I agree. For those who believe in category 2, the inevitable conclusion to make: there is no merit to debate anything which has political implication. Evidence or conclusions based on the evidence will be tainted by political bias to the point where no evidence can be considered meritorious by an opposing viewpoint. In other words, the debate will aways boil down to: "I don't accept the evidence because the person collecting the evidence self-identifies as a member of political group X (or I identify them as a member of group X)." Another way of putting this: there are no facts; just agendas. Or perhaps more exactly, there are facts as long as they conform to my political bias. I suppose this segues nicely into Hayleyanne's discussion of post-modernism and relativism. Ironically, however, it appears to be those self-identifying with a conservative political outlook who are following this doctrine.

To backtrack a bit...

The vote count is brought into doubt because of it's disparity with the exit poll projections... and then considering that fallibility of the vote count, the exit polls projections thus give us reason to doubt the vote count? This reads rather circular to me. Anyone else?



Because of the realities of politics, I don't think any investigation with hopes of affecting the 2004 election results would either manage to get rid of bush OR fully investigate election procedures... And I think it would be horrible at this point for such an investigation into election procedures to fail because they're a real problem. So presently, I couldn't support any investigation with retroactive effect as it's goal because a failure could would set such bad precedent for such a critical issue. And I'm not saying this as a conservative... I didn't vote for Bush, and I was greatly disappointed to see him be elected. Though, not quite disappointed enough to regret not voting for Kerry.

No, the vote count is always subject to a small degree of doubt. If the exit polls agree with the vote count, then we have greater confidence in the vote count. If the exit polls disagree with the vote count, then we have less confidence in either one.

For example, suppose that the exit polls showed Kerry winning by a landslide, and the vote counts show Bush winning by a landslide. Wouldn't such a result cause you to wonder where the problem was? Simply dismissing the exit polls out of hand is foolish. When there's a serious discrepancy, we should investigate to see where the problem is.



One thing at a time. If -- IF -- an investigation turns up evidence of electoral fraud, then we can take up the question of what we do about it. At this point, such a thing is speculation, so there's no point in discussing it just yet. I am not calling for a lynching, nor even a conclusion that the election was fraudulently manipulated. I am calling for an investigation into the possibility that the election was fraudulently manipulated.

On the contrary, I take this issue extremely seriously. Election fraud on the federal level is an enormously serious charge. Such charges when leveled should be considered from all aspects and that includes the possible political motivation of those leveling the charge. Yet, when the possibility of bias on the part of those who performed or published the study is raised, their argument is discounted as an ad hom attack. Look at the opening paragraph in this thread.....



"Sober"? What does that mean? Does it mean perhaps unbiased or completely objective? Are we, the great unwashed conservative masses not to question this? Apparently so, at least in this thread as the following exchange indicates.....





"Reprehensible"? There's a rational argument for you. But, Sleeper does his homework and comes up with the following from one of these "sober" folks cited by Erasmussimo.......



To which we get the following "rational" response......



The lady admited she was biased for goodness sakes! Quoting her own words is now a personal attack?

So, let's go back to your words, Logo......



I am suggesting that you are wrong. This debate isn't moving forward because anytime someone dares to suggest that these serious charges of election fraud might possibly be politically motivated we hear screeches of "ad hom" from those who disagree. I think it is possible to generate evidence and make conclusions based on that evidence independant of political bias. I also think that was not the case here. One of the researchers cited here has already admitted bias in her analysis.



Oh puleezzee..... Watch out for that high ground you're standing on, Logo. It's about as stable as the hills in Laguana Beach......

Aquilla, the long argument you present that ends with these words:

is founded upon a complete misconception of the notion of the term ad hominem. An ad hominem argument is not a personal attack. It is a rejection of an argument by referring "to the man", not to the evidence. It is usually taken in the negative sense, as in "your argument is wrong because the person providing the positive evidence is a bad person"; however, it is occasionally, for reasons of symmetry, applied to the reverse type of argument, as in "your reasoning is wrong because the person providing the negative evidence is a good person". Both such arguments evade the actual evidence in favor of arguing about the character of some person. That is what you are doing here. It is illogical.


