IS THERE A LAWYER IN THE HOUSE??? Please keep kicked!

[Bill Bored]

Are you all waiting for the next stolen election or can we talk about how to prevent it?

The following section of this new NY State Law explains how to get hand counting of VVPATs or VVPBs:

28 § 16-113. Audit of voter verifiable records. The supreme court, by a
29 justice within the judicial district, or the county court, by a county
30 judge within his or her county, by any candidate or his or her agent,
31 may direct a manual audit of the voter verifiable audit records applica-
32 ble to any candidate running for office within such judicial district or
33 county where (1) the uniform statewide standard promulgated by regu-
34 lation by the state board of elections pursuant to subdivision three of
35 section 9-211 of this chapter with respect to discrepancies between
36 manual audit tallies and voting machines or systems tallies requires a
37 further voter verifiable record audit of additional voting machines or
38 systems or all voting machines or systems applicable to such election,
39 or (2) where evidence presented to the court otherwise indicates that
40 there is a likelihood of a material discrepancy between such manual
41 audit tally and such voting machine or system tally which creates a
42 substantial possibility that the winner of the election as reflected in
43 the voting machine or system tally could change if a voter verifiable
44 record audit of additional voting machines or systems or of all voting
45 machines or systems applicable to such election were conducted.

Assume that the BoE (option 1 above) tells you to buzz off and you have to go court.

I'm interested in knowing if the following would be sufficient evidence for the losing candidate to get a court ordered manual recount to prove who won the election with mathematical certainty:

1. A manual random audit of (only) 3% of machines or op scanners in the affected jurisdiction(s) but with no discrepancies found;

2. A reported victory margin of say 5% i.e., 52-47, or any other one you'd care to use as a hypothetical;

3. Enough votes cast on each machine or scanner such that a discrepancy in a SUBSET of the remaining 97% of unaudited machines in the affected jurisdiction(s) could be enough to result in a reversal of the election outcome (feel free to speculate as to the size of the subset);

4. Undisclosed software counting the votes (but the contest of election can force its disclosure in court according to another part of the law).

I'd be interested in your opinions of this law, how it compares to other states, BoE regs, etc. that you're aware of.

Is it adequate protection, assuming the candidates are be willing to use it and that they have whatever funds might be necessary to pay for such recounts?

Is this law good enough to be used as an example for other states?

How can it be improved?

Keep in mind, the context is one in which there are VVPATs or VVPBs, no paperless e-voting, mandatory manual random auditing of 3% of DREs or Op Scanners in each jurisdiction and undisclosed vote-counting software (which could also be disclosed in court as part of the contest of election).

So would you give your right arm to have this law on the books in your state, is it junk legislation, can it be improved, or what?

Thanks in advance for your legal opinions counselors!

And please help me keep this kicked on GD since no one will visit this forum until the next stolen election.
<http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=104&topic_id=4221389&mesg_id=4221389>

[RaleighNCDUer]

1. Here's a kick for you.

Kick.

[Melissa G]

2. kick for BB! n/t

[autorank]

3. Bill don't take this personally, but I'm going to KICK you right now!

:kick::kick::kick::kick::kick::kick:

[loudsue]

4. Suggestion: ANY laws on the books of ANY State, need to include

that FORENSIC INVESTIGATION of programmed vote counting machines will also count toward the auditable recount.

Until we can get rid of these fraud-happy hackable machines, we at least need to include forensics in the mix. It also helps to send a message of the insidious nature of hacking a machine that counts votes.

:kick::kick::kick:

[Bill Bored]

5. Interesting, but I was more concerned about GETTING the recount. nt

[FogerRox]

6. The MAndatory #5 audit -- HOlts language on how to select

thos precincts is probably the best.
It cant be realy random-- to be fair.

[Bill Bored]

7. That CT law looked pretty random too as I recall.

But none of those laws say what to do after the 2, 3 or whatever % is audited. Presumably, if there is no discrepancy, you do NOTHING. That's not good enough. And what exactly is to be done if there is a discrepancy?

The above law addresses the issue of the election result possibly being overturned. That's really the bottom line, so I think it's useful and important.

The question is, given the rest of the laws, how much evidence would one need to get a manual count? I hope some legal minds will weigh in on this, otherwise I'll have to PM them or something.

What a difference a few months makes! Back in Nov/Dec, they'd have been all over this. Maybe they really are just waiting for the next stolen election. Human nature.

But rest assured, the Repukes have probably already hired a team of NY lawyers to exploit this law in their favor. I hope DU isn't reflective of the broader election reform community at this point! (I thought we WERE that community!)

