Audit Protocol: Input Requested

[Einsteinia]

Here is a comprehensive audit protocol for California:

>>> http://www.califelectprotect.net/GSAP_16d.pdf <<<<

The hierarchy of principles and using "Consilience" as a paradigm for
"checks and balances" are NOT negotiable, BUT the details about how these
principles would be applied are modular and negotiable, including
the name of the protocol.

If you have the time or inclination, please take a peek and
read it through completely before forming any conclusions.
(It may not end the way you think.)

If you have any insights, please write asap (today or early tomorrow):
[e-mail]


Here are some points this protocol addresses:

The Gold Star Audit Protocol will explain why:

o We must prepare our highschool students--our next generation--
for maintaining democracy at home with more than R.O.T.C.

o The issue is NOT whether ballots shall be
hand counted -OR- computer counted

o We may use slightly customized “off-the-shelf”
computer software and hardware
to save money without sacrificing security

o The biggest paradigm shift we’re about to encounter is
whether the control of our ballots will be
centralized—under the primary purview and of the county
-OR-
decentralized—first under the purview of ALL the local precincts,
then with clear divisions of power
between the various government entities

o We must adopt a firm hierarchy of principles
and a new paradigm for “checks and balances”
that will guide all our the audit protocol methods

o There is wisdom of building oversight functions
that overlap multiple ways

o An audit protocol is only as strong as its weakest link

o This “Manhattan Project” for a bona fide democracy
is not only possible—it is mandatory
IF we want to continue to call ourselves a democracy

Thank you to all that have time to do this. . . .

[OnTheOtherHand]

1. whoa, they need comments by tomorrow? today?

I'm not wedded to the implementation of the "Step A" 10%-of-each-precinct count, but -- if indeed the human resources can be made available for it -- it certainly offers one way of finding likely anomalies in the count. The general concept that multiple methods are more robust makes sense (although I'm not sure what the point is of calling this "consilience," but hey, whatever).

There is no such word as "Foreward," AFAIK.

It really does require and reward a close reading, because they've paid close attention to many details. I'm sure I haven't picked up on all the nuances.

[Einsteinia]

3. Thanks

Yikes! Yes, of course, it's the "Foreword"!!!!

Well, that's what we need you for!

About the 10%--it is only 100 ballots and it is only one part of the system.

This works a bit like the "Manhattan Project" where no one has all the pieces until the end--so literally no ONE person can cheat the system.

It may seem complex, but it really isn't.

[OnTheOtherHand]

5. yeah, I like the emphasis on decentralizing as much as possible

It's an excellent design principle. And if people in every precinct have at least a pretty good idea what the the vote totals were, big hacks should be a lot harder.

I have no idea whether the Voteneer concept will fly, but hey, people say they support high school community service, so we might as well find out. (Even if people don't go for Voteneers, we might find something else very uesful for high school students to do.)

I'm a bit fuzzy on the specifics of getting from the first to the second step -- should I stare at that?

[furrball]

15. Just read it--I think this is the BEST solution yet.

This solves all the problems right now:

OVC--moot
hcpb vs. machine--moot
proprietary software--moot

verifiable democratic process that is transparent to all.

Thank you for ALL this work, and I think it very exciting the Prof. David Dill supports it.

[Einsteinia]

6. Oh, the deadline--early Thursday is okay

But it's cutting it short. However, please just take your time and get it right, because we may be able to amend the details later. BUT the sooner the better.

[Wilms]

2. Thanks, E.

This is a neat document. But I have to say I'm not sufficiently knowledgeable to make many comments. I hope the DU Statistician Corp can take a moment away from exit poll pie-fights to review your effort.

That said...

I agree that e-counting is expensive, but you may want to differentiate between DRE's, precinct-based OpScan, and Central Count OpScans as I suspect they each have their distinct capital requirement.

I appreciate the analogy of Voteneers to ROTC members but the reference to ROTC (sans Voteneer) in the forward seems sarcastic, and thus not appropriate for scholarly work. It may be best to stick directly to the subject at hand, and save the critique of our ailing democracy for a different time and separate treatment.

Similarly, the term "Gold Star" seems more about branding. Perhaps a technically descriptive term would better suit the title.

Thanks, again.

[Einsteinia]

4. Good insights. . . .

It's hard to know how things are taken. I was trying to give a reason why a elector might want to expend some more dollars on an audit protocol over, say, the closing of a health clinic. Hard choices. But to remember how much people have paid with their lives. Thus, the gold star sort of invokes the military need for bona fide democracies, plus the "gold" means top rate. Right now the Carter Center says that El Salavador has excellent election hygiene, whereas we don't.