And that is because you insist upon relying on ad hominem arguments. The evidence speaks for itself; I am not asking you to trust anybody nor do I tolerate your attacking the people in order to evade addressing the evidence. If you have questions about the evidence or the analysis, it's all out there in the open for you to examine. My case does not rely on the probity of the individuals you attack; it relies on the evidence itself. If you can't address the evidence, then you have no case.


Then why don't you do that? Examine the evidence; don't evade the evidence by talking about people.


No, you distort her words. Here's what she wrote:

Yes, she was biased in favor of finding one result -- but her logic led her to the opposite result. She is saying that her analysis was not affected by her bias. You have reversed the meaning of her words.

For the umpteenth time, I ask you to address the evidence. You have not yet done so.

So attacking the person (your words) isn't the same thing as a personal attack? And you want to lecture me on "logic"? Spock's pointy ears are spinning....



In another thread you started you asked people to consider the impeachment of the President because of a single word in a Downing Street memo - the word "fix". He "fixed" the intelligence data to suit his forgone conclusion was your claim there. For this you call for Congress to consider overturning the vote of the American people. Obviously there was a dispute to your claims there.

Now you come here and allege election fraud in the 2004 election that placed the President in office in the first place and ask us to accept the "evidence" presented as "truth" without regard to the possible bias and agenda of those presenting the data in the first place. These people are "sober" (your words). Even when one of them admits they were biased.

So one wonders why you demand that we accept the exit poll study data without question when you are willing to consider the impeachment of the President of the United States based on a single word concerning data. Is there an agenda kinda thing going on here perhaps? Inquiring minds want to know......

Your previous usage was BOTH an ad hominem argument AND a personal attack. It is certainly possible to combine the two, but using one to infer the other, as you did, is not logical.


If you wish to debate that point, please do it in the appropriate topic.


Indeed I am. If the evidence is wrong or the analysis flawed, you should be able to find such flaws. The fact that you have failed to do so suggests that you can find no flaws in the evidence or the analysis, and therefore resort to illogical ad hominem arguments.


"Even" one? One entire, complete person out of hundreds who have worked on this? You are extrapolating from a single case to the entire set of researchers? This is not logical.


By all means, question the evidence if you can -- but you have refused to do so. And your problems understanding the use of the term "fix" does not demonstrate that I lack intellectual integrity.


I follow the evidence. If the evidence leads to the conclusion that the election was fraudulent, then that's where I go. If the evidence leads to the conclusion that there was no fraud, then that's where I go. Are you willing to do the same?

I guess I lied. I have something more to add...

Erasmussimo I assume you have heard of Occam's Razor?

Occam's Razor states that one should make no more assumptions than needed. When multiple explanations are available for a phenomenon, the simplest version is preferred. A charred tree on the ground could be caused by a landing alien ship or a lightning strike. According to Occam's Razor, the lightning strike is the preferred explanation as it requires the fewest assumptions.

So you tell me which is more likely. The national election took place and there was a clear winner in the actually polling, but the exit polling which only used a very limited sample came up with a different result. This different result was in error because of bias or poor polling methodology.

OR....

The election was stolen by using malicious code in voting machines and tens of thousands of people were silenced from coming forward,(because an election of the magnitude will have to have so many people working on it).. and the exit polls caught this variance and exposed the truth, even though only using a random sample of the whole.

According to Occam's Razor, which is more likely?

Actually, Occam's Razor requires the inclusion of all relevant data. If you see a charred tree, and wish to choose between lightning and flying saucer hypothesis, you should choose the simpler lightning thesis in the absence of any other evidence. If there is additional evidence, you should take that into consideration as well. Thus, if you see three deep indentations in the ground near the tree, precisely spaced at 120 degrees, equal in size, shape and depth, then you might take the flying saucer hypothesis more seriously and investigate (just as I am suggesting). If further investigation shows the presence of a high concentration of short-lived radioactive isotopes, then you have even more evidence.

