sets out to accomplish, and also whether
"gold standard" is a fair description. The fact that more and more
people are getting involved in the movement and asking for most or all
of the right things does not at all tell us whether a particular bill
actually delivers those promises.
The Holt bill has vetted its 2% audit requirement with "computer
scientists" which is great, but it also needs to be vetted from
statisticians for sure and perhaps even consumer fraud attorneys, and
I think it fails in that regard. (My co-author Dr. Jeffrey Hoffman
concurs as to the 2% opinions below, but isn't involved with the
rest).
THE GENERAL PROBLEM SEEMS TO BE THAT ALTHOUGH MOST PEOPLE'S #1 GOAL
IN PROTECTING ELECTIONS IS AN ANTI-FRAUD MOTIVATION, HOLT APPROACHES
IT AS A QUESTION OF MAKING TECHNOLOGY WORK. This may be because we
often don't want to insult the public officials, and it's hard for
Congress to sit down and have a serious discussion about "how do we
protect the people from the politicians and election crooks". But as
Jon Roland said a wise man prepares for the probability of corrupt
administration during a time of good administration, if that's what we
have now. Plus, to make an analogy, I don't think security experts in
banks stand for too much bullshit about "there's no risk of
embezzlement from the inside, just bank robbery from the outside".
Same with elections.
So please let me give some examples based on co-authoring a paper
actually scientifically and statistically auditing an election result
(I know others here have done so as well but I don't think everyone
has), followed by some legalistic observations. With a little overlap,
these are in addition to the 6 or so concerns I posted earlier, most
of which I've not heard any response to. The first one below on the 2%
audits is somewhat of a repeat and an expansion of the concern:
(1) SECTION 5 of Holt (Mandatory Manual "Audits"): The 2% audits on a
precinct or polling location basis means that the samples are
clustered, since all the votes in the entire precinct selected are
counted. The problem with Holt is not the 2%; it is that it is 2% of
precincts. One cannot do, with precinct-wide samples, what one must do
with a true audit: establish a firm margin of error and a "tight"
confidence level so that if the sample exceeds that, we can sound the
alarms. So, it won't be possible or it will be very unlikely that even
a cheating margin (in a reasonably close race at least) can come to a
statistical conclusion that something was wrong. (Thanks to attorney
and non-Ph.D. statistician Jonathan Simon for this) Press releases
will issue, and it will simply be announced, or we will be forced to
announce, that the results are 'within the margin of error' because
the margin will be quite wide indeed.
Moreover, although the EAC can do more audits if "cause" appears for
the same, when something is "within the margin of error" it is
probably not "cause' for further investigation as a matter of law, so
a court might well shut us down or shut the EAC down with the argument
that UNDER ITS OWN STANDARDS AND RESULTS there is no cause to look
further. The fact that citizens are invited to bid on the contracts in
order to truly waste their time with a futile exercise (with the
contract itself being a control mechanism and probably no access at
all in any event to trade secret software) is not an improvement.
It gets worse. Section 5(b) requires at least one precinct in each
(microscopic) county, which further forces small precincts to be
selected, further making the 2% precinct selection problem even worse,
though it the purpose of this provision appears to be to help spread
the sample throughout the state, we simply have no rational interest
in having statistics mimic the distortions of the US Senate, with 2
from each state regardless of size. Legislation simply does not alter
the rules of statistics as to proper sampling, and should not attempt
to do so...
In a nutshell, though an "official audit" be completed, nothing can
be proven from it due to wide margins of errors created by precinct
sampling, so that's the empty end result of the huge audit
investigation, and anybody who continues to be unsatisfied is a wacko
conspiracy theorist and attorneys will raise their prices and
encourage you to reconsider your desire to check more. It will give
audits a bad name, and the conspiracy theorists will be 100% correct:
the audit proves nothing.
(2) Even if the audit could prove something, which it almost assuredly
can not, all it proves is that there's a DIFFERENCE between paper and
electronic, not where that difference comes from. It will be argued
that the problems are with the paper, or in a real pinch even with the
DRE printing an incorrect ballot that the voters (according to one
claimed result in I believe an MIT study) don't really check any more
than they check their paper receipts at the grocery store (a glance
perhaps). Because the paper records under Holt are stored "in the same
manner" as other paper records in the relevant jurisdiction, there's
no improvement as to that (it may be good or bad with chain of custody
issues).
In a nutshell, the audit is almost guaranteed to be within margins of
error, and discrepancies (if any) can be blamed on the paper (with
DREs, primarily). Although "inconsistent' results mean the paper
prevails, the results will be very unlikely to be inconsistent, even
in cases of actual substantial malfunction or fraud.
(3) Although in Section 5(a)(1) it says audits will be "random,
unannounced, hand counts..." two paragraphs later the Holt bill
provides that the EAC shall determine the precincts "and then
*announce* the precincts in the State in which it will conduct the
audits." Sec. 5(a)(2)(A). This does not seem "unannounced" it seems
"announced".
Even if the unlikely audit-proven discrepancy can not be successfully
blamed on the paper records, the Holt bill requires the precincts or
polling locations to be "announced" "no later than 24 hours" after the
state posts final results. Now, presume that the elections officials
have engaged in election fraud, since they are the ones with the
access and if anyone in the past has been successful at hacking, they
or their friends are now the insiders. Given as little as 4 hours I'm
sure they can make paper match computers for certain precincts. This
game was allegedly played in Ohio just recently. Maybe the officials
aren't even guilty, they just are curious to see if they will pass the
audit, so they check the ballots themselves, are horrified by getting
caught seemingly with their pants down, and so they do the human thing
and pull their pants back up, even though someone else caused the
problem? Here again, paper matches result though the officials are
innocent, at least of the fraud or irregularity, though not of the
cover-up.
(Note: You can view every article as one long page if you sign up as an Advocate Member, or higher).
