On Alternative Counting Methods
| By Julia Rosen, Online Political Director - Feb 27th, 2008 at 5:22 pm PST |
| Also listed in: Election Protection California |
Our friend Brad Friedman has advanced a potential method to count more of the "double bubble trouble" ballots. This is a fairly complicated issue, but let me give you the overview. The ballot design was so terrible that there was an overlap between the American Independent Party candidates and the Democratic candidates on the ballot. In precincts where there were there no voters who requested an AI ballot, all of the votes can be counted. In the others, those votes cast for candidates in the 8-10 bubble slots will not be counted because there is no way to tell if a person intended to vote for an AI candidate or a Democrat.
Friedman came up with a potential method to try and count more votes:
We had our lawyers take a look at it and here is what they said:
Given the above, we are not pursuing what Brad is advocating, but in the interest of a full and open discussion we wanted to share this with our members.
Friedman came up with a potential method to try and count more votes:
Rather, as we suggested previously, due to the small number of NP-AI votes cast overall in the election, nearly every unbubbled NP ballot with a Presidential selection on it can be counted accurately if all such ballots are first counted as Democratic votes, with the number of unaccounted for NP-AI votes in any precinct then being subtracted from each candidate's total in the precinct.
We had our lawyers take a look at it and here is what they said:
We have quickly reviewed Mr. Friedman's suggestion and, actually, earlier considered, then rejected, a variation of his proposal. We believe the approach suggested by Friedman, is flawed. First, as opposed to the method now being used to count nonpartisan votes, which requires an examination of voter intent that is clearly and unambiguously demonstrable on each ballot, the alternative proposed does not require such a ballot-by-ballot examination. Instead, assumptions about how a particular ballot was cast and how groups of ballots were cast are necessary. And while the small numbers of votes cast by American Independent voters means that most of the crossover nonpartisan ballots were cast for Democratic candidates, the problem is that with respect to particular ballots, the assumption is not conclusive. Under the law, election officials may not make those assumptions. To engage in the vote subtraction method poses additional problems in that if a similar process is not utilized for AI voters, equal protection constitutional concerns also surface. Even if the method were applied to both parties, it would necessarily result in more votes cast than voters, thereby creating additional legal problems and liability. Ultimately, though, any of the variants of Friedman's suggestions, create serious legal problems with respect to ascertaining voter intent.
More importantly, by straying from election law jurisprudence related to ascertaining voter intent in the manner suggested by Friedman, dangerous precedent is created - legal and otherwise - for situations that may arise in the future. If election officials are given the freedom to make such assumptions about how particular ballots were cast, the abandonment of standards for ascertaining voter intent opens the door the potential for great mischief or intentional manipulation of election results.
While unfortunately, the ballot design flaw will undoubtedly result in some crossover votes where voter intent is not ascertainable, it appears that the vote counting procedure now underway will result in a significant number of votes being counted where voter intent is unambiguous and not founded on assumptions on how a voter cast a particular ballot.
Given the above, we are not pursuing what Brad is advocating, but in the interest of a full and open discussion we wanted to share this with our members.
Comments are closed for this post.
For instance, they say that in the Registrar's plan "voter intent ... is clearly and unambiguously demonstrable on each ballot". I would say that while this may be true for ballot positions 11 through 15, it is true for ballot positions 8 through 10 IF AND ONLY IF the poll workers recorded the non-partisans' intent to cross-over in the Roster with 100% accuracy.
What if a precinct worker failed to record that a NP voter intended to cross-over to the AI primary, or mistakenly recorded his/her cross-over choice as Dem instead of AI (and then failed to mark bubble #5 or 6)? If he/she were the only AI cross-over in that precinct, then under the Registrar's plan, his/her vote is going to be recorded as a Dem vote.
Any time you use something other than the ballot itself - any time - to infer voter intent, you are making "assumptions about how a particular ballot was cast and how groups of ballots were cast". The assumption with the current plan is that the pollworkers kept the Rosters perfectly. And as a pollworker, I submit that in the vast, vast majority of precincts, that is not a sound assumption.
The ballot and ONLY the ballot should be used in determining voter intent.
But hey, that's just me being churlish again.
(Semantic quibble from a logician: "...the problem is that with respect to particular ballots, the assumption is not conclusive."
Can an assumption *ever* be conclusive?)
"We have quickly reviewed Mr. Friedman's suggestion and, actually, earlier considered, then rejected, a variation of his proposal. We believe the approach suggested by Friedman, is flawed. First, as opposed to the method now being used to count nonpartisan votes, which requires an examination of voter intent that is clearly and unambiguously demonstrable on each ballot, the alternative proposed does not require such a ballot-by-ballot examination."
Not sure which method he's referring to as requiring a "ballot-by-ballot examination", but my proposed method requires no different form of examination than the one Logan is currently proposing.
No difference.
So with that, and a few other apparent misunderstandings about my suggestion, I'm concerned the "quick review" of it by Mr. Reyes may have been a little bit too quick. I'm happy to answer his questions about any of it, of course, at ###-###-####
As well, you do know that Logan, as I understand it, is preparing to program the machines to count the ballots, as opposed to a "ballot-by-ballot" examination, right? Either way, whether it's counted by machine or hand, my suggested method is no different from determining votes in Logan's suggested method.
"Instead, assumptions about how a particular ballot was cast and how groups of ballots were cast are necessary."
Not in my suggested method. No assumptions are made, and every vote counted is counted 100% accurately (presuming accuracy of the method used, machine or hand, etc.)
"And while the small numbers of votes cast by American Independent voters means that most of the crossover nonpartisan ballots were cast for Democratic candidates, the problem is that with respect to particular ballots, the assumption is not conclusive."
Yes. It is.