The problem we face here is that the evidence falls outside the intellectual ken of most people. When I learned that somebody had found a hook inside the ROM of the vote counting machines, I instantly said, "That's it!" -- because I know something about programming and I know what hooks are used for. Somebody with no knowledge of programming would not have the same confidence that I have. The same thing applies to the statistics. Because people are not familiar with statistical evidence, they can't see just how convincing all this evidence is.

If this is the case. Then why is not one of the major news networks all over this. Or do none of any of the Networks or Newspapers investigative journalists know what you do. If I were you I would run not walk to my nearest newspaper or television station.

This would be even bigger than watergate.

I think we can both agree that election fraud is a serious charge. However, I believe you are being disingenuous about the ad hominem issue. If you have evidence that the person doing the study has fixed her/his statistics or in some way has purposefully biased the data to generate a conclusion or even just plain mis-concluded, then please, please, please cite it. Just stating that, because person X may self-identify for a given political philosophy and therefore the data/conclusion must be inherently biased, is by definition an ad hominem argument. You need evidence not innuendo to make your case.


The only thing I am saying is that the allegation of bias alone (even self-admitted) is insufficient justification for discounting either the data or conclusions based on the data. If the evidence is wrong or bad or the conclusions are incorrect, then argue it. Do not say that because person X identifies with something you don't like, person X's data/conclusions are therefore wrong as well.


