Big Presidential Vote Count (Ballot Definition File) Error Found and Fixed in New Mexico

[Wilms]

Big Presidential Vote Count Error Found and Fixed in New Mexico

A test in Santa Fe County finds and fixes an error that could have cost Democrats thousands of votes.

By Steven Rosenfeld . Posted October 4, 2008.

An electronic voting machine test in Santa Fe County, New Mexico, on Friday revealed a programming error that, had it not been caught and corrected before the start of early voting next week, would not have counted hundreds -- or possibly thousands -- of votes for president and U.S. Senate in this Democratic stronghold.

The software error concerned straight party voting, where voters fill in one oval on their paper ballot that indicates they want to vote for all the candidates from a political party. The test revealed that the precinct optical-scanner computers, which read hand-marked paper ballots and compile the precinct vote totals, were not counting "straight party" votes for president and U.S. Senate.

"It was a simple error," said Rick Padilla, a senior system supervisor for the Santa Fe County Clerk office, which runs county elections. "When they did the programming, they didn't link the oval to the (presidential and senatorial votes on the) straight party ticket."

snip

What is unsettling about the test in Santa Fe County on Friday was the fact that the error affected the two most hotly contested races on the ballot -- president and U.S. senator. A more likely programming error would have have either affected all the party's candidates globally or a single race.

snip

Thus, hundreds if not thousands of potential presidential votes -- most for Democrats -- could have been lost had county officials not discovered the software error in testing.

A ballot definition file error could cause miscounts or lost votes on any electronic voting system. The only check on the accuracy of these files occurs during pre-election testing.

The M100 tabulator is used in numerous swing states such as Montana, Iowa and Indiana, according to VerifiedVoting.org, a nonpartisan group that tracks electronic voting issues.

"The main thing is this is a recoverable error," said Pam Smith, president of the Verified Voting Foundation. "In New Mexico they have paper ballots. They can recount them if you need to. New Mexico has a (vote count) audit provision (in state law). In another state, if this happens, you could miss a ballot definition file error."

snip

http://www.alternet.org/story/101567?page=entire



Talk about being "Fooled Again". Starting years ago, efforts have been made to point out this vulnerability. Unfortunately, none of them included a big name so there probably wasn't much attention paid.


http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=402806

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=438488

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=401378

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=348683

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=342985

[msongs]

1. democrats need to KILL electronic voting once and for all nt

Msongs
www.msongs.com/political-shirts.htm

[quiet.american]

2. K&R nt

[liberalmuse]

3. Why is it always Democrat votes that get dumped?

This pisses me off. We need to dump those goddamned machines in the ocean.

[Jackpine Radical]

5. Why? Divine intervention, of course.

Do you need any further proof that God is a Republican?

[WillYourVoteBCounted]

23. because they promote straight ticket voting - big time

we have the problem in NC where straight ticket does not count for President.
So many people don't read the extremely long ballot instructions on all of the exceptions for straight ticket voting. Hence our high undervote for president.

[cui bono]

4. We had TWO YEARS to fix this mess. And what did we do?

NOTHING!!! We need new Dem leadership.

:grr:

[garybeck]

7. WE?

the democratic party did nothing

WE did everything we could.

[cui bono]

14. Yeah, that's what I meant.

[diva77]

30. this mess IS a FIX, and everytime we try to repair it, those with access will FIX things again

[Festivito]

6. Those ovals have to go.

All they need do is misalign the paper or the reader in certain areas and no one will catch it.

Hand counted paper ballots.
Certainly recounts of all electronically counted ballots.

[GuvWurld]

8. Repeat after me...

federal election results are unprovable and the media should not report them as fact.

the media should be the greatest advocates of hand counting paper ballots because this method of counting allows media greatest access to observing and documenting the process, affording their reporting the greatest credibility.

K&R

[garybeck]

20. hey bro i got a question

is this true about CA audit:

Sampling Method:
One percent of randomly selected precincts or one precinct in each county, whichever is greater.


I had thought it was just the flat 1%.

[GuvWurld]

25. I'm not certain, but

I think neither of those is quite right. There is something about every race having some amount of audit. Not all races are on all ballots. So in Humboldt, a small county, 1% of precincts is then surpassed by counting other places that allow for all races to have some audit. I don't have the relevant election code at my fingertips so don't quote me.

