Sunday, November 18, 2007

Say It Isn't So... IT'S SO! Diebold Crashes AGAIN in Official Count

At the final, official 11/6/07 election count today, Sunday, 11/18, there was no reason to believe that the CCBOE Diebold/Premier GEMS tabulator, and their other equipment and election "help" would not flunk out again.

In today's preparing for the board's Election Certification 11/21, Wednesday morning meeting, there were the same amount of memory cards that needed to be uploaded into the Diebold/Premier tabulator, as were done for the unofficial, 11/6 election night count.
The layout was the same as approved by the Diebold tech ...and as was used for all previous elections. And because all of our 11/6 votes needed to stay sealed in the 10 day waiting period between the unofficial and today's official count - nothing could be touched - no, even cursory, analysis could be done about the election night GEMS crashes; nor about the 10 highly worrisome to election security, corrupted memory cards that came in blank, but which gave no poll workers any warning of memory card problems (an updated number from former post,); nor about all the other election day and night Diebold failures,

There was no reason to believe that this 10's of millions of dollars to Diebold system, (at least $20 million to Diebold just from 11/05-11/06) - which also literally holds our very elections, and our public election data in their "confidential," "trade secret" hostage - would not just crash again, bringing "their secrets" into greater ugly light.

How shocking, especially since most of those local millions still pour from county taxpayers' coffers, though Cleveland remains one of the poorest cities in the nation with greatest public financial, survival needs - for taxpayers - not for this irresponsible, incompetent, money-hungry company; and all while our (and all electronic) election systems quite deservedly are garnering increasingly less voter "confidence," thus participation, in a time we so urgently need a reawakening to democracy's basic principles. (On 11/6 there was only about 18% voter turnout in Cuyahoga, similarly very low in most Ohio counties, and across the nation.)

Yes it was "expectable," making it ever more unfathomable -
that this company remains in business, still spreading their easy election manipulation potentials and operational failures across the country, instead of being sued, with their products recalled, so taxpayers could be recompensed our money, so we can develop something that brings accurate election results, provable to us, back into our hands, and securely. And the losses to the nation are far greater than even financial. To not be able to trust our "new, improved" election tools anymore - especially with '08 "presidential" elections AGAIN right around the corner - means cynicism, numbness, and often hopelessness in the face of power manipulation and abuse, replaces real hope for more.

And fail Diebold did - again!
And truly unbelievable, today, the CCBOE found MORE - "mysteries," and increasingly serious security and operational problems.

Also different, today however, (a Browns game day) we didn't even have the fleetingly, pretend furrowed brows of Diebold higher ups on hand, waiting to "help. " Guess they also knew what to expect....

Archie Williams, our tech help on hand was there. But even Larry Calvert, our regional tech rep who lives in nearby Canton,OH, though aware that today was CCBOE's official count day, was glaringly AWOL.

What happened today....
Basically at a final, official count, all the ballots previously counted get re-counted, plus the ballots that could not be counted on election night are included, the latter including the absentees that arrived at the BOE after 11/3 (this election, 5,000+,); provisional votes that needed to be verified after the election, but before the final count (this election, around 3,500); curbside votes; and votes from voters overseas.

Yesterday, Saturday, a full complement of workers were at the CCBOE, opening, sorting into precincts, and double counting (the ballot quantities, not the votes on) the almost 9,000 paper ballots not previously scanned, preparing them for inclusion into the Sunday count.

Yesterday late, the scanners were also again tested with a test deck that tests to make sure they can still properly read write-in votes, over- and under- votes, split precinct votes, numbers where there are multiple winners, etc.

One can't make too sure with Diebold.

After all, Cuyahoga leased enough to have 55 scanners working this election. But out of the scanners Diebold delivered, a full 19 failed their preliminary tests, (compared to what was said as 10-12 similarly failing last November) and some scanners were even sent with the wrong connector hook-ups. All those had to be put aside/packed back up, leaving us short again.

( The question of payment to Diebold for these scanners may come up again publicly with this new board, who've previously expressed huge displeasure and refusal to pay for such negligence.) Though I don't remember the exact number of scanners the CCBOE owns, thus, the number also leased, I know it was expressed that what Diebold delivered represented a 31% failure rate, (!) before the scanners were even used. ( I will seek a copy of the letter I understand was already sent by the CCBOE to Diebold, expressing that extreme displeasure.)