I wasn't aware I was standing on moral high ground . I was merely making an observation that relativism seems to crop up in many different threads.

~~~~~~~

I was going comment on the Occam's Razor argument that Sleeper proposed but Erasmussimo beat me to the punch. So, to restate it: given two (or more) choices where the weight of evidence is equal or absent pick the simplest one. Given that we're discussing evidence here of possible problems, Occam's Razor doesn't apply.

You'll have to ask them why they're not carrying this story. My guess is that they suffer from the same problem I cited above: an inability to appreciate just how strong the evidence really is. Perhaps when the technical report on the opscan machines comes in, they'll take notice.

I disagree with both points here. First, it is because I know about programming that, while I will join you in investigating the real source of this 'hook', and seeing why it is there and if it can/should be removed, I also understand that having the hook, and using it to simultaneously manipulate thousands of voting machines are two completely different things. One in no way indicates that the other happened. Second, it is because I am familiar with statistical evidence that I see that the paper cited didn't demonstrate electoral fraud at all. It merely used a set of simulated data to show that one 6% shift could be simulated by another to get the same results. Nothing in that paper even set up vote manipulation in the hypotheses...it wasn't until the conclusions were written that it was inferred. In short, they used good methodology to arrive at faulty conclustions. If the paper truly supported the conclusions drawn, they would have stated it in a hypothesis, and tested it. The evidence wouldn't support their conclusions, so they omitted the hypothesis. You yourself have stated that there are multiple potential reasons for the discrepancy between the exit polls and the vote count.



There is lots of evidence indicating issues with exit polls. None of them have been addressed in this thread. Until they are...their simply isn't sufficient reason to assume vote count fraud when faulty exit poll results suffices to explain the discrepancy. How many times in the United States have exit poll results been compared to actual votes (not vote counts, but votes)? The answer is never...since it not possible to trace the actual votes submitted by those who took the exit poll. All that can be verified is how they say they voted on the exit poll. Most people will probably indicate that truthfully. However, is there any evidence out there that contradicts the following hypotheses (which also fully explains the discrepancy, and indeed would be backed up using the very same methodology employed in the paper you cite):

H1: The discrepancy between exit poll data and actual vote count is due to people misrepresenting how they actually voted in the exit poll.

Occam's Razor would, IMHO, indicate that this hypothesis would be preferred over the extreme complexities and scale of assuming massive vote fraud.

Finally, I find your statements regarding the intellectual inability of others to comprehend this issue condescending, to say the least. If this were truly the case, there is no reason to carry on this debate, as that view would state that the other side is simply intellectually incapable of understanding the issue. No amount of debate would resolve this problem...therefore it would be irrational to even try.

Apologies in advance for a legthy quote excerpt, but I note that, in a prior post, I actually dug into the original subject of this thread and did not find my criticisms refuted.


I then noted:


And the goalposts began to twich. Or at least some progress was made toward ad infinitum. The thread about the Mitovsky data was moved into a "smoking gun" of a Diebolt ROM hook (courtesy of the sober Bev Harris) and then into the unimpeachable vouching for her study by Ion Sancho. I won't analyze any more statistics as I revert to my original conclusion. No matter what the evidence, or lack of it, or disproof, or whatever, there will be constant emergence of "proof" as to election fraud. See JFK assasination for details.

I do not mean this to sound simplistic, as clearly there are posters here with expert knowledge in the field. As someone who has done just a smidge of graduate work in statistics and uses them in my work - if it can't be explained in a manner clear enough to interest rabidly anti-Bush members of the US media, it either didn't happen or my theory on crying wolf has come to fruition. In which case the fraud proponents have brought their lack of credibility upon themselves, and done all of us a disservice.

I agree completely. The existence of the hook doesn't prove fraud. It proves criminal intent to commit fraud on somebody's part. And we both appear to agree that further investigation into this hook is called for.


The paper did not set out to prove fraud -- it set out to disprove the hypothesis that the discrepancy between the exit poll data and the vote count can be explained by the "reluctant Bush responsdent" hypothesis. It accomplished this fairly well, although such issues as the Febble Function do raise some questions.


Again, we have to focus on the particulars. Sure, it's easy to claim that exit polls are wrong. But why and how did they go wrong? So far, the only hypothesis that Edison/Mitofsky supports is the rBr hypothesis, and the USCV study casts serious doubt on that hypothesis.