BUT, if it's not hitting those concepts--any better would be much appreciated.

Do you or does anyone have any suggestions?

I know this is time consuming, but if WE can get it right, we'll all benefit. . . .

[Wilms]

7. Other than Consilience, which I don't think I'd actually recommend...

...nothing else is coming to mind.


Was just reading again and this jumped out...

"This “Manhattan Project” for a bona fide democracy is not only
possible—it is mandatory
IF we want to continue to call ourselves a democracy"


I agree, but some reading it may take offense.

I'd comb the article for anything that indicates political bias, or could be construed as lecturing the reader.

If something isn't absolutely necessary to cover the subject, it doesn't belong in the paper.

[Einsteinia]

8. Good points, but I'm curious

why you're against :"consilience?" The whole thing is structured on this concept of two different methodologies overseen by two different groups (or more) must come up with the same result. We cannot use "doube entry accounting," because we lack the transparency of transactions, and so this offers methodical way to assure the results are verifiable.

[Wilms]

9. I'm embarrassed to say...

I just don't like the word.

:blush:

[Einsteinia]

12. That's valid--and I

sorta agree that it sounds pompous-- BUT
'
it serves such a marvelous purpose to put a name on a methodology to arrive at truth. It's hard to find a concrete basis for a system when there is not transparency of transactions.

E.O. Wilson posed it as a remedy to the relativism of French deconstructionists who make everything relative--to the point where nothing is knowable. This philosophy has filtered into our day-to-day life in so many ways--including a lack of rigor in our election systems. It's really true with what Paul Krugman points out in the crisis of journalism where "flat earth" theories on put on par with "round earth" theories.

So. . . . I reluctantly think I need to keep it in there for the time being BUT if another word or concept could replace it, we'll keep our minds open. BTW, Prof. David Dill says he wants to work on it with us.

[Wilms]

13. Excellent news.

Perhaps he'll have an idea.

It could just jump off the page as you develop it further. Use the force.

[AmBlue]

38. I agree.... sounds like:

su·per·cil·i·ous ( P ) Pronunciation Key (spr-sl-s)
adj. Feeling or showing haughty disdain. See Synonyms at proud.

[Einsteinia]

40. It may sound fancy, but the distinguished

Harvard biologist E.O. Wilson said he prefers it over other words because the other words have multiple meaning that may be misconstrued. So as Einstein said, "State things as simple as possible, but no simpler" wouldn't you prefer haughty over ambiguity?

Also, it is a paradigm based on a specific way to arrive at truth, which E.O. Wilson devoted an entire book called, "Consilience."

I like it because it is so very specific in its meaning that it could stand up in court.

But with all the said, do you still think it is a problem word?

[Febble]

16. Edited contribution

from a departed soul, pies deleted:

Einsteinia's proposal says "The PRECINCT Audit Model examines a small random selection of ballots from ALL precincts,"

This is NOT an audit. It is "post-election voting machine testing" and is NOT a check on the accuracy of any election's vote COUNTS.

Auditing randomly selected ballots does NOTHING to ensure the accuracy of any vote counts because voting machines can trivially be programmed to count accurately except during an actual election, so testing randomly selected ballots "after" an election is useless to ensure vote count accuracy for counts created during an election. Auditing ballots does not check any COUNTS that were created during the election at all.

Someone needs to inform Einsteinia that their audit proposal needs to be modified to be an audit of "vote counts", not "ballots" unless they want the election tamperers to jump for joy if their "audit" proposal passes.

Einsteinia needs to be informed that it would be, alternatively, OK to audit 10% of "machine counts" in every precinct, but NOT "ballots".

However, then they will not be able to independently audit the precinct counts that way, so it is not a preferable plan.

<more snipped pies>

...An audit never gives a 100% chance of detecting vote miscounts, but it comes pretty close.

FWIW

[Febble]

17. BTW

Good job, Einsteinia! It's a great start.

[Einsteinia]

23. Thank you! This now needs everyone's

input to take it to the next level and make it truly airtight.

[Febble]

25. Great. I thought I'd missed the boat.

I'll take a look, but you seem to have lots of people on to it.

[OnTheOtherHand]

18. is it fair to wonder

how the departed soul thinks that exit polls can provide virtually irrefutable evidence of vote miscount, but sampled ballots are utterly useless? I don't quite understand the principle at work here.