Vigilance!

[sunshinekathy]

8. Mathematical & Practical Answers need funding

The NY Law Says:
(2) where evidence presented to the court otherwise indicates that
40 there is a likelihood of a material discrepancy between such manual
41 audit tally and such voting machine or system tally

1.
In order to produce that evidence, we need to fund the National Election Data Archive project, which, thus far, has only received an average $1400/month in donations and not enough funding to hire the technical or administrative staff to create it. (See http://electionarchive.org)

2.
None of the scenarios you mention would trigger a full recount, obviously. Please read this information to understand the probabilities of finding a miscounted precinct using a random audit:
http://uscountvotes.org/ucvAnalysis/US/paper-audits/Paper_Audits.pdf

The combination of 1. and 2. above would catch nearly all miscounts (and probably all miscounts that might affect the election results) because random audits would catch any miscounts that were spread amongst a substantial proportion of precincts, even when small, and an analysis of the data enabled by the National Election Data Archive project if funded, would catch any precincts that were substantially out of whack.

We really need help obtaining the funding for creating the National Election Data Archive soon, if we are to build it in time to prevent the wrong candidates from being sworn into office following the Nov 06 election. We only have about 40 sustaining members so far who are willing to donate $10/month to protect our democracy and we need at least 1,000 sustaining members to accomplish this big, but do-able project. We've designed the database and systems, but cannot implement them without full-time staff.

Please help spread the word by asking folks to subscribe to our mail list:
[e-mail]

to donate to the National Election Data Archive project:
http://electionarchive.org/fairelection/donate.html

to print & distribute our brochure and sign up as members:
http://electionarchive.net/docs_pdf/info/US/USCVbrochure.pdf

I cannot continue to do full time volunteer work and to put my own meager savings at $500/month like I have, into the effort to implement this system for much longer. I've had to pay for system administration, travel to conferences, the NASS.org conference, next the NASED.org conference, etc. plus worked over 60+ hour weeks on this without pay and I have no one to support me. The donations have not been enough to pay expenses to get what we've done so far done, and what we can get done without money to pay staff is causing the pace to be too slow to have any hope of implementation by Nov 06. We need a lot more folks who are patriotic to commit to the effort to prevent the wrong candidates from being sworn into office in the U.S., not just talk about it. If we don't solve this problem following the Nov 06 election, it may become impossible to solve it afterwards.

Harper's August issue mentions the National Election Data Archive project twice.

Best,


Kathy Dopp
http://electionarchive.org
--------------------------------

Please email [e-mail] to receive our email announcements.

Mission:
The National Election Data Archive is a scientific project whose mission is to investigate the accuracy of elections through the creation and analysis of a database containing precinct-level vote-type election data for the entire United States. By making detailed election data publicly available and, when warranted, by informing election officials and candidates of probable errors in local vote counts, our goal is to ensure that correctly elected candidates are sworn into office in future elections.

[Fly by night]

9. kick

[Land Shark]

10. My answers to your questions

I can't definitively answer these questions as to the regulation of the election board that is referenced (as subpart 1) because I don't have that information, but I'll answer with that uncertainty in mind. HOWEVER, I DON"T AT ALL LIKE THE "LOSING CANDIDATE" RESTRICTION ON STANDING because I consider losing candidates to be heavily deterred for political reasons from questioning results because of the glib and easy arguments available to the other side if a challenge is brought. Citizen rights are much better they are not so easily deterred from raising questions.

YOUR QUESTION #1, CAN A LOSING CANDIDATE GET A RECOUNT WHEN... A manual random audit of (only) 3% of machines or op scanners in the affected jurisdiction(s) but with no discrepancies found;

Nope. Hard to see how this would violate the unstated Board of elections regulation or would show that it is likely our losing candidate was actually the winning candidate, before getting to look at VVAR.

YOUR QUESTION #2, CAN A LOSING CANDIDATE GET A RECOUNT WHEN... A reported victory margin of say 5% i.e., 52-47, or any other one you'd care to use as a hypothetical;

Not under this statute, unless the regulation adopted by the Board of Elections provides for recounts when the margin is 5% or less, or the losing candidate has evidence other than the ballots and VVAR that shows it likely the losing candidate is actually the winning candidate, without examining the VVAR at all at the point the request is made to the judge.