I have been careful not to single out any individual on this point (although in a few cases it has been difficult). Instead, I cast my aspersions broadly, at the media, at a group, and in such a way as to permit individuals such as yourself to fall outside the purview of my comments. Surely you will agree that there are few who can appreciate the enormous signficance of a hook?


Once again we see the statement, "Don't bother me with evidence, I've already made up my mind."

Let me remind you that I am not claiming proof of electoral fraud. I am claiming that we now have evidence sufficient to justify further investigation. In order to support your opposition to this claim, you must show that the existing evidence has no substance. I don't have to prove that there's electoral fraud to support my claim -- you must, in order to support your claim, prove that the existing evidence does not justify suspicion. But you won't even look at the evidence!

Are you going to comment on my couple of criticisms of the stats above, which I've now posted twice? Maybe I am off base, but they seem to impeach the study which launched this thread.

Let's see if we can clarify this further....



We do both agree this should be investigated...I am not ready to claim it proves criminal intent to commit fraud. Lack of indictments would seem to back my claim up. It is entirely possibly this has a routine funtion....a 'backdoor' to allow machine reprogramming, for instance. However, it could also have other purposes. However, in either case an investigation seems warranted.



Agreed. However, their conclusions exceed this limitation.



This is true. However, there are other potential explanation besides vote manipulation. Several have pointed out in this thread that these other reasons are more plausible than vote fraud. Wouldn't it then be logical to investigate those first?



Actually, I think most people would readily grasp that. To the point where its significance could be overblown, as the remaining difficulties in implementing it would not be discussed. Further, though, I don't think that the issue itself is really germane to the topic. As I indicated in my previous post, it removes the need for debate, which is self-defeating in this forum. If there is indeed a 'smoking gun' or some other evidence worthy of consideration, it should be possible to explain its significance even to those unfamiliar with the complexities of the issue.



This is misplacing the need for proof, I think....the status quo is always the default choice. In this case, that is no investigation...one must show convincing evidence that one is needed, while others get the luxury of sitting back and casting aspersions upon it.

However, even if the need for further investigation were granted...there remains the question of what should be investigated. It would seem that many on this thread would say that would still be the exit polls themselves....I think it would be difficult at this time to counter that reasoning.



Ahh..but I did. In doing so, I am not even convinced they refuted the E/M hypotheses...it seemed more that they offered a counter proposal that seemed to fit the data better. I didn't see any statistics indicating that this better fit was outside of any confidence interval expectations....so there really couldn't be any conclusions drawn as to the statistical superiority of one method over the other. Even if their refutation were granted, though, we'd be a long way even from having that indicate an investigation into electoral fraud. Rather, I would think that would mean that other explanations for the discrepancy would need to be proposed, and tested. The one I propose I don't think even could be tested, although it is still plausible...that being that some people simply lied to the pollsters. Without access to actual votes (which would violate election law)...there really isn't any way to disprove that hypothesis. One could certainly turn the tables, and caste aspersions at it...it would eventually be left to judge which of the remaining hypotheses (vote fraud being one of them) seems the most worthy of investigation.

Yes, the Febble Function does cast doubt on the exit poll analysis by USCV. I don't think it impeaches the study, because it does not arrive at any quantitative estimate. Rather, it demonstrates that there remains an unknown element that could -- could -- serve to disprove the USCV disproof of the E/M rBr hypothesis. It depends on the magnitude of the Febble effect. She has robbed the USCV study of logical compulsion, but she has not crippled it. Without the Febble Function, the USCV study constitutes solid disproof of the rBr hypothesis. With the Febble Function, that disproof is no longer absolute -- but it still has significance.

I want to emphasize: the exit poll analysis does not constitute proof of electoral fraud; it does not even prove that the exit polls were right and the vote counts were wrong. All it does at this point is demonstrate that there's something fishy in the discrepancy between the exit polls and the vote counts. Febble's Fancy Function enfeebles the fishiness demonstrated by the USCV study, but does not eliminate it. If the USCV study were all we had to go on, I would recommend that further investigation be confined to more thorough academic studies, and especially the release of the E/M raw data -- but nothing more at this point.



You're willing to place your trust in a prosecutor who surely knows less about programming matters than you or I? Yes, there are some rather exotic scenarios that could make this hook innocent -- but they are pretty exotic.


