Try It Again: Why ATMs are Not like Voting Machines

Here's the text from the link, in full. My introductory comments were previously posted at:

<http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x401778>

FYI: this incorporates a few of my ideas and the "no basis" for confidence frame started I think by GuvWorld, but this piece is not written solely by me and the author has chosen to not put a name on the piece:


WHY VOTING MACHINES ARE NOT LIKE ATMS

1. ATM software is open, but voting software is proprietary

Banks insist that all code in ATMs be fully disclosed to them and they won't trust their money or their depositors’ money with anything less. Voting software by comparison is considered proprietary by companies that make both ATMs and voting machines and proudly boast of their open source software for ATMs in advertising. This closed-source situation could conceivably be changed by demanding that voting software also be fully disclosed, but there are other reasons why open source code is not by itself sufficient to make voting machines like ATMs. For example, there would be the necessity of matching the code on all voting machines to verify identity with the true open source master code immediately prior to each election, but even then any diskette or other similar device can introduce a virus that deletes itself and human beings can not observe the vote counting even in open source environments.

Open source code is not necessarily "knowable". One can think of the law as being open source "code", free of copyright and at least in theory available to all in free libraries. However, like the extensive areas of code in computer programs that often have unknown functions or utility, even a lawyer who spends his life studying the law doesn't understand how every bit of the "open source" law works, nor can we the people realistically understand even a fraction of exactly how the open source code would work. Thus, we would be required to accept election results on trust or faith, which is the opposite of checks and balances.

Were the code of the voting machine vendors suddenly opened up or disclosed, it would take a long time to understand it, we may in fact never understand it, and those who do understand will only be a handful of experts with a lot of time on their hands, probably paid by the government or a vendor and not loyal solely to the public.

2. Individual ATM transactions can be tracked, but individual secret ballots cannot be tracked

Every transaction in an ATM is completely tracked with redundant account numbers traceable to the account holder, and your transaction is photographed or videotaped for security purposes. In contrast, a secret ballot cannot possibly be associated with such an identifying number and still remain secret. The very secrecy of the ballot creates a virtually untraceable system that is wide open to both fraud and the cover-up of material irregularities. It is not feasible to provide a receipt in elections to prove a transaction because of concerns about using it to sell votes, though this concern might be addressed by making verification available only to the voter in secure locations like the elections offices.

To make ATM banking perfectly analogous to the process of voting, you'd have to have every account holder at a bank make a non-traceable (secret ballot) cash deposit on the same day (election day) by dropping this anonymous deposit (ballot) into a large bin (ballot box). Bank officers would then calculate the total amount of money deposited in secret with no public oversight, but not start counting until after the bank (polls) close. It would then be claimed that the account holders (the voting public) only main concern would be to come back at the closing of the business day (election night) with the media in tow demanding instantly reliable bank balances and overall account results within minutes or hours of the closing of the bank (polls). Bankers (election officials) would insist along with some in the media that the convenience of speedy results was far more important than accuracy in one's bank account (election results).

The insane rush to count the bank deposits (ballots) within minutes or hours on election night would them be used as a primary argument for making the banking deposits invisible and unverifiable by converting them to mere electrons, so that they could be processed all the more quickly and conveniently. Hopefully you can see that in elections we are putting intense pressure on a very fragile and inherently unauditable system that can work, but only at deliberate, and visible, speed.

3. ATM errors typically have no consequence for users because of opportunities to correct them, but ballot tabulation errors have very serious consequences that are often not correctable

With banks, you have at least 60 days after receiving your statement, if not much longer, to contest and challenge the transactions involving your account. With voting, there is no possibility at all of correcting your vote after you leave the polling place. In fact, voters are considered legally incompetent to contest their ballots with extrinsic evidence under stringent anti-challenge provisions. Election contest laws are subject to extremely short statutes of limitation such as ten days. At any rate, you couldn't locate one's own specific ballot anyway for purposes of challenging it's tabulation, and some elections officials have preemptively cited academic research purporting to suggest that significant numbers of voters "don't accurately remember their own votes" after having voted, in order to cast doubt on any member of the public who may question the tabulation of their own vote. Thus, it is most likely that NOTHING WILL EVER BE ALLOWED TO IMPEACH OR CONTEST THE RUSHED COUNT, not even the voter themselves were they somehow able to show their own ballot counted incorrectly.