YOUR QUESTION #3, CAN A LOSING CANDIDATE GET A RECOUNT WHEN... Enough votes cast on each machine or scanner such that a discrepancy in a SUBSET of the remaining 97% of unaudited machines in the affected jurisdiction(s) could be enough to result in a reversal of the election outcome (feel free to speculate as to the size of the subset);

Probably or maybe not. "Could be" enough does not seem to meet the "likely to change the result" standard which is higher than something that simply COULD be true. Likely would probably be interpreted to mean "more likely than not" or 50% probability plus a hair. Ask a statistician what discrepancy would have to be shown in a 3% sample to create a 50.0001 percent probability of result change.

4. Undisclosed software counting the votes (but the contest of election can force its disclosure in court according to another part of the law).

Unlikely to violate the election board's regulation (though I can't rule it out totally). Undisclosed might count for something in the judge's probabilistic calculation of likelihood but its weight depends on its combination with other facts.


BREAKDOWN OF STATUTE'S COMPONENTS

28 § 16-113. Audit of voter verifiable records.

Which people have rights?

A judge of a “supreme” court (the local trial court of general jurisdiction often called “Superior Court” in other states) or a county court judge, or any candidate or his or her agent.

What may these people do?

”Direct” a manual audit

Of What?

of the voter verifiable audit records

Any restrictions?

Yes, those VVAR “applicable to any candidate running for office within such judicial district or county”

Any further conditions that have to be met before triggering this?

Yes. EITHER the regulation of the state board regarding discrepancies is violated OR other evidence outside the VVAR indicates the RESULT WOULD LIKELY CHANGE and someone has sued for an audit

FULL TEXT. Yes, EITHER (1) the uniform statewide standard promulgated by regulation by the state board of elections pursuant to subdivision three of section 9-211 of this chapter with respect to discrepancies between manual audit tallies and voting machines or systems tallies requires a further voter verifiable record audit of additional voting machines or systems or all voting machines or systems applicable to such election, OR (2) where evidence presented to the court otherwise indicates that there is a likelihood of a material discrepancy between such manual audit tally and such voting machine or system tally which creates a substantial possibility that the winner of the election as reflected in the voting machine or system tally could change if a voter verifiable record audit of additional voting machines or systems or of all voting machines or systems applicable to such election were conducted.

[Bill Bored]

11. Now we're talkin'!

Thanks Land Shark!

Let me address your points:

Losing Candidate vs. Citizen Standing:
I agree and perhaps this is something we can work on. On the other hand, if the candidate doesn't want the job badly enough to challenge an unverified election, we probably shouldn't be voting for him/her in the first place, esp. in the Primaries. So I think there is some room to make this an issue in future campaigns so there won't be such a stigma attached to challenging otherwise unverified results. I.e., no more Kerry-Edwards '04s! We should expect nothing less than for our candidates to contest unverified results.

The rest:
You may be misunderstanding my hypotheticals. I was thinking that taken together they might be enough to get the recount in court, and not necessarily a full recount either. Just enough to rule out the possibility (or the likelihood?) that the result is wrong. Please read on.

The 3% random audit is already required by law, but if it's clean, we must assume that with a 5% margin, there will be no BoE regs that will trigger any additional auditing. Those regs haven't been promulgated yet so there's no point in discussing them now but let's assume they will say that if the 3% random audit is clean, with a 5% margin all that's left to do is to conduct the swearing in ceremony! So I'm assuming there is no discrepancy in the 3% audit so that no BoE regs will apply.

Now the evidence to present to the court is this:

With only a 3% random audit, 97% of machines are unaudited (virtually all of them).

With a 5% margin of victory, depending on the number of votes per machine and the number of machines in the jurisdiction, it would be a simple matter to determine how many votes on each machine would have to be switched on average to change the outcome.

Let's say there were 10,000 votes, 100 machines and 100 votes average per machine. The 5% margin = 500 votes, so a 251-vote swing would change the outcome. On average (since there are no fractional votes) that's 3 votes per machine, or 6 votes on half the machines, or 11 votes on a quarter of the machines, etc.

The only thing the 3% random audit proved was that vote switches did not occur on EVERY machine, because 3% didn't have any.

The software is proprietary, so without forensic analysis or independent auditing of the unaudited machines, there is no way to know how they actually counted the votes.

So there is definitely a possibility that the outcome of the election is incorrect. Auditing enough additional machines to rule this possibility out would confirm that the outcome of the election is correct.

Now when you start talking about the word "likelihood" instead of "possibility", you've got me there! But perhaps we can think of a few ways, such as the checkered history of DREs in general and those made by the vendor in question in particular, that a "possibility" can be changed to a "likelihood". What do you think?

Otherwise, perhaps we can just focus on changing that ONE word in the law itself.

As always, thanks for your time and attention!