Just a bit about Provisionals - more later...
At shortly after 9am the official count of all paper ballots began. Our Candidate and Voter Services Department, though having finished verifying the bulk of the over 3,000 Provisional ballots over the past 10 days, was still working on some final "few" that were in question, trying (as personally observed) to let as many votes as possible be counted. All the rejected ballots had been checked at least 3 times to make sure that there was no way they could find the voter or count the vote, by law.

One example: for those they could not find in the registration database, (who would have had their vote not counted for not being registered,) they cross checked birth dates, street addresses, voter histories, etc. to make sure the names had not been "merged" with someone else's in the new statewide base - which at least for counties using the local Diebold registration system (DIMS,) now statewide uses another Diebold product (!) - DIXI.
(I have to do much more research on this statewide system... to know the prevalence of DIXI use.)

It seems that sweet-sounding DIXI, like (appropriately) dim-sounding DIMS (Diebold Information Management System,) also lets operators accidentally merge one voter's registration with another's, with no warnings, which would mistakenly wipe a person off the registration rolls/not have the vote counted, each time that happens.

At least now, with LOTS of local BOE time and diligence - only for those who are willing to pay attention and apply it - those merged names do not get trashed, but are held in "history" files, so voters can be properly reinstated. (In the relatively short time I watched the verification process, last Wednesday, I witnessed 2 such searchings and then appropriate registration reinstatements.)

So, at shortly after 9am today, about 50 workers started the official count - re-scanning the almost 24,000 previously scanned and first scanning the approximate 5000 + "new" absentee votes.

They then moved to the verified Provisionals, again around 3,000+ - a total of now around 32,000 paper ballots.

(I don't believe that any absentee votes came in from overseas (including military) in the 10 day waiting period for this rather local election, but still a fact to be checked, regarding whether proper outreach was made, or if overseas voters are just not caring about local elections any longer.)

They finished that paper ballot phase at about 2:30 pm.
The provisionals took the longest per ballot, since for every precinct, the workers must doublecheck that they have the correct number of ballots/ the tabulator is reading each precinct correctly, etc. But for provisionals there are usually only a few to zero ballots per precinct, while all the scanning double checking still must occur.

I will make a fuller report about provisional numbers later, and the still large problems they reflect/(they cause?) after the final numbers are released on Wednesday.
What I do know and can state definitively here is that once again, the two most frequent reasons for needing to reject provisional ballots, by law, are voters voting in the Wrong Precinct (though almost half the time they were in the right room (!), just the wrong table;) and voters who were not registered.

What has become known by many as the "Voter Suppression Bill" - last year's Ohio HB3 - nonsensically demands detailed ID's, and a statewide voter registration data base (see DIXI above) but still rejects the entire ballot of a person who gets to the right location, but the wrong table! Certainly there must be a way to code elections that truly are precinct-specific ( like certain zoning issues) and to let all other vote's count for all that apply for those people - even so long as a voter is in the right county, let alone, in the right room.

More Diebold Tabulator Crashes...and Worse...

After a lunch break, most workers returned for the infamous memory card re-upload. That process started out at about 3:30 pm, "rather swimmingly," actually making people question even more what happened on election night. Nothing had substantively changed - in the cards, the process, the layout. On election night, the tabulator crashed after only about 1/2 hour.

People were thinking/looking - what was different on 11/18; and what was different on 11/6 from all previous elections run on that system.
Some of the conjectures were that on election night there are usually a number of memory cards which had not been properly "ended"/closed-out at the polls, that are found by inserting them for uploading, then after rejection, sending to another tabulation station for "ending" so they can be uploaded. That was different on 11/6/07 from 11/18, but not different from previous election nights.
Another conjecture was that on 11/6 those 10 corrupted/blank memory cards were found by trying to upload them.

Another was that the sun was shining on 11/18....

But the questioning optimism began faltering a bit only about 15 minutes into the process, when many operators were noticing they were getting an "error/time out" message on their TSx screens - which seemed able to be made to go away, with no compromise to results, just by inserting the next memory card. Given Diebold's track records, this obviously could be quite worrisome.
That problem was determined, by it's approximate 5-minute intervals, to be probably caused because something called the "J-Report" (sp?) which runs in the background and updates by default every 5 minutes. The J-Report, is what puts results into reporting format, with charts, for projection onto the large screen in the CCBOE media area. It was felt that was probably a memory resource problem.