Broken voting machines have disenfranchised many, many people who have had to get back to work or school before a functioning one could be made available to them during limited voting hours. A broken ATM just means that you have to go to another bank branch or supermarket, at any hour of the day or night. WIth voting, the machines are expensive bottlenecks where you are usually forced to stay in a long line in the same polling place and can not go elsewhere and vote in the usual manner (but in some states a provisional ballot MAY be allowable).

In summary, you vote untraceably (assuming that you aren’t turned away unable to access a functioning machine, or by long lines), you're not allowed to challenge or change even your own vote, you're not trusted to remember it, and then the elections officials will refuse to disclose their data (ballots) or their analysis methods (counting software) on the grounds of trade secrecy, but will only release their conclusions (election results).

Such a system has absolutely none of the safeguards built into ATMs, which have quadruple redundancy. If you take out $100, you can count the five crisp $20s, check the receipt, cross-reference it with your bank statement listing individual transactions tagged with unique numbers, and if necessary, request the photo of you making the transaction.

4. ATMs have extensive real world testing that vote counting systems can never have

Principles of elementary systems analysis dictate that any complex system, whether mechanical or electronic, is highly unlikely to ever be free of bugs. Such systems can, however, eventually be made robust and reliable by banging them against reality hard and often. ATMs are part of a complex system that has had most of the bugs worked out of it by being constantly tested in the real world, billions of times an hour, 24/7, 365 days a year. Even so, they still malfunction occasionally, though if you run into one that isn’t working it’s usually a minor hassle to find another one.

In contrast, voting is something we do a couple of times a year, and letting machines with complex hardware and software do it for us must inevitably always be a beta test. This is why you rarely hear of ATMs that don’t work because of heat or cold or humidity, but commonly hear of voting machine breakdowns for those reasons and many others. If we only drove our cars for a couple of hours once a year, they'd suck pretty badly too. Beta test mode is absolutely unacceptable for something as important as voting. If you run into a voting machine that isn't working, it increases the length of the line you stand in, and you may never be counted.

Moreover, even the billions spent on ATMs would not allow us all to use the ATM in the same 14 hour voting day, even with long lines. Commiting our country to electronic voting means committing to standing in line instead of a 5 minute service guarantee we are used to in stores. The "promised land" of electronic voting promises convenience for election officials, inherent invisibility of mistakes which appeals to both vendors and election officials, and replicating the situation with school systems whereby rich districts get great service and poor districts get poor service. The ultimate in e-voting then is structural disenfracnhisement of the poor by bottlenecked expensive e-voting machines.

5. We can safely entrust others with tracking ATM transactions, but we can only trust ourselves to supervise vote tabulation

The current situation is this. We now have no basis for confidence in election results because data and the methods of its analysis are never disclosed, only conclusions (election results) are disclosed. Voters are considered legally incompetent to change or challenge their votes, or even to recall what those votes were. Voters are widely considered by elections officials to be the cause of the machine malfunctions themselves, resulting in delayed responses to fix the machines, especially since the pollworkers are not supposed to observe the voters and therefore can't verify that it in fact is a machine problem and not a voter-problem.

We need to fight for democracy here in our time, meaning that the government serves the public as the ultimate source of political power, and not the other way around. Government "servants" should not seek their own convenience and insultation from accountability for mistakes, but should instead be rewarded for falling on their swords and reporting problems voluntarily and immediately.

We the People must insist on vote counting methods that are transparent and public, that have robust checks and balances, and that keep fully in mind the very unique features of elections that make them not analogous to much of anything else. Thomas Jefferson anticipated every generation would need a revolution in democratic values to remember the inalienable rights of We the People and assert them against the government officials who (quite naturally and even understandably) believe their full time status entitles them to superior rights, because that is the route to something other than democracy, something other than We the People being in charge.

For more information see
http://www.votersunite.org/
http://www.ecotalk.org/VotingSecurity.htm

How undemocratic can you get?

(but still have nominal 'elections')

Our so-called "elections" are better described as simulated competition, like professional wrestling or the Harlem Globetrotters. The more closely they resemble actual elections, the easier it is to create inherent uncertainty about the outcome. That is, if we can't know for sure what the true results are, the public is bound to wind up in dispute. This creates a rift in the perception of reality. This is only one way that we see the government creating intentional divisiveness. Read Blueprint For Peaceful Revolution.