[Land Shark]

12. One could say that 3% audits will check for "distributed error"

but do not check much for "undistributed error" or concentrated error. There should be other types of checks (deviation from past election results for example) to create alerts for non-distributed errors. This appears to be another way of re-phrasing what you are saying immediately above.

A "possibility" could be very small and its core meaning would be a probabilit of under 50% (we don't think of rainfall happening now as a "possibility" of rainfall 2 minutes from now since the probability is very high).

Something that is "likely" generally implies a probability OVER 50%. This is the basic difference I am pointing to in possibility v. "likely".

Possibilities get changed into likelihoods by adding facts, adding analysis or adding argument, or some combination of all three of these. (When I refer to "adding" facts I mean this in the most innocent way, not like the way they "add" votes in some jurisdictions)

[sunshinekathy]

13. distributed error in only 5 to 15% of precincts

You're right - but the distribution does not have to be very wide.

Please take the time to read this explanation of random audits which shows how surprisingly high the probability is of catching at least one miscounted precinct - even with a not so widely distributed error rate. I spent many days and many hours figuring this out and writing this for folks like you - so it is disappointing when people who claim to have interest in the subject don't bother to study it.

http://uscountvotes.org/ucvAnalysis/US/paper-audits/Paper_Audits.pdf

Please look at the charts and download the spreadsheet and play around with it so that you understand it completely.

You can see why a random audit in a small percentage of precincts, along with a mathematical study of ALL the detailed election results could catch almost any vote miscounts large enough to swing most elections.

Audits catch anything that is distributed - as machine vendor staff programming errors would most likely be, and the mathematical study of election results catches all those very out of whack precincts.

[FogerRox]

14. I talked to the Executive director of NJ Appleseed, she likes the

HOLT version of randomness, over S55 in Conn.

A truly random selection of precincts is not really what is needed.
That may not generate a reasonable "sample".

NJ has S29, there is no mandatory audit---
Which sucks big time.

[Land Shark]

15. one remaining problem is that 'catching' a miscounted precinct

while seen quite clearly by one with the mathematical credentials to see the significance, can just be negligently or willfully misunderstood by elections officials who, if one is to believe their email streams, focus an awful lot on how much time off they are going to have around the Thanksgiving and Christmas holidays.

So in a given statutory system i would look not only to what definitions are in the statute but as a practical matter how does it force attention to be brought to the error.

e.g. If by meeting a certain test the statute declared that this would constitute

!!!DEFCON TEN DEMOCRACY-BEING-HIJACKED ERROR CALL NATIONAL GUARD!!!

then i'd tend to think the statute would be effective.

technical stuff is so easy to ignore or obfuscate however, and lawsuits are slow and expensive.

The legal term is
REMEdIES
REMEDIES
REMEdiES

You don't want to have a right without a REMEDY. Nor a violation without a clear siren of some sort attached to it.

I don't know if any one has a list somewhere, but there are lots of error types that are NOT recount-detectable, and people ought to know that while recounts and audits are great, we still need to watch out for the types of irregularities that are not detected by audits or recounts.

[Bill Bored]

16. This is a very useful discussion!

First of all, the distinction between distributed and concentrated error/fraud is crucial and this is the first time I've seen it discussed here at ElectionFraud.net (which is the new domain name that points to this place in case you guys don't know that yet!).

The distinction between possibility and likelihood is also crucial.

I thank you Kathy for posting the random audit paper (instead of the usual exit poll stuff -- not that there's anything wrong with that) and in fact, if you recall, I emailed you about this sort of thing some time ago. Please do not assume that you are the only one thinking about this problem. You might want to include a 3% table just for the sake of convenience as there are some laws out there that require this and not all readers will take the trouble to use Excel.

It's clear that in addition to random auditing, which would detect most distributed fraud, we need ways of detecting more concentrated fraud.

Clearly a single outlier precinct that could change the outcome of an election, would attract some attention, but the real threat is a subset of corrupted machines that are only slightly more askew than every machine would have to be to change the outcome. This is easier to achieve with a close election since in such cases, the perpetrator would have to corrupt fewer machines than to overcome a landslide, and thus would be more likely to avoid detection.

Such tampering might not be detected by either random auditing or the usual "Poli-Sci" sort of analysis. E.g., instead of 1 vote per precinct being shifted, one could shift 4 votes in a quarter of the precincts, or 25 votes in only 4% of the precincts. In the last case, according to Kathy's paper, the odds of it being detected with a 2% audit approach 50/50. I'm not willing to take that chance!