Yes, backdoors are useful for onsite debugging and such, but for secure applications backdoors are a serious no-no.


Yes, we should start off with the easiest, cheapest-to-check-out hypotheses. I sincerely hope that we find something innocent at the bottom of this. Maybe it will be some psycho coder who just went wacko, but did no damage. Maybe it will turn out to be a criminally incompetent programmer -- I'm sure you know plenty of those!


Yes, but... we need to traverse a slope of comprehension here. Right now only a few tens or hundreds of thousands of people in this country can appreciate the evidence. That tiny group has to engage itself in the debate -- and carry out early investigations -- and determine whether this is serious enough to be brought to the attention of others. If the expanding investigation continues to uncover additional evidence, eventually it will suffice to convince even the <shudder> mass media. But we've got to get the ball rolling somewhere.


I concede your point. Yes, we have to assume status quo unless something otherwise arises. The question here is, how convincing does the evidence have to be to justify further investigation? I have raised just three of the most significant points of evidence -- there are some others.

At the risk of taking us off into a nasty digression, I'll ask how far off the exit polls would have to be to justify suspicion. For this election, the exit poll discrepancy was 5 percentage points. What if the discrepancy were 10 points? 20 points? Where would you draw a line and say, "This is just too weird -- we should investigate this." ?


I'm sorry, that comment was directed at carlitoswhey, not at you.

I pointed out Febble's Fancy Function, the lack of multivariate analysis to explain the uniform response bias, and the fact that they assumed a normal distribution of precincts, then based calculations off a normal distribution. I'm sure that there are a few other nits.

Why do you continue to ignore the forest when it's clear that the trees are all leaning far left and screaming "conspiracy" into the wind? The mere fact that we have a cottage industry focused on an exit poll shows just how far down the list of potential conspiracies we have sunk.

Yes, you did respond to that evidence, and so I apologize for the misplaced jibe.

On the matter of the assumption they made, let's remember Occam's Razor. The normal distribution is the simplest assumption to make, and so the one to prefer in the absence of other evidence. Have you any evidence to suggest that a normal distribution is inappropriate?


Hmm, if I take your analogy correctly, it is an exaltation of ad arborem reasoning -- arguing against the trees, rather than the evidence. I really don't care if the people providing the evidence are child-raping commie terrorist litterbugs. I decide based on the evidence alone. Shouldn't you?

Erasmussimo,

What possible evidence is there that there was no electoral fraud? In order for a hypothesis to be a valid, it must be disprovable.

What evidence would, without question, dismiss this theory?

I will posit that this theory is not disprovable and thus is not valid.

I still maintain that the polling data was in error. Either it was not a representative sample (I am sure you know that for any statistical analysis to be valid, the sample must represent the whole) or there was an error in methodology in the data collection.

Sounds like a great idea to me! Let's disprove it! All we need is some further investigation. With luck, we might be able to disprove it without a big investigation. Here are some things we could do to disprove the theory:

1. Look at the E/M raw data so we can see what really happened with the exit polls. If the E/M data checks out well, that would surely go a long ways to discredit the hypothesis that the exit polls were more accurate than the vote counts.

2. Examine the source code for the Diebold op-scan machines and make sure it corresponds to what's in the ROM; if there's no funny business in the code, then we could reject the hypothesis that the Diebold op-scan machines were rigged.

3. Examine all processes by which the memory cards for the Diebold op-scan machines were formatted and handled. If these processes meet standards for information security, then we can reject the hypothesis that they were used to manipulate the vote counts.

4. Do a full hand recount of all votes cast in Cuyahoga County, Ohio. That county showed numerous statistical discrepancies that suggest fraud. The public prosecutor has instigated an investigation into some aspects of this election; a full recount could demonstrate that the election really was conducted fairly.

5. Do a full hand recount of the voting records for Palm Beach County, Florida. This is another one of the most suspicious counties in the nation; if it were to pass all tests then we could safely assuming that less-suspicious counties would also pass. You can find some of the information related to electoral anomalies in Palm Beach Country right here

I'm sure that there are other ways to establish the reliability of the vote counts. Yes, it's true that it is impossible to prove a negative, but an examination of some of the most egregious problems could go a long ways to establish the credibility of the electoral results.

So whaddya say we investigate some of these things?