"guaranteed non-representation for 49.9% of the population" a politician has to straddle such a wide political area to form a majority that our leaders inherently have damaged integrity if they win. (By appealing to so many different subgroups, they lose internal consistency and over-promise) now in the age of video, those promises in tension can be placed side by side in video clips, revealing the "hypocrisy" of the politician, when they are just trying to cobble together a coalition.

On the other hand (otoh), in a more parliamentary system the general criticism is that with so many more parties, the coalition GOVERNMENTS may sometimes be less stable.

We have a choice: less stable coalitions, or less stable politicians. THe answer is not immediately obvious, but I prefer to stabilize the individuals and let the coalitions shift a bit more, rather than have politicians themselves be individually "shifty".

Requiring a society to work on keeping its coalitions together is the opposite of intentional divisiveness. It is a challenge for inclusivity and a recognition of the value of unity.

"Requiring a society to work on keeping its coalitions together is the opposite of intentional divisiveness. It is a challenge for inclusivity and a recognition of the value of unity."

the strong american tradition of individualism yet the relative unity that has always existed in times of crisis should reassure those who like to bring up the example of Italy as some kind of horrible place to be.

Oh, man....now reading Andrew Gumbel's "Steal This Vote"
which has a lot of history of elections processes in the US.
You won't believe the corruption which has been a part of the
system forever. Our mythologies regarding "the sanctity of the vote" seem built on the flimsiest of beliefs and evidence. Very, very sobering...
We must have paper ballots hand-counted twice, precinct and BoE, in public view. Or certified open-source software, protected to the highest security possible. Otherwise we're never gonna see "one person, one vote.'
And delete the Electoral College, which is long past its usefulness...convincing the smaller states to agree to support
the idea of a republic.
Isn't there an honest whistleblower in the election system vendor
world? How much a reward or bounty would it take to smoke one out? Jes' sayin'

can't either, open source or not. Open source MAY WELL BE ADOPTED but since it is only about 1 of 20 factors needed for a secure computerized system, which will never allow publicly observable elections because the electrons are invisible, it's adoption would be almost nothing in terms of progress because of all of the other co-factors that would have to come with it in terms of checks and balances.

And in the end, the Defense Dept. can't protect its computer system if its networked. And, as of a year ago, plugging a computer into an electrical outlet is NETWORKING IT. see homeplug.com

Febble wrote: There is a fundamental reason why the ATM model is completely inappropriate for voting, which is the principle of the secret ballot.

You couldn't run a banking system if the bank wasn't allowed to know which accounts belonged to which customer.

The whole idea of electronic voting is absurd.

Bill Bored wrote: ...but the main difference is with the ATM, both participants in the transaction (the customer and the bank) are known to each other and the customer gets a statement confirming the transaction at the end of the month that shows the money was deposited as cast, uh I mean counted as deposited, uh...well you know what I mean! No secret ballots with ATMs so you verify your vote, er...deposit!

by definition create a radically non-auditable system, because you are not allowed (by penalty of the law) to trace a ballot back to a particular voter to check to see if it is done correctly or not.

Thus, it is illegal to have an auditable and accountable ballot system. So the only solution is total observation and transparency with robust checks and balances OR getting rid of the secret ballot in some form. Most prefer the former.

One possible solution is to allow the voter to identify a unique PIN number which they can then check their ballot with privately, perhaps only at secure locations to avoid the problem of "proving" one's vote to a potential vote purchaser (a possible, but not huge, risk). This is worth considering, although there's still the risk of bad faith and what the remedy would be when someone discovers an improperly tabulated ballot. (i.e. what happens in response to the "So What?" test).

However, as I've said many times, don't expect a perfect system that doesn't exist, just make the system as secure as possible without making voting difficult or burdensome, and then structure it such that the payoff or reward to a cheater is as low as possible, such as one single vote. the single vote is the typical payoff in a paper ballot system with appropriate checks and balances. It's not that I like the "old technology" of paper, it's the fact that the paper provides the modern and incredibly slick advantages of transparency, obviousness, provability and robust checks and balances. It is thus the most ADVANCED system we have FOR DEMOCRACY. Technology may be "advanced", but it ain't advanced FOR DEMOCRACY, and that's where we are applying it.

and cheap means freely available.

Expensive resources are inevitably rationed, and when things are rationed, the poor get less.

My mantra is: voting should be as cheap and available as fresh air and clean water.

It's why we use pencil and paper in the UK, and hand-count the votes.

and don't require you to stand in an electronic bottleneck line.