THEN another problem appeared - a memory card that had been OK on election night, was giving the error message on 11/18 that it was "bad media." The many various ways that could cause a card that had been locked up, to 11 days later, show up as corrupted could indicate a very serious system security problem, not just an operational one.
Further, that card, one from Holy Redeemer Church in Cleveland, meant that again, the CCBOE had to on Monday, go out and retrieve all the approximate 12 machines from the warehouse or if not back yet, from location, re-set them up, match the corrupted card to the right machine, and again, burn another memory card to retrieve all those votes - just as they had to do for 10 cards/10 locations, until 11:30 am the day after the election to report those unofficial results.

(I checked. When they burned those new memory cards they did label them as recreated cards, and kept them with the original corrupted ones.)

Shortly after - but presently certainly not indicating definite cause and effect - and almost an hour into the uploading, the message that previously indicated the tabulator was about to crash appeared on the server.

I didn't quite catch the error message, something I know I can find out, but it was something like 00008.... a memory error...something like the "memory that the system was trying to reference could not be found!"

GEMS then crashed at about 4:25.
This time, unlike election night, however, it did not bring the Tsx machines down with it.

After the crash and reboot, MORE problems showed up.

Three operators then noticed that when they attempted to upload cards they had not uploaded
( they have a system to keep track of what's done,) they got a message saying those cards had already been uploaded.
- At times those precincts DID show up in the tabulator.
- BUT when they checked the paper trail running on each machine, those precincts were NOT represented there!!!

Big and various errors, with the paper trails not even matching up....!!!

They stopped the entire process at that point again, ran a whole report, and checked it against the unofficial results.

And as they were finishing all 5,793 cards, they found that some of the cards that were in process while the tabulator was crashing, were showing in the end, as "missing precincts." Again, there were no pre-warnings. They had to go back and re-upload those again.

At about 6 pm, when no one was uploading anything, because they were finishing, and some workers were signing out...ANOTHER GEMS crash...

At about 6pm, with just the approximate 40 "missing precincts" left to do; about 20 more provisionals that had been in question and the CVS department had found they could verify, left to scan; and with the machines that needed to be brought back from Holy Redeemer to burn a replacement memory card left to do - the Diebold GEMS server crashed again.

The "perfect" end to the Diebold day.

Is this what you want tabulating our votes???

In the end, I saw that the CCBOE did everything they could possibly do - for hours - to make sure that the vote counts were accurate while being limited by these Diebold major failures. But is this what you want tabulating OUR votes???

I understand that Diebold DOES now want to come in after all recounts are done, to do their own analysis.
I'm sure they do...! .... to make up more of the same years long stories about what the CCBOE did "wrong" with their "fabulous" equipment.

And I think that the CCBOE needs to charge Diebold - Diebold-like exorbitant, rental fees on our machines - for doing their R&D, which should have been done years ago, before they paid their "certifying" labs for "testing", and before they sold their election-manipulatable stuff to anyone, including us.

The CCBOE needs to do their own independent evaluation, possibly now scheduled with the Secretary of State's testing. (Chris Nance, Assistant Secretary of State was there yesterday, witnessing the whole debacle, as you'll see in the video below, as soon as I can get the Google video uploader working again.) Maybe to get some


Diebold is no "Premier" company - far from it! We need to get our money back, trash these Diebold products, and move on...and quickly!

Hand counted paper ballots, that are humanly marked and readable and verifiable - for OUR elections, including for President - is the very best answer.

I would accept at this point, precinct based scanners - preferably from a company with pristine accountability, and preferably with open-source software (that which is not secret and can be checked by many computer experts, that all it's essentially doing is counting, not changing the voter's will.) - absolutely NOT from Diebold.

That way, everyone marks a paper ballot even at the polls, the ballots can be scanned at the polls to check for under and over-votes, and to create a preliminary count which would later be audited by hand; the ballots drop into separate, locked box; the precinct scanner results can be posted for viewing at every precinct on election night, so we can follow our results from the polls, to check for reasonableness against final totals; and with a truly very high speed scanner (thousands of ballots per hour) that also photos every ballot, the CCBOE can do it's pre-audited central election night count.

Any other conclusion - including that taxpayers would again have to pay for some other even good or another debacle decision that that may come from the Secretary of State's so far too closely held testing from the public - only including election directors, vendors (like Diebold) and testers some of whom are of unclear origin, but not regular citizens - would be completely insufficient for OUR elections.

Below is Official Count video, taken when I was NOT in the tabulator room as an official observer, but from outside the glass tabulator room windows - from the media area.

No comments: