You can read The BRAD BLOG’s previous detailed explanations of the current Los Angeles County “Double Bubble” ballot debacle here and here.
If you’ve yet to sign the Courage Campaign’s petition demanding that these ballots be counted immediately, please do so right now!
As the clock ticks towards the date required for final certification of as many as hundreds of thousands of currently uncounted ballots in Los Angeles County — due to an idiotic, ill-conceived, and possibly illegal scheme by the county’s Registrar of Voters — it is imperative that those ballots be counted immediately.
While the acting Registrar has claimed, incorrectly, that it is “impossible” to accurately determine the intent of the voters on those ballots, he is wrong.
Almost every single uncounted ballot can be counted today with absolutely certainty that the intent of the voter is being accurately recorded.
With current the miscount/error rate of those uncounted ballots — on which the voter’s intent has currently been inaccurately recorded by the optical-scan voting machines as “no vote” — now standing at 100%, almost every single ballot counted will only lower that rate at best. At worst, in a tiny (almost infinitesimal) percentage of ballots, the current error rate will be not be lowered, though it will not be increased.
Thus, the error rate can be reduced to almost 0% if the Registrar’s office will simply begin counting the ballots — immediately.
Here’s what everybody, (including apparently the Registrar!), needs to know about the currently miscounted ballots…
The short background for the “Double Bubble” debacle
In short(ish), working from the very conservative numbers offered by the LA County Registrar’s office, at least 170,000 non-partisan (NP), also known as decline-to-state (DTS), voters showed up in the county to vote on Super Tuesday. Of those, at least 100,000 requested to vote in either the Democratic or American Independent Party open primaries as they were welcome to do.
When an NP voter asked to “cross-over” and vote in one of those primaries, the procures required that they be given a paper NP “InkaVote” ballot (a card with only a series of empty numbered bubbles on it), slip that ballot into one of the Dem or AI template booklets in one of the booths, and then select BOTH a Presidential candidate, as well as fill in a second bubble meant to specify — to the computer optical-scan counting machine — that the ballot should be tallied as either a Dem or AI vote, depending on which bubble the voter filled in.
If that second bubble wasn’t filled in, the computer didn’t bother to count the presidential selection on the ballot.
Since the great wisdom of recently-resigned Registrar Conny McCormack determined that both the Dem and AI presidential candidates would be recorded in the same line of bubbles on the same NP ballot, the second bubble had to be filled in, else there would be no way of knowing — if only looking at the ballot — which primary, Dem or AI, the NP voter was intending to vote in.
The result has been (again, using the exceedingly conservative numbers of the LA Registrar) some 50,000 NP ballots with a selection for Presidential candidate on it, but no selection for either Dem or AI in that second bubble. Therefore, the machines did not count the Presidential selection on those ballots.
For the moment then, some 50,000 voters in Los Angeles County have had their votes for Presidential candidate currently miscounted. An intended vote for Hillary Clinton, for example, has not been registered as a vote for her. She has lost that vote for the moment, and the voter has been disenfranchised. Needlessly.
Moreover, current acting Registrar Dean Logan is claiming that, due to the fact that the same sets of bubbles were used for both Dem and AI candidates, it’s “impossible” to determine with absolute certainty the intent of the voter. But he is wrong. In almost every single case.
The current miscount/error rate for those 50,000 ballots is now at 100%. Thus, any ballot counted at this point will only lower the current miscount/error rate.
Since almost every single one of those ballots can be counted accurately, as per the voter intent, beyond a shadow of a doubt, it’s an absolute absurdity and outrage that Logan is claiming that none of them can be, as he argued in an absurd report [PDF] delivered to the County’s Board of Supervisor’s on Monday.
How Do We Know Most Ballots Can be Counted With 100% Accuracy Immediately?
The precinct procedure for handling NP cross-over voters, when they came in and requested to vote in one of the open party primaries, was to cross out the NP next to their name in the poll roster at check-in, and write either DEM or AI next to the crossed out NP in the book.
In almost every single precinct there were zero NP voters who requested to vote in the AI primary.
I’ll repeat that another way: If you check the poll rosters in most precincts, you will find that every NP cross-over voter intended to vote in the Democratic primary. Therefore, any NP ballot in that precinct which has a Presidential candidate bubble filled in, can be counted immediately, with 100% certainty, that it was meant as a vote in the Democratic Primary.
For a start, those ballots should be counted immediately, in full, and tallied as part of the official Democratic Party primary count, as the clock is ticking towards the official certification date, just over two weeks from today.
What about precincts in which the poll roster shows some NP-AI crossover voters?
When the poll rosters are checked to see if anybody at the precinct was an NP voter who requested to vote in the AI primary, as mentioned, most such precincts will show zero such voters. How many NP voters do you think bothered to go out to vote on Primary Day in order to vote for a minor candidate in a minor party that almost nobody has ever heard of?
Nonetheless, there may be a few precincts where one or two, or even a handful of NP voters, might have requested to vote AI for some reason.
In those rare (perhaps minuscule) number of cases, there are essentially two choices if one wishes to comply with the California Constitution, which requires that “A voter who casts a vote in an election in accordance with the laws of this State shall have that vote counted,” and the California Elections Code which requires that code requirements “shall be liberally construed so that the real will of the electors will not be defeated by any informality or failure to comply with all of the provisions of the law.”
Remember, filling in that second bubble is not part of either the law or the elections code, and the requirement to fill it in may even be against the law.
So the two options, in such rare cases, where there happens to be an NP-AI voter or two at a precinct, are as follows:
OPTION 1) Count all NP ballots at the precinct as Democratic votes. Yes, that would disenfranchise a few AI voters, which is not good. Nonetheless, those voters are disenfranchised already if the Registrar continues to refuse to count any of the ballots, as he is currently doing. And, again, counting any ballot at this time would most likely reduce the current 100% error rate that exists now for the currently uncounted ballots.
OPTION 2) Count the ballots as Democratic votes and reduce the count by a proportional number. Though this option requires a bit of fairly simple math to account for a tiny number of AI voters, the county routinely uses such fractional extrapolations in other areas of vote counting. For example, just 1% of ballots are counted by hand for the state-mandated post-election “audit” of ballots, meant to check for accuracy of machine counting. As well, the acknowledged error rate on the optical-scan counters is already some 1% or more. Any errors factored in to the count of NP-DEM ballots, where a small fraction would be reduced proportionately to the tiny number of AI voters, at a rate commensurate with the overall results for each Democratic candidate in that precinct, would still likely be far smaller than the “acceptable” error rate of the county’s computerized optical-scan counting machines.
Either option, of course, would reduce the current 100% error rate, to nearly 0%.
[UPDATE 2/16/08: A reader writes in to offer yet one more option as to how to count these ballots in such a way that would ensure that no votes are given to any candidate inaccurately:
Yup. Sounds like that would work! A handful of potential votes for Dem candidates might still go uncounted using this method, but every vote counted would then be ensured as a vote for a Dem candidate.]
As to other elements that can be factored into the already existing error rates, we’ll note that many poll workers, thankfully, misunderstood the Registrar’s ridiculous scheme and actually handed Dem ballots, instead of NP ballots, to NP voters who asked to vote in the Dem primary. Those ballots, even though they were “erroneously” given to the voters, have already been counted as Democratic Primary votes.
So as it stands now, many NP-Dem cross-over voters — perhaps hundreds of thousands, since LA County has some 761,000 NP voters registered — have already had their votes counted “more equally” than all of the others, simply because they were handed Dem ballots, instead of NP ballots.
(Please see the addendum now posted at the end of this article for further details on this, and how the off ice of the Registrar — who has had the temerity to blame poll workers for this mess! — issued contradictory information to poll workers about how to handle NP cross-over voter ballots.)
To Review: This is What Must be Done Immediately to Avoid Disenfranchising at least 50,000 LA County Voters
In precincts where the poll roster shows no NP to AI crossover voters, count ALL NP ballots with votes for President on them immediately. There is no legitimate justification for doing anything other than that.
If there are any precincts with NP-AI cross-over voters noted in the poll roster, the NP ballots with presidential votes on them can either be counted in full as Dem ballots immediately, or (the more conservative option) they can all be counted as Dem ballots, and then reduced by a proportional fraction to account for however many NP-AI voters there may have been signed into the precinct, as noted in the poll roster.
To reiterate: We currently have, conservatively, some 50,000 voters who have been disenfranchised. 100% of their ballots have been miscounted. Any counting of those ballots — even if an infinitesimal number of them are counted contrary to the voters intent — will only decrease the existing miscounted/error rate for those ballots which now stands at 100% until the acting Registrar of Voters directs that those ballots be counted, as is his legal requirement.
While the clock ticks towards certification day, any delay continues to risk the permanent and inexcusable disenfranchisement of thousands and thousands of voters. That prospect which is wholly unacceptable on any level.
The excuses must stop. Dean Logan must get to work and start counting. NOW.
Any questions?
ADDENDUM: As mentioned above, at some precincts poll workers gave Dem ballots, instead of NP ballots, to NP crossover voters. Those Dem ballots have been counted in the Democratic Primary tally, despite the fact that the procedure was supposed to be for poll workers to give NP voters an NP ballot and direct them to vote on it in the Dem InkaVote booth.
Despite the Registrar’s claims to the contrary, the instructions given to poll workers were both confusing and incomplete, and, as we reported in an earlier story on this, did not include clear instructions to direct NP-Dem voters to fill in that second bubble on the NP ballot.
Worse still, as the email just sent from a precinct inspector where poll workers gave out Dem ballots instead of NP ballots details below, the Registrar’s office issued contradictory instructions on which type of ballot should be given to NP cross-over voters. All of which makes the suggestions, made by acting Registrar Dean Logan, that poll workers and/or voters were to blame for this mess — instead of his office, where the blame squarely belongs — all the more reprehensible.
I’m the inspector that wrote that I used the wrong method for the nonpartisan voters [by giving them Dem ballots, instead of NP ballots]; I just found out that the written instructions given to us were contradictory. One of the clerks that worked with me just dropped off the “Election Guide and Checklist” that was issued to all poll workers at the workshops provided by the County. On page 21 it states:
“Ask all nonpartisan voters if they wish to vote for candidates from either the American Independent or Democratic parties and, if so, cross out the ‘NP’, in the party column, and enter voter’s party preference.”
The very next step listed is:
“Announce the party listed for the voter so that the Ballot Clerk can hear and select the correct ballot for the voter.”
That means, the County did not intend for the NP voters to receive nonpartisan ballots to take to the Dem or AI booths; but to use either the Dem or AI ballots in the Dem or AI booths. That is the method my poll used.
But on page 23 it states:
“If a nonpartisan voter is voting for American Independent or Democrat, direct them to take their NP ballot to either the AI or Dem party booth.”
Did the County want us to give the NP voter the ballot that corresponded to the choice they made (NP or AI or Dem) or were they supposed to receive only the NP ballot? The instructions contradict themselves.
I voted in the primary election in Los Angeles. At the top of my ballot was the word:
DEMOCRATIC
Below was a very small box with one bubble and words similar to “Check this box only if you are NOT Democratic” so I didn’t check the box as I am currently registered as Democratic.
I have been carefully following this story since the moment it hit the news — which was before the election. So I was actively looking for the box on my ballot.
I STILL have NO idea if I should have checked the box or not because I still don’t understand why the box with the bubble was there.
SD #1
Now you know that the election religion, where we have FAITH in all the election officials do, is a false religion.
Oh, it is a religious faith alright, it is just that it is the religion of despots throughout history. Have faith in us is the phrase of those who want to enslave you into serving them. And they always become harsh taskmasters.
Our forefathers bequeathed to us a framework whereby, if we keep diligent and DO NOT HAVE FAITH in government, but hold them to the supreme law of the land – the constitution, then we will grow into the ultimate democracy. They work for the good of all of us … or else.
If not, we will devolve into rank demockcrazy, where they think they can feed us any old bullshit any old time.
Am I correct in saying that the three AI candidates correspond to votes for Dodd, Gravel, and Edwards? Or are the positions different on the real ballots?
awesome reporting, brad. keep up the good work.
Brad, if you want all the double bubble’s popped, if you want fiasco ballots, if you want voters disenfranchised just call on me.
BVAC asked:
There were 3 AI candidates and 8 (as I recall, I’d have to go check) Dem candidates. As to your question though, I believe that the ballots was ballot rotation across the county. So, no, the bubbles that crossed over between Dem and AI candidates would NOT have always been Dodd, Gravel and Edwards. Unfortunately.
since LA County has never divulged what their chain of custody is for elections, I have little confidence in the results whether or not the ballots get recounted. Yes they should recount, but they should let voters know that the path the ballot takes from precinct to final result is anything but transparent
Unless the state intervenes, it seems the purpose of delay is to let the clock run out.
Does California have an AG willing to uphold the law on the behalf of the voters?
Badger #8
I am amazed that AG Jerry Brown has not put the hammer down.
If they follow your instructions using precision counting they will still have an error rate but it could well be the lowest error rate of any county in the nation. Their other option is to not count and have the higest error rate in the nation. Which route will they feel most comfortable with??? 😉
As one of the King County Washington voters who had to endure Registrar Dean Logan’s nonsense, I have to say that the man is incompetent, however he is canny in pushing the buttons of elected officials. If you want to know what he pulled in King County, read the Seattle Times, Seattle P-I, and, I think, The Stranger. It was never his fault, always someone else’s, preferable a woman who worked in a subordinate position. I suggest that everyone involved keep hammering on him, and Dems and Repubs alike should sue.
HI Brad,
Heard you on Air America last night. Did you say that along with the bubble problem there is also a big chunk of absentee ballots that have not been counted too? Most of them delivered to the polls by voters on election day? (one is mine and one my husbands).
I heard that 49% of the cast were from absentee ballots this year. Many of my neighbors and I have been voting absentee for years because we have been told it was safer- paper trail- and also because of the inaccesability of polling places in the canyons. This is one of the first years that we have had a local place to vote. One of the reasons I went ahead and walked them in.
Also the week before the primary Randy Rhodes was telling people to walk in ballots. A lot of Edwards supports were holding back and and I don’t know how many followed her instructions.
How do we make sure those votes are counted??
Who do we call? Or light a fire under?
BVAC asked:
Am I correct in saying that the three AI candidates correspond to votes for Dodd, Gravel, and Edwards? Or are the positions different on the real ballots?
Brad Friedman replied:
There were 3 AI candidates and 8 (as I recall, I’d have to go check) Dem candidates. As to your question though, I believe that the ballots was ballot rotation across the county. So, no, the bubbles that crossed over between Dem and AI candidates would NOT have always been Dodd, Gravel and Edwards. Unfortunately.
Hardy asks:
If different precincts use ballots with the candidates listed in a different order, the counting machine must know how to count the different orders. How? (Why didn’t Logan show the different orders of candidates in his attachments to his report?) It should still be possible to count ballots that have a mark in one of the five positions that do not overlap and assign those votes. There could be NO question on the intent on those ballots. The only question would be where the three overlap. Or Logan could use one of Brad’s suggestions: count all the ballots as Democratic since hardly anyone (maybe zero) declared themselves as Independent Party.
The position of candidates names are indeed rotated from location to location. per state law. The reason is that people will tend to vote for the first name on a ballot. (This was a big issue during the recall election because since there were so very many names on the ballot, some were going to never make it to that first position, etc.) Where I voted in Torrance, the Dems in positions 8, 9, and 10 were Joe Biden, Barack Obama and Bill Richardson.
“…Or Logan could use one of Brad’s suggestions: count all the ballots as Democratic since hardly anyone (maybe zero) declared themselves as Independent Party.”
In the 1% audit that LA County Dean Logan has already done, and reported on, there were validly marked non-partisan ‘cross-over’ ballots for AI candidates in 25 of the 48 precincts looked at. Certainly not “hardly anyone”.
Except that by your evidence, if only ONE person in a precinct voted AI in the precinct you make it seem like over 50% did.
The number of precincts that had an AI voter is not in dispute, it is the number of PEOPLE within that precinct that is tiny.
Appendix C of the Registrar’s report shows 50 validly marked non-partisan AI ‘cross-over’ ballots (it also, oddly, shows 59 ballots with only bubble #5 marked, but no marking of any bubble in the ‘presidential’ range of bubbles.)
Taken together, one might say that that means 109 ballots marked with at least some intent from a non-partisan to cross-over to the AI primary, out of 1887 total NP ballots found (I think; that total number in the pdf didn’t reproduce too well). That’s a little under 6%.
Now, that doesn’t mean that the idea of looking at the Roster of Voters in the precincts for notations of intent to cross-over can’t work. But it does mean that a corresponding number of ‘AI’ notations must be found. And that seems, IMHO, to make the whole idea less likely to work (given that the poll workers, myself included, didn’t perform other tasks very well or accurately that day).
Here’s a link with the strength of the AIP in Cali
Increase in membership is 8000 in the past year.
LA County is not among their top 3. They have only two counties at over 4% registration.
How many voters should be disenfranchised in order to satisy a party with less than 330,000 voters in the state?
Topanga Lib #12 –
You should only ever vote by absentee under two conditions:
1) You are actually out of town on Election Day, and/or can’t actually show up to the polls to vote.
2) The only option you are given to vote on Election Day is with a DRE (touch-screen) voting machine.
Since ALL California voters are entitled to vote on a paper ballot, at the precinct, on Election Day, there is no reason, short of item #1 above to vote Absentee. It’s a terrible idea, for a number of reasons that I won’t detail here for the moment.
A paper absentee ballot is likely counted by the same, unreliable type of optical-scan machines that count paper ballots cast on Election Day. The difference is that there is much more tracking for it, to help assure that it actually gets counted, as opposed to disappearing in the mail, at the Election office or anywhere else along the line.
The counting of Election Day ballots is also generally done with many folks from the public watching (which you are also welcome to do!) Whereas Absentees are sometimes counted in the days leading up to the election, or during the day on Election Day when few are watching, or many days after Election Day when ever fewer are likely watching.
If you retained your stub from your Absentee ballot, you may be able to call in and check with your local county registrar’s office to see if they have a tracking system to let you know if it was received and/or counted.
Whether it was counted accurately is another matter entirely, and given the systems that most counties use (op-scan) it’s frequently impossible to determine if the ballots were counted accurately at all. Unless you’ve got thousands of dollars to invest in a requested recount.
Robert Earle #16 said:
The reason he would know that “25 out of 48 precincts” tested had such NP-AI cross-overs is because the ballots had to have the bubble checked for AI.
When comparing those precincts to the poll books, how many more NP-AI cross-over voters were there?
That’s what Debra Bowen, who has now echoed many of our assertions above, has now asked Dean Logan to find out [PDF].
I have added this as a mid-article update to the original post above, but in answer to Robert Earle’s points, I’ll repost it here, as it may help answer those concerns:
[UPDATE 2/16/08: A reader writes in to offer yet one more option as to how to count these ballots in such a way that would ensure that no votes are given to any candidate inaccurately:
Yup. Sounds like that would work! A handful of potential votes for Dem candidates might still go uncounted using this method, but every vote counted would then be ensured as a vote for a Dem candidate.]
Now there is a story in the NYT that Obama did not receive one vote registered in 80 districts. Well, put that together with NH, the 100,000 independents in California, and poll workers taking ballot boxes home overnight in NM and I’m starting to wonder “rigged election?”. Where is Jimmy Carter when it comes to elections in his own country, in his own party?
Brad, re: your comment at #20
I’m not at all sure what the proposal is that you (and your reader) make at comment #20, but if it in someway intended to ‘solve’ some problem I’ve brought up relating to “validly voted NP cross-over to AI” ballots, you’ve misunderstood me.
The reason I pointed out that there are validly voted AI cross-overs was to show that the idea there were going to be ‘virtually zero’ AI cross-overs was incorrect.
But validly voted AI cross-overs is not where your problem lies.
Maybe the best way to do this is with an example:
The precinct represented on the first line of Appendix C in the Registrar’s report shows a total of 35 non-partisan ballot. Of that 35, 13 showed marks only in the column(s) corresponding to the propositions; that is, they were truly non-partisan voters who on their ballots made no indication of an intent to cross-over whatsoever.
That leaves us 22 ballots with some indication of intent to cross-over. Of those 22, 14 were validly marked; 13 as cross-overs to the Democratic primary (meaning they had bubble #6 marked and a mark in the ‘presidential’ range of bubble 8 through bubble 15), and one valid AI cross-over (meaning bubble #5 marked, and one of bubbles 8 through 10 marked).
The ballots that we’re all interested in are the remaining 8 ballots, the ones that had a mark in one of the ‘presidential range’ bubbles, but with neither #5 nor #6 marked.
Now, my understanding of what you (and Rick Jacobs at Courage Campaign) want to do is look at the Roster of Voters, hoping to find 22 notations for NP voters who informed the poll workers of their intent to cross-over. Further, you are hoping to find exactly one notation of an intent to cross-over to the AI primary.
If you find exactly that set of notations – one of ‘AI’ and 21 of ‘Dem’ – in the Roster, you can then infer that the one ‘AI’ notation corresponds to the one validly voted AI cross-over ballot, and 13 of the 21 ‘Dem’ notations correspond to the 13 validly marked Dem cross-over ballots. And lastly, that the remaining 8 ‘Dem’ notations in the Roster correspond to the remaining as-yet uncounted mis-marked ballots, because now we can be sure that they are indeed ballots intended to be votes in the Democratic primary.
So long as the notations in the Roster match up correctly, the fact that there is a validly voted AI cross-over is not a problem.
The two problems I think we ARE going to discover – if not specifically in that particular precinct, then in LOTS of other precincts – are (1) that there are fewer than 22 notations in the Roster, and /or (2) the notations that are there don’t properly ‘match up’ to the validly voted cross-over ballots found in that precinct.
If either of those two conditions are found, the inference of voter intent fails, and the mis-marked ballots become (or remain) uncountable.
The whole scheme depends on the poll workers uniformly across the county accurately marking the Roster, and voters actually marking their ballots the way they intended, and the way they told the poll workers they intended – two things that I have very little faith actually happened.
(If there’s any way to apply the ‘solution’ put forth in comment #20 to my example, I’ll certainly consider it.)
– Robert
If a NP ballot is marked once in bubbles 8 through 10, then it is either a vote for an AI or a Dem. If a NP ballot is marked once in bubbles 11 through 15, it has to be a vote for a Dem, right? Seems to me a mark in bubbles 11 through 15 shows intent and the 11-15 marked ballots can be counted WITHOUT question.
(I totally reject Dean Logan’s suggestion that a NP voter might have voted his/her ballot in a Republican booth using a Republican template. By that logic, any Republican might have used a Democratic template and vice-versa, making all the votes uncountable!)
After the 11 through 15 ballots have been counted, one of Brad’s solutions (Option 1 or Option 2 above) could be used for the 8 through 10 ballots.
If not, the bubble 8 through 10 ballots could still be read as an “either/or” vote: A mark in position 8 would indicate a vote for AI candidate “Smith” or Dem candidate “Jones”.