[diva77]

29. here's what's in the California elections code regarding audits

CALIFORNIA CODES
ELECTIONS CODE
SECTION 15360




15360. (a) During the official canvass of every election in which a
voting system is used, the official conducting the election shall
conduct a public manual tally of the ballots tabulated by those
devices, including vote by mail voters' ballots, cast in 1 percent of
the precincts chosen at random by the elections official. If 1
percent of the precincts is less than one whole precinct, the tally
shall be conducted in one precinct chosen at random by the elections
official.
In addition to the 1 percent manual tally, the elections official
shall, for each race not included in the initial group of precincts,
count one additional precinct. The manual tally shall apply only to
the race not previously counted.
Additional precincts for the manual tally may be selected at the
discretion of the elections official.
(b) If vote by mail ballots are cast on a direct recording
electronic voting system at the office of an elections official or at
a satellite location of the office of an elections official pursuant
to Section 3018, the official conducting the election shall either
include those ballots in the manual tally conducted pursuant to
subdivision (a) or conduct a public manual tally of those ballots
cast on no fewer than 1 percent of all the direct recording
electronic voting machines used in that election chosen at random by
the elections official.
(c) The elections official shall use either a random number
generator or other method specified in regulations that shall be
adopted by the Secretary of State to randomly choose the initial
precincts or direct recording electronic voting machines subject to
the public manual tally.
(d) The manual tally shall be a public process, with the official
conducting the election providing at least a five-day public notice
of the time and place of the manual tally and of the time and place
of the selection of the precincts to be tallied prior to conducting
the tally and selection.
(e) The official conducting the election shall include a report on
the results of the 1 percent manual tally in the certification of
the official canvass of the vote. This report shall identify any
discrepancies between the machine count and the manual tally and a
description of how each of these discrepancies was resolved. In
resolving any discrepancy involving a vote recorded by means of a
punchcard voting system or by electronic or electromechanical vote
tabulating devices, the voter verified paper audit trail shall govern
if there is a discrepancy between it and the electronic record.

[diva77]

28. turns out the former LA County Registrar, Conny McCormack was NOT selecting precincts randomly

have to watch everything like a hawk (that is, whatever is possible to watch)

in my opinion, it is too easy to rig audits; you would have to watch auditors like a hawk and the rigging is easier for an audit than for counting all of the ballots. The sort and stack method of HCPB which incorporates a 100% audit (ie people doing the counting double check each other), and is conducted at the precinct level on election night, is by far the most transparent, cost effective, verifiable way to get the actual results.

Unless the current registrar is observed choosing random precincts, we won't know whether the choices are actually random. And then there are more ballot security breaches possible above and beyond that.

[garybeck]

9. Are there audits in NM?

This would have been caught if there are decent audits in the state.

That's why we need good audits in every state.

[OnTheOtherHand]

10. presidential on 2% of machines, I think

http://ceimn.org/state-audit-legislation-reference-guide/new_mexico

So, yes, that probably would have caught this problem. And, yes, that's why we need good audits in every state.

[garybeck]

13. Looking at their audits, I'm not sure they would catch the problem

at least not in time:

"The audit must occur within five days of the completion of the county canvass."

This is a big problem. We have the same problem here in Vermont. It's crazy to conduct an audit after the canvass report. The canvass report should include the audit results.

It's a little unclear to me exactly what point an election is "certified." But in Vermont it seems the canvass report pretty much lays it down. there are a few days after that, during which a candidate can ask for a recount, and then that's it.

So conducting an audit after the canvass report is issued, seems only to diminish the importance of the audit.

Not to mention, we've seen candidates get sworn in on election night... well before the audit or canvass report.

In this NM case, the audit would be taking place well after election night, well after the presidential election is conceded.

The point is, there need to be GOOD audits, not just any audit. A flat 2% audit taking place after the canvass report is not ideal.

[OnTheOtherHand]

15. I agree it's not ideal!

With respect to the timing of the audit vis-a-vis the "canvass," I think in New Mexico the canvass has to be completed prior to any audit or recount. It's not too late for the results to be corrected, although the way the NM audit law is written, I'm not sure the results necessarily would be corrected -- it might depend on the magnitude of the miscount. In many states, if the audit were conducted prior to the "canvass," there would be no presumably authoritative results to be checked by the audit hand counts. (ETA: I'm sleepy and expressing myself stupidly, but my point is: if the preliminary count isn't 'locked in,' there's no clear basis for comparison. That can play out in lots of different ways.)

Even a flat 2% audit should be good enough to catch this particular problem, but I agree that it isn't simply "good." There are all kinds of problems with this audit design.

[garybeck]

16. doing the audit before the canvass report

i see no reason why the canvass report has to be done before the audit. you can do the audit, and compare the results to the canvass report when it comes out. I just don't understand the logic in doing the audit after. you don't need to have someting to compare it to. Compare it to the unoffical results, until the canvass report comes out. As soon as the canvass report comes out, then you can do the comparison and decide on the trigger.