I have no problem at all with 1, 2, or 3. Hasn't 5 already been done? I thought that was one of the counties that did have a full recount done (post election). For 4, I would just like to make sure what the objective of the recount was, and limit the scope...once we start recounting one county, it would be far too easy to have that cascade into a massive call for recounting, which would be a massive waste of time and effort.

I don't believe so, but I could be wrong here. I know that they did an audit and the audit came up with some infelicities. The link I provided earlier provides some of the details. However, since it's from Bev Harris, you have to ignore her own representations of the facts and look only at the facts themselves.


I agree, it just won't do to have random fishing expeditions. There were lots of problems in other counties in Ohio, but Cuyahoga County seems to be the worst, so it provides us with a good test case. I think that there should be nothing more than a strict recount of all the ballots cast, using the very thorough methods they used in Washington State, where there were multiple referees and representatives from each party overseeing every single ballot counted. We should expect that recount to yield results different from the actual election count, and we should establish in advance two levels of threshold: 1. below this threshold, we call the election fair and suspend further investigations; 2. below this threshold, we decide that further investigation is called for. Above the second threshold, we declare the election in that county flawed and initiate investigations into other counties.

I recommend that threshold 1 be set at 1,000 votes, and threshold 2 be set at 10,000 votes.

What exactly would make the data "check out well"? What kind of criteria is that? Do we have a pre-conceived notion of what the data is supposed to look like? Is there some way to know if there were flaws that would make this sample non-representational? How do you "check out" data to make sure it is "ok"?



My home county. I myself was surprised at how many voters talked about Bush while I was in line (about 45 minutes to an hour). There is sure to arise some discrepancy between the recorded vote and any recount. It is simply inevitable. How much error would make one believe that fraud is present? What if the count error was in Kerry's favor rather to Bush's?

Here is an article about What Didn't Happen In Ohio

It links to this analysis of the election results versus the exit polling data. Interestingly, we find this tidbit in the .pdf document:



The whole thing is a good read. It addresses many of the points raised here in an objective (rather than ad hominem) manner.

By the way Erasmussimo, your 5 points do not address my question, which was what results would disprove the election fraud theory. You gave methodologies, but no qualifiers. "Let's examine" doesn't prove or disprove a hypothesis.

Question for both of you - why would a re-count be more accurate than the original count? I'm sure that we all agree there is some margin for error, whether computer, mechanical or human. We would just end up with a different set of data, one slightly different than the actual vote count on election night. When this happened in his county in Michigan, curmudgeon was convinced for months that there was something foul afoot. We also saw this in Florida 2000, where hand counts were slightly apart from machine counts, and then varied even more based on the system employed - hanging chads, light through the card, etc.

Yes, this is exactly why I would want the objectives of the recount clearly stated. This should then lead to any methodology being clearly stated. I also think that in going through this process, the problem you indicate (highlighted above), and which Amlord I think also eludes to, might lead to the conclusion that no recount will help resolve the problem. If this seems to be the case, then I don't think the recount should be done. Assuming that an approved methodology for doing the recount...is validating exit polls sufficient reason for doing so? That in itself is a question that is open to debate. Certainly validating the results themselves is a worthwhile endeavor...but would that really be what would be done? Remember the recounts taken post election in Florida. Using any of the three methods sought by the Democratic party, Bush's lead in Florida actually increased in all recounts. There was a fourth method, not one that was proposed, that would have swung the votes in Kerry's favor. So, determining how to do a recount is not easy. And the method chosen might vary depending on whether the goal was to verify the exit polls, or verify the results themselves. For example, if verifying the polls is the goal...I think discarding any questionable vote is reasonable. This cannot be done if validating the actual vote is the goal (or can it?).

I have revised my criteria below; the new criteria should now meet your requirements.


As I wrote, any discrepancy less than a thousand votes (obviously, in favor of Kerry) would dismiss suspicions of election fraud in Cuyahoga County.



The first article doesn't impress me; it's entirely too partisan and its wordings are a bit too slippery for my taste. However, the second link really is quite good. Brady does a good job of demonstrating flaws in Freeman's paper. Indeed, he pretty well trashes it. However, Brady also supports the basic conclusion of the USCV paper:




OK, I'll reword my proposals with firm tests:

1. Perform statistical tests of the E/M rBr hypothesis against the E/M raw data. If the raw data supports the hypothesis, then we can dismiss the exit poll discrepancies.