I think you could kill e-voting entirely with a cost benefit study on what it would take to guarantee five minute or less service to every voter in a general election. There's just not enough dough for that, even if e-voting had no other problems!

here by an ex DU legal eagle:



Actually, I'd go one further: with paperless voting, three entities are merged.

In a paper ballot system you have three entities:

1. The ballot
2. The act of casting the ballot
3. The act of counting the ballot.

When you vote on a DRE all three entities are merged (I don't know how lever machines work - but is the same true of levers?)

And I don't know about legal rights, but certainly the ability to audit is lost when the three entities are merged. As long as the casting of a ballot involves a ballot that is validated by the act of being cast, the ballots can be recounted as many times as you want.

Once you lose the ballot as an entity, not all the checksums or paper receipts in the world will allow anyone to recount what was cast.

Anyhow: check out the lawsuit referred to in the link.

to be corrupted.

I'm sorry. I'm not a machine code programmer, but I can tell you that a voting program would NOT be that difficult to write.

I recall one story from 2004 saying that one type of these machines used a run-time version of FOXPRO. In other words, this ain't rocket science. A voting program would be an CIS UNDERGRADUATE PROJECT.

There is NO reason that instead of the various state gov'ts paying a LOT for these devices to the private companies, they instead pool their money and hire a HANDFUL (should it really take more?) of computer programmers to whip up a voting program of some kind whose source could then be MADE PUBLIC for ALL states to use, recommend changes, etc.

I'd LOVE the computer community, esp. Steve Gibson http://www.grc.com to take a LOOK at what was going on 'under the hood' and report to US.

Having this whole thing run by a handful of private companies, in total secret, is INSANE.

Being a computer consultant I can honestly say that the old punch cards (yes, the "chads") were much less risky than this wholesale "blackbox" solution...

it is nevertheless hard to secure. I tried to make that a little clearer in some edits just made above. The present code used in Diebold, I'm told, has huge regions that nobody understands what they do (from someone who got a peek at it, I'm told). this is from piecing together some oommercial software and building on other foundations, not all of the code for which is necessarily used.

Money and votes are different. I know how much money is in my account and I can challenge the bank on an error.

But with voting my vote and the votes of millions of other dump into a central pile. Then the results are announced. How do I tell if mine is even in there?

Because there is no connect between my vote and the final tally then safeguards and even MORE important than in banking.

I say, and hopefully others have all ready been down this road, that when you vote you get a vote number for that election. With that number you can go check the database, either printed or online how the election officials recorded your vote. Nobody knows how you voted, unless you gave them your number. The voter could validate his results. That database could be viewed analized by anyone concerned about the election results. Canidates could call on his/her voters to check there votes to see if someone has played with the results.

or probably in some other post is the under-appreciated need for (and under-made demand for) more and more DATA being disclosed EVEN UNDER PRESENT CONDITIONS. We are really being shorted on data that could be disclosed, even when only results are disclosed the data is left cryptic, quite often.

Economics as praticed by BushCo: is a return to a feudal society. A Feudal society cannot sustain a democracy.

Thusly open and transparent elections are more important than the Banking Industry and its ATMs.

Thank you, Land Shark. Wish you'd cross post to GD!

:toast:

has no place in a Democracy?

Great summary.

Thanks for this thread!

K & R

Interesting contrast there between saying one is welcome (could be interpreted that one is "invited" or "entitled") and saying "be good". Just like You're welcome people say the equivalent in finnish so often that they don't think too much about the literal meaning of what they're saying, though.

most of the exit poll discrepancy was calculated on precinct counts not tabulator counts. There are a number of other reasons to think that tabulator level hack is not compatible with the exit poll data. My hunch is that if electronic is lurking in the data, it's at machine level.

move....provided nobody "gets to" the ballots prior to the official recount. And,even if they get to the ballots prior to recount, an accurate exit poll would show a discrepancy. Course, that may be disputable as well so long as we have reluctant Bush responders and the smattering of circumstances to support that inference/argument.

is whether anyone can/does check that the precinct totals at the end of the day match the county tabulation for that precinct. I have to confess that your system is so bizarre that I can't quite imagine how there could ever be any discrepancy that wasn't caught, or at least catchable. Although I presume, from the the fact that a proportion of the WPE numbers were apparently based on county tabulations that for those precincts an actual precinct total wasn't available. It certainly strikes me that it would be incredibly useful for partisan precinct watchers to do this kind of check (I think autorank described doing it in VA for the recent elections). And I think Bill Bored called it "canvassing" (which means something quite different in UK English).