And of course, if someone were serious about corrupting a vote count, they'd have their Poli-Sci-based cover story in place beforehand to explain away the vote shifting. For example, an anti-gay-marriage referendum on the ballot could be said to have "brought out the Republican Base" as an alternative explanation to shifting of votes away from the Dems. Or, if Early Voting results are the opposite of Election Day Voting results, the Poli-Sci explanation is that one party had an "aggressive early voting campaign" while the other party just showed up in droves on Election Day. The more of these alternate explanations there are, the more it dilutes the Election Fraud Hypothesis. And they can all be set up in advance along with the vote- shifting algorithms themselves.

The question this thread is concerned with is how to present to a court, enough evidence, or doubt, to recount the voter-verified paper ballots or records to be certain the election outcome is correct.

So, how about this: Grant a recount (either partially or completely) whenever there is a mathematical possibility that an error large enough to reverse the outcome is NOT LIKELY to be detected with a random audit or by any other means??? I realize I've just rewritten the statute at the top of this thread, but that's one reason why we're here. So what do you think?

As far as ways to steal votes that would not be detectable via recounts, I'm afraid this is beyond the scope of this thread, which is complicated enough at the moment Land Shark! But I hear you and there are some ways to address that too. I just think it's better to start another thread about it. I hope we can continue working on this one too though!

[sunshinekathy]

17. National Election Data Archive would catch Other vote Miscounts

The combination of independent audits in a small percentage of precincts along with the National Election Data Archive (NEDA) project would enable us to detect most, if not all, vote count errors that would cause the wrong candidate to be sworn into office immediately after the election (assuming we can convince candidates not to concede until after looking at the data - which should be easier after we've collected and analyzed the 2004 data by building NEDA).

i.e. independent audits of a small percentage of precincts catches distributed errors while mathematical analysis of detailed election results from all precincts would catch significant errors that were in any precincts.

If we can obtain enough funding and donations to build and test the National Election Data archive at least a year prior to the next election, then we would have a nation-wide database of detailed election results for every precinct in the U.S. that would permit us to detect all probable vote count errors that were significant, immediately following elections, prior to candidates conceding.

See this paper and the spreadsheet data that was collected and organized by Warren Stewart from the New Mexico 2004 election, for a demonstration of what can be uncovered (phantom votes in one vote type and suspicious under-votes in another) when detailed election data from every precinct is made available. We could detect probable vote counts post-election in time for candidates to contest their elections and provide court-worthy statistical evidence where necessary.

http://uscountvotes.org/index.php?option=com_content&task=view&id=70&Itemid=63

Please help us build this system by encouraging folks to sign up for our email announcement list, donate, or distribute our brochure (See the "Take Action" page on http://electionarchive.org ) I've been working full-time without pay on elections issues now for over a year, plus donating hundreds each month from my meager retirement savings to pay our expenses, and I have no one to support me. I cannot continue this much longer, plus I am overwhelmed with work and need assistance. We really need help - possibly PR/marketing help, grant-writing, data gathering, etc to build the system we spent months designing after the Nov 04 election but now need full-time paid staff to implement. We don't have much time left if it is to be ready by the Nov 06 election.

Thanks.

Kathy
http://electionarchive.org

[Bill Bored]

18. OK, I get it: Archive needs funding.

But what I'm trying to discuss here is how to get laws on the books that will allow the recounts once discrepancies large enough to reverse the election outcome are suspected.

We can have all the quantitative analysis imaginable but if the law (or a judge) says you can't have a recount, then you can't have a recount.

And of course many states still have nothing to recount (no paper). I suppose some forensic analysis of the DREs might be feasible but there's no guarantee that this will reveal self-deleting code, if it's done right. Nevertheless, in the states that will have paper to recount, the next step is to get some laws on the books that will actually allow it to be used for something.

I think we are on the right track in NY (we also have a law that can force source code to be revealed in contests of election), but there is room for improvement.

Your random audit paper will help. Thanks.

[Land Shark]

35. the Clermont 91 precincts would be less than 15% of ohio's

753 precincts in the 2d CD. I suppose an objecyion might be raised that these 91 precincts were already hand counted the First time (under humid conditions and all...)

[kiwi_expat]

19. Lessons from Ohio Recount re. randomness, over/under votes, and timing

Hi Bill

The disastrous Greens-Libertarian recount in Ohio has demonstrated the importance of randomness, over/under votes, and timing.

RANDOMNESS
You state that the NY legislation context includes "mandatory manual random auditing of 3% of DREs or Op Scanners in each jurisdiction". Who decides what method is used to randomly select 3% of the machines in a jurisdiction?

Ideally, a uniform selection method should be stated explicitly in the appropriate NY State legislation.