(EDIT: Someone is thinking, "But, wait, doesn't he say exit polls are useless?" Well, the trouble with exit polls is that we can't really force people to respond at random. EDIT 2: I forgot to point out explicitly: no, I don't say that exit polls are useless; I just don't trust them. Anyway, back to the subject at hand: getting a 10% random sample of ballots might not be as simple as it sounds either, but it looks like the Gold Standard team has taken a pretty good crack at it.)

But I certainly agree that a 10% sample on its own wouldn't do much good -- of course, that isn't the proposal.

Once I understood it, I got pretty interested in yowsa^3's idea that one could defeat a lot of op-scan hacks and problems by doing a partial hand recount and then machine-scanning (1) the hand-recounted ballots, (2) all other ballots, and (3) the entire batch to make sure the totals reconciled.

[Febble]

20. Well, I think one argument

is that voters are sampled before the count, whereas the audit occurs afterwards.

One thing I liked about Einsteinia's proposal is the principle of:

"The Importance of Tethering ALL Votes to Neighborhood Precincts", which addresses a point the dear departed has made elsewhere about absentee miscounts cancelling out at-the-precinct miscounts.

What does seem extraordinary to me, as a Brit, is that you potentially have a lot of ways to check for consistency between levels of counting, from the precinct itself upwards, and yet people are able to allege that vote-switching in 2004 could have occurred in either place. It ought to be pretty simple to check where discrepancies occur if you have two opportunities to observe the tally - as long as someone does.

I also like the fact that the document pays serious attention to the varieties of "miscount" that could occur, especially the ubiquitous salami slice, which would be the ideal fraud if you could pull it off, and by far the hardest to detect.

[Einsteinia]

32. What is the salami slice?

It sounds intriguing? I think I want some.

[Febble]

35. Well, it's good on rye, but not on elections.

I mean trimming a very few votes from as many precincts as possible. I hope it's not possible, because it would be very hard to detect, and plausibly deniable if it was.

Very thin slices, so thin no-one would notice, but adding up to a whole salami.

But it would be difficult, if not impossible, to pull off on a scale large enough to matter, IMO. Depends what gives you nightmares.

[Einsteinia]

42. And now that you've given it some thought

Do you think this overlapping effectively closes all the holes?

We're not statisticians, but intuitively it just seemed to solve the problems.

Also, we have a scheme to make it generally random, BUT if that got subverted the 3rd audit isn't random--it's targeted based on the precincts with the most discrepancies between the hand and machine count on the precinct level.

It seems iron clad, but is it?

[Einsteinia]

22. This is only 1 of 3 separate and overlapping

audits that are contained in this. Look at the graphic towards the end.

[Febble]

24. Yes, I saw that.

I think it is an excellent document. Congratulations.

[Einsteinia]

26. Wow! You caught me off guard--I was all ready

to defend! Thanks!!!

[Febble]

31. Well, what I posted wasn't my critique

it was Kathy Dopp's. As she's quite widely cited on the subject of audits, I thought I'd pass it on.

Edit: There's more critique from Kathy on Progressive Independent, but you'll have to do a bit of shovelling to get at it.

I won't link.

[Einsteinia]

33. Hmmm? We did get input from Kathy Dopp and

it informed one of the three audits that overlap, and so I'll be curious if after she actually reads it (if she does), if she'll be for or against it. I'm curious. . . .

[Febble]

34. Well, at present

she says she thinks that OTOH and I are in the pay of election tamperers, so be warned.

Of something, at any rate.

[Einsteinia]

36. Huh? Can you. . . .

elaborate so that if I hear such allegations that I might be able to respond. Or is it the just another dimension in Twilight Zone that a few of our allies seem so partial to falling into now and again?

[Febble]

37. That's the one.

I'll PM you the link. I won't dignify it with a public one.

[OnTheOtherHand]

21. I was trying to think of a statistical question to address...

I think many issues about audit protocols hinge on empirical questions that I don't have the answer to. (I'm not sure that anyone does, although I know some folks know more than I do.) For instance, I would like to have some sense of the ranges of observed proportions of op-scan ballots for which the voter intent is clear (by some explicit standard) but which machines will not count in that manner. We haven't even gotten into that issue with respect to the Gold Standard -- dunno, maybe I was just too sleepy.

Here is something that I can point out about the 10% portion of the protocol. Suppose that a precinct has 700 votes, which actually divide evenly -- 350 for candidate A, 350 for candidate B (I will abbreviate that 350:350). But 70 votes are shifted from A to B, so the vote is reported as 280:420 -- a 10% vote shift (from 50%:50% to 40%:60%).