But one reason why I was initially skeptical of (although attracted by) the notion of a mass hack was that it seemed to me it was only seriously executable at tabulation level, and that's were it doesn't seem (to me) to have happened. There could have been plenty of interference with precinct machines but the precise mechanism would have to be different for different machine types, I would have thought, or else hacked individually if wired (or wireless). I dunno. The more I think about it, the more I difficult I find it to get my head round the way it could have been done (especially given those damn non-correlations). Though it seems clear that move to DREs would make it more possible, not less.

But as far as 2004 goes, it seems to me more likely to have been opportunistic, and relatively small scale, though if targetted right, it could have helped swing Ohio. On the other hand there are plenty of other non-hack methods that could (and certainly did) cost Kerry votes.

whether anyone can/does check that the precinct totals at the end of the day match the county tabulation for that precinct.

We match the tape tabulations for each precinct to the voter sign-in sheets. At the polling place I worked at, we were off by 4 extra votes out of about 1500 cast, which we attributed to a couple of provisional ballots being successfully put through the tabulator. I understand that this result was typical in 2004. If the sign-in tally is higher, it is usually because a few people sign in at the wrong precinct and don't delete their original signatures after signing in at the correct one.

We had a total hand recount of the governor's race, in which votes were added to the totals of all three candidates. The hand count was within 0.01% of the two machine counts. This is what we should expect with optical scanning, which tends to a slight undercount for the same reason that your computer printer sometimes picks up an extra sheet of paper.

Unfortunately, there was no recount in the DRE using Snohomish or Yakima Counties of the poll votes.

so whether or not there is precinct tabulation at all depends both on the technology and the jurisdiction. Individual counties vary, so the existence of precinct level tabulation with public posting or without varies enormously.

There is however a typical pattern when it comes to DREs, most of which print results totals at the end of the election day.

For example though, comparing precinct level printouts to precinct reported totals in snohomish county reveals "phantom" votes, meaning there were, say, approximately 930 votes on election night for all machines where I was personally at (according to official reports of all pollworkers as well as my personal notes), but then when the 5 precincts for that polling location are reported and added up there are 931 votes, adding one extra phantom vote. where did this vote come from???? It had to have come from "outside" the polling location, or somehow be added electronically after the printouts at end of election day.

Now, the whole notion eridani identified above as an explanation for discrepancies between signatures and votes that somebody must have signed and forgotten to delete their name, or not signed, is plausable at the level of 1500 voters. But it is not necessarily the explanation. you see, there are these checks and balances but the fact is that when there are both an innocent and a guilty explanation the innocent one is readily accepted and nothing more is thought about it. This is also why touch screen calibration fraud is brilliant: "reasonable" people will all ascribe it to mechanical difficulty or voter incompetence even though it can also be deliberate misprogramming or miscalibration and look all teh same.

Manjoo, speak of the devil, CASTIGATES mark crispin miller for providing some "easily disputable" fraud methods. Well, Manjoo, these are the VERY BEST KIND if you can do them. And i fully recognize that some miscalibration is truly accidental and on the surface it is indistinguishable from non-accidental miscalibration.

My thought here is that the strongly held notion that an "easily disputable" situation will be blown off by "responsible" people like manjoo and election officials makes this the perfect potential crime.

Bottom line is that anything goes, in elections. When we design systems we have to design them such that any discrepancies are in fact red flags and not readily dismissable. Note that people like Manjoo and politicians generally are very attuned to the demands of political advertising and politics generally in which "incontrovertability" is a major value, i.e. what one says should not be readily debatable or turned around. lawyers like to be in a strong position too. But this will also mean that a really smart criminal, by doing ambiguous things likely to be interpreted innocently by the relative Pollyannas and elections officials of this world who certainly don't want publicity and headaches, will have a free hand with our elections, acting essentially unmolested.

Remember the radical nonauditability of the election systems (all of them) based on the universal secrecy of the ballot and you are virtually guaranteed easy disputability at many if not all, turns along the way. This constitutes real serious protection for election criminals, and Land Shark's evil twin is truly gratified by this situation. It's quite heartwarming when he thinks about it.

I completely agree with you about the plausible deniability thing (raised by Yowzayowzayowza a while back.

It's just that it's hard to find 3 million votes that way. Enough to swing the electoral college I will certainly grant as a possibility.