UNDER/OVER VOTES
A very big question is, how are over/under votes handled in that 3% mandatory manual random auditing? If the BoEs have the discretion to decide how to handle those votes, the under/over votes will not be counted! That is what happened in Ohio.

However, if the under/over votes in that 3% audit are examined for voter intent, and included in the recount totals, most jurisdictions would probably have discrepancies with the original counts, and thus automatically have a full manual recount. (Ideally, examing under/over votes should be required to be done by neutral persons with no stake in avoiding a full manual recount.)

Under/over votes were hugely important in Ohio, which uses mainly punched cards plus some Op scanners. I'm not so sure about NY with its DREs.


TIMING
If the 3% manual recount includes under/over votes - thus forcing full manual recounts in many jurisdictions - the law needs to ensure that the full recount will be done *before* vote certification. In Ohio, Blackwell demonstrated how the clock can be run out by using the latest possible date for each phase of the counting process.

The NY law should be modified, if necessary, to force 3% recounts to be started *before* the absentee and provisional ballots are counted - and for the resulting full manual recounts to be completed *before* vote certification. Then, hopefully, you won't have to worry so much about changing the legislation which provides for overturning the certified vote. :-)


Cheers!

[Bill Bored]

20. Good points.

But I'm assuming that the 3% random audit really is clean and there is still undetected fraud. In that case, we need another way to get the recount.

That said, the types of discrepancies you are suggesting would be covered by as yet undefined BoE regulations -- not by state law. I prefer to assume that these regulations will be useless, until they are issued and proven to be otherwise. If the BoE comes up with some fool-proof regulations, then so much the better, but I am not optimistic about that. They like to trust their machines, otherwise they wouldn't buy them in the first place. One of the problems with the existing law is that it gives too much discretion to the BoE.

Therefore, I am mostly concerned with making it easy for the BoE to be overruled by the courts if there is a possibility that the outcome of the election is being decided incorrectly, even if the 3% audit was spot on! That way, a candidate can always challenge the count.

[kiwi_expat]

21. The challenge needs to be reviewed by the Court in time....

...for resulting recounts to be completed before the Electoral College convenes.

Can your proposed legislation mods stipulate a time frame for the Court review and subsequent recounts?

As I recall, the election challenge in Ohio was not addressed by the Ohio Supreme Court until after Bush was elected by the Electoral College. The Court has yet to hear the Greens' recount challenge.

[Bill Bored]

22. I agree. All I can say is that our SoS doesn't run our elections here.

And in fact I don't think there's any one person or party who does that in NY. The BoE is bi-partisan and there are two head honchos under another new law.

The courts of course are another story.

We'll try to do something about the time factor. It could be that the term "material discrepancy" addresses this because the discrepancy wouldn't matter if it wasn't discovered in time to correct it. I.e., it's only material if it's found before the Electors vote. My main problem is the use of the term "likelihood." In some cases, you don't know if you have a likelihood of a material discrepancy until you do the recount.

Also, keep in mind that the Supreme Court mentioned in this law is NOT the highest court in NY, as Land Shark has pointed out.
See: http://www.courts.state.ny.us/courts/structure.shtml
So we are thinking in terms of fairly quick action, i.e., not having to appeal to a higher court, unless of course the judge says "no." That's why the law needs to be worded to make it easier to get a "yes" to any reasonable recount request.

One other thing about Ohio is that Mr. Kerry was unfortunately not a party to that recount, right?

[kiwi_expat]

25. Kerry joined the "recount fray" AFTER the 3% manual recount.

"One other thing about Ohio is that Mr. Kerry was unfortunately not a party to that recount, right?" -BB

Kerry to Enter Ohio Recount Fray By William Rivers Pitt, Thursday 23 December 2004
http://www.truthout.org/docs_04/122404Y.shtml

[Bill Bored]

27. Too little too late.

So what happened? I know Will was writing about this and it gave some of us a lot of hope, myself included. But so far this hasn't gone anywhere, has it?

This recount was never taken seriously by the bureaucrats in Ohio because Kerry didn't want it. He was a miserable failure on this issue. Three other states, not even including FL, should have been recounted too. They were much closer than Ohio and they would have tied the Electoral College.

"Specifically, Kerry will be filing a request for expedited discovery regarding Triad Systems voting machines, as well as a motion for a preservation order to protect any and all discovery and preserve any evidence on this matter."

So what did they discover expeditiously?

[kiwi_expat]

26. Ohio recount mess was not just the result of an evil SOS running the show

The BoEs, themselves, are motivated to avoid randomness and under/overvotes.