[OnTheOtherHand]

17. it's like that game show: "is that your (voting system's) final answer?"

If you do the audit before the results are frozen, and compare the audit results to the canvass results when the canvass is complete, you appear to run the risk that the canvass results will be 'conformed' to the audit results, so that miscounts identified in the audit will not be treated as such. That could gut the audit. It's quite possible that I misunderstand your meaning, but that's the issue I'm addressing.

I'm not saying that the audit always has to come after something called the "canvass"; it all depends. For instance, if ballots are coming in and/or being processed for days after the election, it should be possible to freeze the counts of the election day ballots and audit those, then audit other kinds of ballots later. Or the canvass might be, by definition, the last step before final and irrevocable certification, in which case obviously the audit would have to precede it!

Since "canvass" means different things in different jurisdictions, we could be basically talking past each other.

[garybeck]

19. certification of election

this has always been confusing to me

as the brennan center recommends, the certification of the election should be contingent on the results of the audit:

9. BINDING ON OFFICIAL RESULTS: Post-election audits must be
completed prior to finalizing official election results and must either
verify the outcome or, through a 100% recount, correct the outcome.

so how do you accomplish this? what is meant by "official" results anyway? Is it the canvass report? in my state, the canvass report sure looks like the official results. you can look at one here:

http://vermont-elections.org/elections1/2006GECanvass1113.pdf

this was done before the audit.

then I look on the "elections calendar" for that year and there is no mention of when the "official" results are announced, so I'm left to think the canvass report is more or less the "official" results.

so it becomes difficult to comply with that recommendation, unless you simply conduct the audit very quickly after the election, even on election night and ensure that the results of the audit are known before anythin "official" is announced. So if they find a discrepancy, they are prevented from putting out any kind of official report that shows numbers that could have a problem.

Maybe if the canvass report had a disclaimer at the bottom stating "these results are unofficial pending confirmation by the audit."

[OnTheOtherHand]

26. we burned hours haggling over that wording

By "finalizing official election results," we mean that all recourse to correct the winner has expired.

I don't know whether or not that is the case in Vermont at the time that the canvass report you linked to is created. In New Mexico, recounts come after the canvass, so while the canvass is in some sense official, it isn't finalized.

We agree that the audit has to come in time to correct who wins, if it needs to be corrected. It's also important for the audit not to come 'too soon.'

[Bill Bored]

22. You should "decide on the trigger" BEFORE the audit is started!

Here's the deal:

If the canvass report finds things like missing precinct tallies, or tallies that were copied incorrectly to the server or some other central reporting system, then this has to be corrected.

If the audit finds discrepancies within the precincts, this should also be corrected and the audit may have to be expanded.

HAVA requires a paper record to be used for audits/recounts. It's NOT the paper ballot or VVPAT. It's the "poll tape" or results report. If the audit compares hand counted tallies to the tallies on those tapes, it can be done as soon as the tapes are available -- even on election night if warm bodies are available. If the poll tapes and the hand counts agree, there is no trigger or escalation of the audit, but the election-night poll-tape tallies must still be compared to the officially reported results for ALL the precincts. That's the canvass (or recanvass).

I think as long as the audit compares hand count tallies to poll tape tallies, it doesn't matter when it takes place, assuming the chain of custody is secure.

Similarly, as long as the central tallies are compared to the poll tape tallies, it doesn't matter when that canvass takes place.

I think the order of preference or control for these tallies should always be:
1. Hand count
2. Poll tape tally
3. Central tally

Any discrepancies between 2 and 3 should use 2.
Any discrepancies between 2 and 1 should use 1.
Any discrepancies between 3 and 1 should use 1 (this can happen if the audit corrects a poll tape tally).

Now, it would be preferable if CERTIFICATION did not take place until after both the audit AND the canvass were completed, but that's another issue and it also depends on when candidates can ask for full recounts if the audit doesn't trigger them.

How does that sound?

[garybeck]

24. That's exactly what I'm saying

this is how it worked in 2006 in vermont:

(see http://vermont-elections.org/elections1/2006_election_info.html)


11/7 Election night - precicts announce their poll tape results publicly (to media)
11/13 Canvass Report issued from SoS
11/14 Audit took place
(I am not sure what date the SoS issued results of the audit)
11/13-15 Date files were last saved that currently show official results on SoS website
11/17 Last day for a candidate to ask for recount
12/1 Date of Official results currently showing on website (actual file was last edited 11/29)

so losing candidate had approx. 1 day to consider audit results and decide what to do (with no automatic triggers).

and again I ask the question - when here are results "certified?" Is it when the official results are posted on the website? It seems we don't use the word "certified" in our state.