Now we draw a random sample of 70 ballots from the precinct. On the assumption that the true vote count is 280:420, there is about a 98.6% chance that our sample will turn out somewhere between 19:51 and 37:33 (this is a 98% hypergeometric confidence interval). But, since the ballots are actually distributed 350:350, there is about a 26% chance that the precinct sample will lie outside the 98% interval (specifically, that we will end up with a sample favoring A by 38:32 or more), which means that it will probably be selected for our 2% sample. (At least, that's what it means if I understand how the 2% sample works.)

A 26% chance of detecting fraud may not sound very good, until you consider that it applies to every precinct. If the fraud extended to twenty precincts, there is only about one chance in 300 that none of them would be selected for the 2% audit. It would be pretty hard to steal a statewide (or even a large countywide) election through 10% vote shift in fewer than twenty precincts. And if we try for 20% vote shift (i.e., the official count is 210:490 although the actual ballots are 350:350), then any one such precinct has about an 87% chance of getting selected for the 2% audit. (The chances will be smaller with smaller precincts, but then of course the vote yield will be smaller, too.)

There is no special significance or importance to the exact numbers I used in this example, and I'm not trying to convince anyone that this approach is foolproof. But it may help someone who didn't quite grasp the original write-up to understand why the double-barreled audit approach is appealing.

[eomer]

10. Here are a few comments.

They are sort of random as I don't have time to do it exhaustively (a deadline looms). Also, sorry to do it late but you said you may still be able to incorporate late suggestions. :)

Bi-Hourly Computer Tally. If, and only if, ANY
person is taking unofficial precinct tallies at anytime
during election day, this information must be made
public, as follows:

Every two hours the unofficial preliminary tallies
from the computerized voting shall be posted with
the time of each tally indicated. The tallies shall be
posted prominently in the immediate proximity of
the entrance to the precinct.

Is the public (that is, beyond those who are in line to vote; including the press) allowed to approach the entrance of the polling place in all jurisdictions while voting is underway under current law? I don't know the details of the 100 foot rules (or however many feet). Do they only prohibit electioneering or do they prohibit presence (and also, are they ambiguous on the point, which would be a flaw that could be exploited).

Also, the preceding point made me think, is it proper to post results in a time and a place that might affect how the remaining people will vote? It sounds like those waiting in line to vote would get a peek at how others in their area have voted, just before they go in to vote. Sounds like a problem to me. Similarly for posting results during election day on the SOS website -- wouldn't that potentially influence voters who vote later in the day?

Also, would it be more clear to say "polling place" in this context than "precinct"? Precinct is a geographical boundary that defines a set of voters; polling place is the physical location where voting takes place for a precinct. I'm assuming you want the manual count to be done at the polling place and not at some other location that someone might designate as the precinct count location. Otherwise you have a more complicated chain of custody problem.

Definition of “Manual” & “Hand”
The terms “manual” and “hand” are used interchangeably to
mean a process conducted by using human eyes and hands to
count ballots (and/or paper audit trails) without computerized
assistance.

Should you say something about the standard to be used? Gore v. Bush said there has to be a consistent statewide standard and, I believe, that it must be specified in law before the election begins. You should probably say here that the manual counting must apply the standard specified by state law. Edit to add: in the Ohio recount there were counties that totally disregarded the standard specified by law and instead used a procedure that was intended to mimic the machine and force the manual count to match the machine count. Just another reason the standard to be used should be clearly specified.

Final Tallies. Upon the close of polls, the precinct’s
Election Inspector will be given a computer
password that will permit access to a link that will
automatically and immediately post the information
on the Secretary of States website, as well as to the
County.

I think you should be specific that the result for each individual precinct must be posted separately on the SOS website so no one thinks that posting aggregated results satisfies this requirement. Of course, I know this document isn't intended to be the statutory language so there would be lots of places like this where one needs to tighten the language as much as possible to prevent intentional misinterpretations (like the non-random random selection of precincts in Ohio).

Paper Stock
i. size: 8.5 x 11” sheets (not on rolls)
ii. weight: 20 lb. minimum
iii. color: white
iv. type: archival for at least 10 years
v. considerations: used use of recycled paper is preferable

I've got to run now. If I get a chance I'll look some more another time. Great job, by the way and thanks for doing it!

Cheers,
eomer

[Einsteinia]

11. Wow! Thanks!

I'll copy these and add these. I appreciate your time.