The individual Ohio county BoEs decided how to "randomly" select precincts for the recounts. Some counties, like Hamilton, made a real attempt at randomness. Others, like Clermont and Cuyahoga, deliberately chose precincts that they thought would be less likely to have problems.

And if examining under/over votes is left to the discretion of the BoEs, the under/over votes will be ignored as much as possible to avoid a discrepancy (and thus avoid having to do a full manual recount).

If there is a CHANCE that some NY BoEs might chose to avoid randomness or under/overvotes, I encourage you to add those two conditions to the four conditions you have already listed in your original message. Something along the lines of:

"1. A manual audit of (only) 3% of machines or op scanners in the affected jurisdiction(s) but with no discrepancies found; and -

"2. The manual audit selection was not based on a valid random-sampling method; or -

"3. The manual audit did not include inspection of all undervotes and overvotes to check for voter intent; or -"


(In my opinion, fraud is more likely to be missed because the BoEs did not do a proper 3% recount, than because the fraud was somewhere other than in the 3% sample.)

[Bill Bored]

28. Points 1 and 2 would be covered by this law.

Kiwi, the whole point of this law is for a court to be able to overrule the BoE, Sos, or whoever else says you can't have a recount. I don't think they had anything like this in Ohio and it appears that their random selection "law" was really just a regulation that Blackwell or a previous SoS had issued, and then violated. I'm waiting for proof to the contrary in another thread:
<http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=387275&mesg_id=387562>

As to point 3, you wrote:

"In my opinion, fraud is more likely to be missed because the BoEs did not do a proper 3% recount, than because the fraud was somewhere other than in the 3% sample."

But the whole point of the 3% audit is to determine if there should be additional audits.

If you're saying that there might be enough undervotes in a 3% sample to swing the election if voter intent were known, I'm sure a recount could be granted based on such a narrow margin alone. If not, then the issue is how to get more audits because we don't trust the 3% (or the other 97%). If it can be shown that the selection was in fact non-random, then that just adds to the evidence for the recount. Remember, we already have a 3% random audit law (not just a regulation).

At that point undervotes on Op Scan ballots, which should be rare, would be detectable and counted if possible.

Voter intent could not be inferred from DRE undervotes though because there are no hanging chads or lightly filled-in ovals to show how the voters would have voted if the machine had counted the votes. Only if the VVPATs didn't match the count on the DRE would there be a discrepancy or correction.

You can't have VVPAT printers behaving like punch cards! There are no chads, there are only voters' choices, which should be identical on the VVPAT, DRE Summary Screens, and in the votes database. If they aren't, then you'll have a discrepancy in the audit and you'll get a recount right away.

The problem of not getting a full recount in Ohio would be solved by this law, which would say you get one no matter how perfect the initial audit was if you can show other evidence that the outcome might be reversed.

What would be the scenario in which a judge would deny a recount if this law were tweaked to change "likelihood" of a discrepancy to "possibility" of a discrepancy, for example? (I know it's up to a judge to interpret all this but let's assume we get the language crystal clear!)

[kiwi_expat]

31. I think this law is great.

I'm sorry, Bill. I mistakenly thought you were considering amending the law to include specific wording such as you indicated in your 4 conditions (e.g., "Enough votes cast on each machine or scanner such that a discrepancy in a SUBSET of the remaining 97% of unaudited machines in the affected jurisdiction(s) could be enough to result in a reversal of the election outcome".)

* * *

"In my opinion, fraud is more likely to be missed because the BoEs did not do a proper 3% recount, than because the fraud was somewhere other than in the 3% sample." -ke

"But the whole point of the 3% audit is to determine if there should be additional audits." -bb

Sorry about the confusion. (That usually happens when I try to be concise.) I agree that the whole point of the 3% audit is to determine if there should be additional audits. What I should have said: I think the idea of having an additional way of getting a full manual recount to detect fraud is great. But usually if there is fraud, evidence would show up in the 3% sample - if the samples are randomly selected. IMO if there is fraud in a county but the 3% manual recount showed no discrepancies, it is most likely that the 3% selection was not truly random.

I might not have made myself clear, even yet. Never mind.


"The problem of not getting a full recount in Ohio would be solved by this law, which would say you get one no matter how perfect the initial audit was if you can show other evidence that the outcome might be reversed." -bb

Yes, that is a great idea. Even better would be for Ohio to have a law like this which ALSO included provisions for a full recount if the 3% selection was not truly random or the under/overvotes were not properly inspected. :-)


"What would be the scenario in which a judge would deny a recount if this law were tweaked to change "likelihood" of a discrepancy to "possibility" of a discrepancy, for example? (I know it's up to a judge to interpret all this but let's assume we get the language crystal clear!" -bb

I can't imagine a judge ever denying a recount if the law stipulated "possibility".