And whatever you think of Manjoo's review, I don't think your assertions as to his motivations are particularly well-supported (speaking as one who has frequently had her motivations impugned). He DID acknowledge that he had been credited in MCM's book, in the same sentence as his second cite of DU, although he did not cite the page you required of him. However, as a declaration of interest, it is there.

So the division looks to me as though it is over the magnitude of the theft, not whether it occurred. Your evil twin may be gratified by escaping detection, but if so, I don't think he stole the popular vote. And he certainly didn't escape either Manjoo's or my own suspicions that he was up to no good.

what is there about deniability that keeps the theft under 3 million votes nationally? (nothing). You may be presuming the specific method outlined in an earlier post but i am raising the post more generally.

The mystery of Manjoo: talk about impugning someone, I'd much rather be accused of being afraid of backlash and being fearful than to be an (often) ostracized conspiracy theorist not worth taking in any way seriously even as a voice in the debate. That's approaching if not consituting outright dehumanization. Fear is at least human.

But the real mystery of Manjoo: his prior to the election writing saying an election can be stolen and "nobody would know it" with his present denials that anything happened. Or at least that there is any evidence of any thing happening, and he puts it in conjuction with the "election not stolen" formulation that he is effectively denying his prior statements that an election could be cleanly stolen.

we'll agree to differ on Manjoo.

If you can suggest how 3 million votes could have been stolen by methods that would be plausibly deniable and not produce that swing-shift correlation, then I'd (genuinely) be happy to hear it. Voter suppression would be the best bet. But vote tampering on the scale required, I just can't see. And I don't simply mean that in my Pollyannaish fantasy I don't believe it. I mean that when I mock such a thing up in a spreadsheet it doesn't compute. But I'll keep trying.

For single state, or even a couple, I'd be more easily convinced.

either accidentally or know about in order to mask the effects? It would be less than 100% I'm sure. So that's method #1 which I guess you don't accept.

Then if you point me to your most complete definition of "swing-shift" correlation I will find another way around it for you.

you have to avoid any precinct where Bush does well relative to 2000 anyway, or it will stick out like a sore thumb.

So you can't risk even doing this accidentally. Three possibilities strike me:

1. You gauge the political temperature in each precinct beforehand (which is going to be tricky, as I don't suggest you trust opinion polls).

2. You use some kind of algorithm in the machines that prevents Bush's vote share dropping below a certain percentage. However, this had better either be reversible (in case you get a glut of Bush voters at the end of the day) or kick in at the end of day tally (as long as no-one's been able to keep an eye on the counters as the day goes by). You can't leave it to the county tabulators unless the magnitude of your fraud is very small as a) someone might notice the discrepancy between precint count and tabulation or b) it will still show up in all NEP precincts where the WPE was calculated on the precinct count.

3. You could try avoiding NEP precincts - but this tend to show up when E-M tests to see whether their polled precincts were representative of the total precincts. If you confine substantial fraud to the non-NEP precincts, they won't be.

In addition, you will have to get very good coverage (unless you avoid NEP precincts) - if you leave out more than, say 30% of precincts*, you are going to be in trouble with the swing-shift correlation again. This means that you have to apply your algorithm to all kinds of voting technology (I'll let you off paper ballots). You will also have to cover all states. Don't leave any out.

If you decide to keep vote switching to a minimum (try it out in a few precincts and see if anyone notices, but I expect they will), then what you could do instead is to simply invalidate (by registering them as over votes or undervotes) a few votes in a large number of predominantly Democratic precincts (make sure you randomly destroy both Bush and Kerry votes if you can). This won't affect the exit polls at all, so it won't show up as redshift in the WPE or as a swing-shift correlation. And if you keep the numbers small, it will be plausibly deniable. You are best placed to estimate how many votes you would have to destroy in how many precincts to get the effect you want. The more pro-Kerry the precinct, the easier the strategy will be too pull off (because most of the invalidated votes will be Kerry votes). And as Democratic precincts tend to have higher residuals anyway, you can probably get a way with quite a bit of pro-Bush fraud this way. And of course, by making sure ethnic minority voters are disproportionately given provisional ballots, or by rationing the number of voting machines in Democratic precincts (say you are basing it on past turnout - that will be an excellent piece of plausible deniability, as you can even throw in the odd extra machine on that basis and point to how generous you were) then you can give Bush quite a boost.

Oh, and if you simply up the spoilage rate of Kerry votes in precincts with historically high rates of Democratic spoilage, it'll show up as redshift, but it won't show up in the swing-shift correlation because you've been doing it for years.