Now consider that this year, the audit will take place "no later than 30 days after the election".

(see: http://vermont-elections.org/elections1/2008ProcRandomPostElectionAudit.pdf)

Let's just say for the sake of argument that it takes place 29 days after the election. If we look at 2006 as a model, the official results should be posted by then.

which is why I believe the audit should take place as soon as possible after the election. I say, do it on election night, or the next day. Our state is so small we're only talking about 4 precincts.

this is just one problem we're facing here. we also have no trigger. even if the audit turns up large discrepancies, there is no assurance anything will be done. not to mention we have no guarantee that the presidential race will even be audited. on election night the SoS will decide which 2 out of 8 state-wide races will be audited.

[Bill Bored]

32. I agree. Consider changing the recount request deadline to a later one.

Also, it may still be possible to get a court to order a recount using the evidence obtained from the audit.

However, all this recount stuff is really a reaction to FL 2000 where it would have changed the outcome of a very important election.

Prior to that, at least in some states, recounts were considered to be the least secure counts, due to increased opportunity for tampering after the polls close and the ballots are no longer under observation. It's only because of the unreliability of software (and current hardware) used in today's voting systems, and the trauma of FL 2000 and a few other cases, that recounts are all the rage in the election integrity community.

Ideally, we should have voting systems that provide reliable, accurate counts on election night. But that seems to be a thing of the past unless you vote on lever machines or count by hand. The problem is the laws haven't kept up with the technology. So we are stuck with having to use software and find ever more clever ways to verify that it behaved as it was supposed to on election night. While this is intellectually challenging and is a good way for some to get grant money, in practice it's a pain in the ass, and no one is really serious about doing it.

[garybeck]

33. i disagree with you and agree with the brennan report

it's not so much a question of moving the date of the recount request deadline

according to my read of the brennan recommendations which are my guide, the posting of any official results has to be contingent on the audit being completed and the discrepancies being within the acceptable range. if not in the range, a pre-determined protocol is followed, for example counting a larger sample.

so the candidate requesting a recount is unrelated. we are trying to create a system that will catch fraud or errors, and not put it in the hands of a candidate to decide what to do.

[Wilms]

27. And what about chain of custody issues?

Someone had to ask.

:yoiks:

[Bill Bored]

31. What about 'em?

Either do the audit on election night (warm bodies needed), or chain the custody up as best you can.

I could make the argument that it's hard to get lots of ballots to exactly match lots of precinct counts, esp. if multiple contests are audited and there's some accounting of the ballots, including the unused ones (like counting unused cash that's lying around). But not everyone buys that argument because supposedly it's still possible to get everything to match in time for the audit. I'm not so sure, but I would certainly say the audit should start as soon as possible after the election even if this means the statistical accuracy might be a bit less than the target because some election data is unavailable.

[garybeck]

34. well since you ask

from the SoS office, responding to my question:

"There will be at least seven days and probably more between the time polling places are selected for audit and the time the audits will be conducted. Vermont law requires that ballots are secured in ballot bags with numbered seals and placed in town vaults on election night. The bags will remain there until the morning of any audit, then transported by two election officials to the place of audit."

I DON'T UNDERSTAND WHY WE HAVE TO WAIT 7 DAYS.

[Wilms]

35. The "probably more" in her response suggests she has some reason.

Not that we'd necessarily agree, but I'd like to know what it is.

Have you asked?

[garybeck]

36. it's worth asking

i see no logical reason why the ballots have to sit there for 7 days or more

however they are tired of all my questions and have resorted to ignoring me.

and an article i just wrote is about to come out, which condemns them pretty badly.

but i'll give it a shot.

[Bill Bored]

21. The link to the text of the law is broken. I'd like to see it before passing judgement. nt

[FogerRox]

11. Psssst, these maqchines dont count very well. The evidence is overwhelming

Study after study shows this. Minimum solution is VVPB's that are audited, Maximum solution is to offload them into the ocean to act as reefs for fish.

[wildflowergardener]

12. blinking

Sorry to be a bother, but is there any way to stop that one line from blinking? It makes it hard to read - it won't stop blinking long enough for me to read it. I just thought you might like to know if you are trying to highlight it so people can read it - it actually makes it harder to do.

Or maybe it's too late to edit. I probably can read it a couple of words at a time.

Editing: Never mind I went to the original article to read it.

Meg

[bleever]

18. K&R.