[Bill Bored]

32. Friendly vs. Unfriendly Judges.

As seen in 2000, it's possible for unfriendly judges to interpret the law in an unfriendly way!

I'm thinking of changing to word MAY to SHALL, which would in effect force the judge to examine the evidence and order the recount if the there's a REASONABLE POSSIBILITY that the outcome could be in error.

I think this combination would cover most if not all of the contingencies, including unfriendly judges who were after all elected using the same system that's being questioned here.

I'm talking to a NY lawyer about this and I hope we can agree to try to move forward with this.

Always interested in other views of course!

[kiwi_expat]

33. "must" vs. "shall"

I agree, "shall" is much better than "may". When imposing an obligation, speak in mandatory terms.

Would "must" be even more emphatic than "shall"? (Both "shall" and "must" are used in NZ law.) If "shall" would in effect force the judge to order the recount if the there's a REASONABLE possibility, could "must" force the judge to order a recount if there is ANY possibility??

Can an unfriendly judge be forced to do anything?

[Bill Bored]

34. Getting kind of hypothetical now aren't we?

Here we are discussing these subtleties of language and there are still states without even a paper trail! What a country!

[Columbus]

23. Some answers from NY attorney

I forwarded your questions to some NY attorneys. I was advised that you will find your answers at the below url (particularly the highlighted portions). I have not read this thread (I have not even read what I am sending, I have very briefly glanced the document) and I noticed there are alot of responses already. I hope the below answers your questions, if they have not already been answered. When I have a bit more time I will read the thread. I hope for now this helps.


http://www.wheresthepaper.org/S5877_A8969WithComments.doc

[Bill Bored]

24. Thanks Columbus

This is Teresa Hommel's website. I know her.

These comments, and others, most of which are excellent, were sent to our legislature and duly ignored!

I'm surprised you've got attorneys actually citing them, unless they wrote them.

The law that was eventually enacted was not changed much from the Democratic version you cite here, at least not in this section. There were a lot of good comments in the document you cite.

Now that the law has passed, it's time to amend it! Never give up!

I'm thinking small steps. Deceptively simple stuff like just get me a hand count when the numbers and auditing are such that the machine count can't be trusted. Do the math after each election and each audit, and if there's still a possibility the outcome is wrong, go to court. You can use the poli-sci stuff to back you back, but first, just do the math and don't assume that the machine count is right -- ever.

I'm not crazy about using exit polls and so on as suggested in this document. There are more substantial and yet general ways to ask for a recount. Any time that it's mathematically possible that the outcome may be incorrect, even given the initial or subsequent random auditing, count the paper until it's not possible! If you do that, it doesn't matter what the exit polls say. We have the paper ballots/audit records. We just need to use them effectively.

[Janice325]

29. Another kick!!

[Bill Bored]

30. Thanks for the comments and kicks! Just to clarify where we're at:

1. By LAW, NY already has VVPATs or VVPBs (Op Scan) that can be recounted to confirm machine or scanner (system) counts.

2. We also have a mandatory manual random audit of 3% of DREs or Op Scanners in each jurisdiction in the state (the counties and the City of NY). This is also a state LAW -- not just a BoE regulation. (I think it's important to distinguish between the two.)

3. It's up to the BoE to write statewide regulations (but not LAWS) about what to audit next, i.e., if there is a discrepancy in the initial 3%.

4. Also by LAW, if there is a full recount, or an individual machine or system has been found to have failed to record votes indicating an "operational failure", the hand count becomes the official count.

5. The law at the top of this thread (which is also a LAW!), states under what conditions a candidate can go to Court to request a recount, i.e., if he/she isn't happy with the BoE's decision.

6. I'm just trying to improve this law to make it a little easier to get hand counts since we know that even a perfect 3% audit does not rule out fraud. The goal is to prevent any fraud that could change the outcome of an election -- not necessarily to count _every_ vote.

7. The term "likelihood" as it applies in this law, may be the biggest problem since it implies some a priori knowledge of the existence of a material discrepancy between paper ballots/audit trails and machine counts large enough to affect the outcome. Such a likelihood can not be inferred unless the requested manual audit takes place. So the law is putting the cart before the horse, IMO, and we need to amend it.

That's where we're at. A simple amendment could go a long way to improving this law, IMO. And just maybe it can serve as a model for other states.