"Under the law, election officials may not make those assumptions."
None are necessary in my method. None.
"To engage in the vote subtraction method poses additional problems in that if a similar process is not utilized for AI voters, equal protection constitutional concerns also surface."
Equal protection does not come into play here, and I'll explain why. But before I do, let me note that EP *already* comes into play with Logan's method. And much more so. He is suggesting tossing out thousands of votes simply because they happened to be for candidates unlucky enough to have been placed in bubble 8 to 10. They will have their votes not counted. While candidates lucky enough to be in the 11 to 15 slot, in the same precinct, none the less, will have their votes counted.
Where is the equal protection for the voter unlucky enough to have selected a candidate who happens to be on a particular line. I fail to see how the EP argument could be made at all, when *supporting* Logan's proposed method.
As to equal protection for the AI voters, the law in CA is clear, as both Bowen and Logan pointed out in their various letters, that, according to Election Code, section 15154(c), "If for any reason the choice of the voter is impossible to determine, the vote for that office shall not be counted."
Indeed, when it comes to AI voters, I can see no way to count their ballots accurately, to ensure voter intent, if there are more unaccounted for (unbubbled) NP-Dem ballots than NP-AI ballots in a precinct.
In those cases, the NP-AI votes will not be counted, because they cannot be counted. According to state law.
That same law applies equally to NP-Dem voters.
Fortunately, there *is* a way to count those votes, at least, as the CA law and Constitution require. As the Courage Campaign helpfully pointed out, the Election Code says it "shall be liberally construed so that the real will of the electors will not be defeated by any informality," and the Constitution states clearly: "A voter who casts a vote in an election in accordance with the laws of this State shall have that vote counted."
Under those two provisions alone, ALL of the NP-Dem votes that can be counted, must be counted. And virtually all of them *can* be counted.
"Even if the method were applied to both parties, it would necessarily result in more votes cast than voters, thereby creating additional legal problems and liability."
I don't have any idea where you come up with the assertion that it would "result in more votes cast than voters". So again, it seems that Mr. Reyes may have a *very* different read on my proposed method.
As I don't wish for him to feel like he has to defend his comments -- truly, I'm trying to figure out how we can work *together* here -- I'll happily take the blame for a poorly written explanation in any of my articles that covered it. It's a complicated matter, as you know, though the solution is simple. But explaining the complicated issues and the simple solution is not easy. I've done my best, but Mr. Reyes comments would suggest that perhaps I failed in that regard.
So, again, I offer my phone to any of you guys, at any time to clear any of this up.
"Ultimately, though, any of the variants of Friedman's suggestions, create serious legal problems with respect to ascertaining voter intent."
I fail to see it. The voter intent would be 100% accurately determined and recorded. If it could not be, I would not be calling for this method to count votes.
"More importantly, by straying from election law jurisprudence related to ascertaining voter intent in the manner suggested by Friedman, dangerous precedent is created – legal and otherwise – for situations that may arise in the future."
Again, I am *not* straying from ascertaining voter intent 100% on every single ballot counted.
I see tossing out thousands of random votes (which could be counted) simply because the voter had the bad luck to chose a candidate on the wrong line to be a *far* more disturbing precedent. I see no precedent problems with counting every vote possible, exactly to the voters intent.
"If election officials are given the freedom to make such assumptions about how particular ballots were cast, the abandonment of standards for ascertaining voter intent opens the door the potential for great mischief or intentional manipulation of election results."
There is NO "assumptions" made about anything. Period. In my method. I would certainly agree with Mr. Reyes concerns if there were, and I'd never support such a possibility.
"While unfortunately, the ballot design flaw will undoubtedly result in some crossover votes where voter intent is not ascertainable, it appears that the vote counting procedure now underway will result in a significant number of votes being counted where voter intent is unambiguous and not founded on assumptions on how a voter cast a particular ballot."
And again: My method would count ALL of the votes that can be counted, not just a "significant number" (and a number that unfairly gives some candidates an advantage over others).
And it would ONLY record votes where the voter intent is ascertainable and 100% certain.
There are NO "assumptions" being made in my method. And best of all, thousands of ballots will not be needlessly, and randomly, and illegally dumped just because it was either inconvenient for Logan to count them (it's really not), or because of a misapprehension of how almost EVERY vote given to a candidate is known, with 100% certainty to be a vote that deserved, as per the accurate intent of the voters. And such that no candidate will receive an undue advantage over any other.
It is true that Friedman's method (which I independently "invented"--indeed, anyone with a decent quantitative background would come up with the same) does not unambiguously demonstrate the voter intent *on each ballot*.
However, in our voting system it does not matter *who cast which ballot*. So a legal standard that says that you must be able to determine intent on *each* ballot instead of the intent expressed by the *collection* of ballots is a broken standard.
I agree that if this is the current legal standard, that it is probably not worth fighting to overcome this technicality which blocks entirely sensible counting measures like Friedman's. But I think that, when advocating fairness in voting, it is important to at least point out when the law is structured to be unfair.
To review, the basic idea is this:
- Count all votes assuming that they're Democratic.
- Count how many votes might have been for *non* Democrats.
- In a worst case scenario, all the other votes would have gone to one candidate.
- Assume that *both* sides had a worst-case scenario. Subtract those votes from both sides.
- Since you can't do worse than the worst case, each Democratic candidate must have gotten at *least* as many votes as those worst case estimates.
- If you are being maximally conservative regarding voter intent, this number is the maximum number of votes for each candidate that you are positive that the voters expressed.
If you're positive that the voters expressed something, it seems a little odd to throw it out on because you can't be sure who expressed what. There are no assumptions about voter intent; it's a logical impossibility that the candidates got fewer votes than this.