And of course you continue to do, as you do every election, to discourage likely Democratic voters from turning up at all, either by messing with their registration, or telling them to turn up on a different day, or threatening them with jail if they have an outstanding parking ticket, or whatever. Let your imagination rip here - it's all free money.

The problem you are going to have, however, is stacking up the numbers in these non-voteswitching ways. Vote switching is extremely lucrative in terms of vote, because one vote-switch both depletes Kerry's column and increases Bush's. All the other methods only deplete, and some deplete both columns, and only benefit Bush by depleting Kerry's at a higher rate.

So, my (retrospective) recommendation to your evil twin is:

Limit the voteswitching to small amounts per precinct, but make it as widespread as you can. Suggest you do it at tabulation level, which will be a lot easier as you don't have to program each machine, but keep the amount very small and very uniform. You need to target around 75% of all precincts where Bush is doing worse than his average relative to 2000. Remember to cover all states, including those without DREs.

Do as much by voter suppression methods as you can. Make sure spoilage rates remain high on the old machines where they are allocated to democratic precincts. In punchcard precincts, make the ballot designs are as confusing as possible, and in multi-precinct polling places, make sure that the ballots can't be matched to machine, and make sure that there is maximum chance that voters will cast their votes on the wrong machine. As long as Bush is trailing this will give you a net advantage (regression to the mean). Issue provisional ballots to anyone who looks remotely like a Kerry voter, and reject it if you can.

Have a shot at switching the DRE defaults to Bush. People will probably notice, but if you don't overdo it, poll workers will just blame the machines. But remember - if you overdo it, it will show up as a swing-shift correlation. Although you might want to give one state a more substantial try, though, as with fifty states, chance alone is likely produce a "significant" swing-shift correlation in a couple of states. One more won't strike anyone as odd. Suggest you try a smallish swing state - maybe NM?


*acknowledgements to my own evil twin.

I wrote:

2. You use some kind of algorithm in the machines that prevents Bush's vote share dropping below a certain percentage. However, this had better either be reversible (in case you get a glut of Bush voters at the end of the day) or kick in at the end of day tally (as long as no-one's been able to keep an eye on the counters as the day goes by). You can't leave it to the county tabulators unless the magnitude of your fraud is very small as a) someone might notice the discrepancy between precint count and tabulation or b) it will still show up in all NEP precincts where the WPE was calculated on the precinct count.

I didn't mean that last b) (I simply forgot which argument I was addressing - this evil twin thing messes with my head....). It won't of course show up in the exit poll if you do it at tabulator level (except where WPE is calculated on the tabulator count). What I meant was that the WPE won't be evidence of it. But that makes it a good move, provided you can get away with it. Just be careful you don't argue that the exit polls support the case that it happened.

The short answer:

In places that use op-scan CENTRAL count (at least as common as precinct-count, from what I've seen-check out the "verifier map" at verifiedvoting.org) there is no independent precinct tabulation. The votes are all counted at the central location. No way you'd see discrepancies between precinct totals and hacked county totals-it's all the same.

And assuming kiwi is right about punch-cards (used overwhelmingly in Ohio in '04-see link in my #45), there would be no way to verify county vs. precinct totals throughout the great majority of the state's counties.

that's helpful.

As far as the exit poll discrepancy goes, apparently about 40% of NEP precincts didn't have precinct totals. The remainder of the precinct discrepancies were calculated on precinct totals collected on election night.

by voting system? Thanks.

by voting system within the E-M data. Information is available in the E-M report:

of the 1250 precincts analysed at precinct level,

40 had paper ballots
118 used mechanical voting machines
360 used touch-screens (presumably this includes push button DREs)
158 used punchcards
573 used optical scan.

http://www.exit-poll.net/election-night/EvaluationJan192005.pdf
page 40.

However, they are not cross-tabulated by state.

we can not rule out tabulator fraud for that reason? (if you accept the premises of my prior posts in that area where precinct totals are a check and balance against tabulator fraud)

Of course, tabulator fraud will be exit poll detectable in most manifestations. But not (for example) where percentages are kept the same and the pie increased, though then there will be disconnect with the poll books at the precinct level.

for around 40% of precincts the WPE was calculated on county tabulations not precinct totals. So yes, tabulator fraud could be the cause of WPE in these precincts. But if so, it should induce a swing-shift correlation.

"most of the exit poll discrepancy was calculated on precinct counts not tabulator counts" -Febble

Are some precincts' CANDIDATE votes counted once at the precinct location and then again on county tabulators?