[Einsteinia]

14. The link has now been updated

significantly based on ALL the great input from all. Thanks so very much.

[OnTheOtherHand]

19. so now you are calling it "Gold Standard" instead of "Gold Star"?

(I do prefer "Gold Standard" -- "Gold Star" kinda takes me back to second grade. Not that I care much either way!)

[Einsteinia]

27. Voteneers! Look what's up now!

On Friday we presented o Senator Bowen's office our Gold Standard Audit Protocol ("GSAP"), which among so many things extols the importance of using high school students ("Voteneers") as poll workers--because they need to learn more about maintaining democracy than just R.O.T.C.

Look what's up today!

#: ) Einsteinia


---------------------------------------------------------------------------------------------------------------


BOWEN INTRODUCES MEASURE TO ENCOURAGE HIGH

SCHOOL STUDENTS TO VOLUNTEER AT THE POLLS ON ELECTION DAY

 

SACRAMENTO –Encouraging students to experience democracy first-hand by volunteering at polling places on Election Day is the goal of SB 1193 by Senator Debra Bowen (D-Redondo Beach), which was introduced today at the State Capitol.  The bill eliminates a provision of state law that cuts school funding when students work at the polls on Election Day.

 

“When students spend Election Day at the polls, they get a hands-on lesson in how our government works that’s certainly as valuable as reading a chapter in a textbook,”said Bowen.  “Encouraging young people to lend a hand at the polls lets them know voting is one of their most important civic responsibilities, which is why taking money away from schools that let their students volunteer at the polls makes no sense.”      

 

Under current law, students can work at the polls on Election Day if they’re 16-years-old and are a high school senior with a grade point average of at least 2.5.  However, if a student misses school to work at the polls, the school loses the “average daily attendance” (ADA) payments it’s entitled to receive from the state for that student.  As a result, many schools don’t encourage students to serve as poll workers on Election Day.  SB 1193 allows students who volunteer at the polls on Election Day to be considered “under the immediate supervision of the certificated teacher of the history or social science course” and allows the school to receive the ADA payment as if the student was in the classroom. 

 

“It makes no sense to put up road blocks to prevent high school students who want to get involved in their community from working at the polls on Election Day,” continued Bowen.  “This is really a common sense proposal that works as an investment in both our students and our democracy.”

 

According to the Secretary of State’s office, some 80,000 volunteers are needed to serve as poll workers during a typical statewide election and it’s common for counties to face a shortage of people who are willing to work at the polls on Election Day.  For example, during the November 8, 2005 special election, county registrars of voters in Alameda, Butte, San Diego, and Santa Clara counties were forced to scramble to find poll workers at the last minute. 

 

SB 1193 is similar to AB 1944 (Hancock) of 2004, which was approved by the Legislature on a bipartisan basis before being vetoed by Governor Schwarzenegger.  In his veto message, the Governor wrote, “While civic and other volunteer activities can offer many educational opportunities to students, these activities should be in addition to, and not in place of, valuable classroom learning time with a teacher.”

 

“Time in the classroom is certainly important, but it’s not the only way kids can learn how our democracy is supposed to work,” concluded Bowen.  “This isn’t any different than a school that takes a field trip to get students out of the classroom and put a ‘real world’ face on what they’re learning in textbooks.  Schools shouldn’t be penalized because they want to let their students learn first-hand for themselves about the nuts and bolts of how our democracy works.”

 

SB 1193 will be heard in the Senate Education Committee in the coming weeks.

 

###

 

 

[FogerRox]

28. just freaggin awesome

[Amaryllis]

29. Bowen is so totally wonderful! She really gets it.

[Einsteinia]

30. Yes is bold. She's our best hope.

I'm so impressed with her and her office. They are NOT typical!

[Amaryllis]

39. Shelley wasn't typical either. You know how to pick 'em in CA! The ones

who will speak truth to power.

[Einsteinia]

41. You're a longtime supporter AND

and you're not even a Californian. And I'd say that the vast majority of Calfiornians are clueless BUT the tide just might be turning. The other night in a local assembly race in Marin County a candidate who had previously been quite spineless on election reform spoke out quite eloquently on the crisis with the Diebold equipment. He's on my spam list and I think he must be finally reading it, because he then went on a long tirade about how Shelley was done wrong. Another candidate then laid into him for supporting that weasel Shelley and the whole audience (I'm told) started booing at him. The guy who stood up for Shelley and election reform was the clear winner of the night's debate. This is quite the talk of the town today. People are finally connecting the dots.