It has been my understanding that, in Ohio punch-card counties at least, candidate votes are counted in ONE location: precinct by precinct in the county BoE offices. The counting is done by county tabulators. The only "counting" that is done at the precinct locations is of the TOTAL number of ballots cast, not the votes for individual candidates.

I don't know about the op-scan counties. But if the op-scan candidate votes are counted at the actual precinct locations, would they be counted AGAIN on county tabulators? If so, why - and how?

there, and then a "tabulation" which is a mere addition at the countywide level.

This way, any citizen can reproduce the work of the tabulating computer if they so choose to do so and the work of fraud is forced to be very diverse and widespread instead of concentrated at the focus point of the tabulator computer.

The procedures actually followed vary from state to state and county to county.

Did each of the 88 Ohio counties have a similar final (hackable) "tabulator" run that did nothing but add the county's precinct candidate totals together to get county-wide candidate totals?

For example, would the precinct candidate TOTALS from a punch-card county (i.e., the precinct candidate totals which were computed using card readers in the county BoE offices) then be added together in the separate (hackable) "tabulator" run for the county? Likewise for summing the precinct candidate TOTALS in the op-scan counties?


If precinct candidate totals were recorded prior to the final county-wide summing tabulator run, it is hard to imagine how a hack of the final run could be expected to succeed.

There are two types of optical scan systems, precinct-count vs.central count. In central-count systems, the votes are not even counted at the precinct.

From New Yorkers for Verified Voting:



See here for a breakdown of Ohio's voting equipment in 2004. Most of the state used punch-cards (which you say are counted only at the central location), and a smattering used op-scan central count (in maroon).

See also Pennsylvania, Minnesota, and Iowa for examples of the widespread use of central count systems.

Probably easily verified, but do you have a source for the one-site counting of punch-card ballots?

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=354121#354997

post 28:

"...#25 - SFAIK from my meeting at BoE on 3/29, the procedure is this: all precinct ballots stay together from poll to BoE, in a tamper-evident pouch. At BoE, they are inspected for chads, damage, etc., then provisionals (in separate envelope in pouch) are added, then absentees are added. These have been filed previously by precinct (precinct is identified on back via rubber stamp and hand entry by staff before being mailed to voter.) This is so voter can verify that they received correct ballot. Then a precinct ID header card is added to stack. Then they are run through the tabulator. The digital file output is transmitted by wire to their in-house compiler / computer...." - liam_laddie

I see that there was one Ohio op-scan PRECINCT-count county (Allen).

And there was only one touchscreen county in Ohio - but there were 6 "Electronic-voting other" counties. Would those be button machines? Do you know if they would have had a paper-trail?

Cheers!

Franklin ("e-voting: other") used Danaher Controls ELECTronic 1242, which involves touch-screen selections, followed by push-button vote. No paper receipt. As used in the state of Delaware:
http://www.state.de.us/doe_ncc/Use_VM/VM_poster.pdf


The system used in Lake county looks as though it might be quite immune to fraud:
http://www.iwantmyvote.com/recount/ohio_reports/counties/lake.php

kiwi - Triad (Systems?) out of Xenia, Ohio, managed, or ran, almost half of OH counties in Nov 2004. Not sure if they had their own brand machines, or used others' hard and software.
Have not heard anything about them in several months. But I believe they had remote access capability, and seems some of
their techies were lurking about both during election and recount.
Ohio is totally corrupted, IMHO.

documentation or followup on Fitrakis' verbal statements in June that Blackwell (and others) had access to the central tabulator in Ohio during the Nov 2004 election?

For example:
http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x378842
thread title: Fitrakis: Blackwell had direct access to central tabulator!

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=377409
thread title: HEADS UP: Bob Fitrakis on 'The Brad Show' RIGHT NOW

ATM source code is open, imagine that. Maybe the banks think that OUR money is too damn important to put in the exclusive control of Diebold.

Well, are not our votes to damn important to put in the exclusive hands of Diebold?

Open the code, open the machines, full impartial auditing from top to bottom...and start with what ever the hell recorded the stunning 40% spread in pre election polls and election results in Ohio's special election.

Enough already. No more bull shit from right wing Republican voting machine companies.

to pass on the fact that closed machine voting is a terrible idea.

So much about election fraud can be confusing, but this is easily grasped. People use ATMs regularly and relate to the analogy.

Recommended.