2. Examine the source code for the Diebold op-scan machines. If it meets all standard requirements for secure code, then we can dismiss the hypothesis that Diebold op-scan machines were hacked.

3. Examine all processes by which the memory cards for the Diebold op-scan machines were formatted and handled. If these processes meet standard protocols for information security, then we can reject the hypothesis that they were used to manipulate the vote counts.

4. Do a full hand recount of all votes cast in Cuyahoga County, Ohio. If the result of the hand recount differs from the reported count by less than 1,000 votes (in favor of Kerry), then we can dismiss the hypothesis of vote fraud in Cuyahoga County, and I'd be willing to go you one better and dismiss the hypothesis of vote fraud in Ohio.

5. Do a full hand recount of the voting records for Palm Beach County, Florida. If the results of a properly-conducted hand recount differ from the final vote by less than 1,000 votes in favor of Kerry, then we can dismiss the hypothesis of vote fraud in Florida.

There, now you have falsifiability measures for all claims.


Because a recount would be done with every single ballot examined by a county worker, rather than run through a machine, with representatives of both parties observing and making any protests to a supervisor. This insures that every single ballot gets a proper evaluation.

Actually it would be the reverse. If we wanted to see how people would have voted in an exit poll, we should count ALL ballot, regardless of questionability.

For a vote to be valid and count towards the actual vote count, there are certain limitations (not more than one vote, must be clearly punched out, things of that nature).

At the risk of showing my "bias" I decline to accept that 3 humans looking at a ballot designed to be counted by a machine are more accurate than a machine. Who gets the final vote? Who is arbiter of disagreements as to voter intent? This is all ultimately subjective, even if you define everything in advance.

Moreover, voters of Cuyahoga County appropriated a certain amount of dollars to afford their citizens the right to vote for President. If the exit poll people want to validate their poll (which it doesn't seem they do) then they should pay for it themselves. I've never heard of validating a sample by counting the data...

Well, I stand by my initial contention...I think your comparison assumes something I do not...that people actually voted the way they said they did in the poll. Therefore, to validate the exit polls, all you need is an accurate representation of the vote, not necessarily the actual vote itself. Discarding all questionable ballots seems the fairest way to do this...unless one assumes bias in the votes cast out this way.

Your statements about the actual votes are correct...but those would be discarded anyway, as they're not valid votes.

One would hope that either method would produce similar statistical results...if not, then there is indeed bias in the discarded votes, which indicates a potentially serious problem in the way we collect votes. This was one of the contentions in Florida, but I don't really find it valid. A busy executive in a hurry, I think, is just as likely to punch a card incorrectly as an uneducated person. But I don't know if there's any data to support that either way.

This kind of problem has long since been worked out. If there's any disagreement, the ballot is referred to the supervisor, who puts the ballot into a "contested ballots" box. Those contested ballots are then gone over by another group. In the end, almost every ballot gets a decision. Ballots that just can't garner agreement are not counted.


The purpose of this exercise would not be to validate exit polling, but to validate the vote count.

Why? The vote in Ohio went pretty much as predicted. Bush won by 1%. Or are you talking about the exit polls?

Cuyahoga Country saw some pretty serious oddities. As I mentioned earlier, the public prosecutor there has initiated an investigation into some aspects of the vote counting there, but I do not know the scope of his investigation.

Here's the most egregious of the oddities: more than a dozen precincts reported anomalously large numbers of votes for third party candidates -- and each of those precincts was surrounded by precincts that voted heavily for Kerry. The suspicion here is that votes for Kerry were somehow transformed into votes for the third party candidates.

There are a number of other oddities about the voting in Cuyahoga County; I can dig them up if you wish.

All of this makes me revert to my initial position which is that the election fraud proponents are throwing everything under the sun out there, in the hopes that something, somewhere will stick. Long lines in black precincts, Dixiecrats voting for Bush, Diebolt hacks. Your 5 points, and your various Cuyahoga county sub-points notwithstanding, there is no amount of investigation that will satisfy you. Whatever the conclusion, you will be able to point to another sober statistician or theorist with a hot set of data clamoring to be examined. Black Box Voting alone has about 40 different theories, and there are scores more in the democratic blogosphere. There will be no end. And meanwhile Kerry has conceded, and the election is over.

I note with some disdain that none of the impartial, sober analysts care to opine on the fact that more votes than voters were counted in Milwaukee Wisconsin (where Kerry won, by complete coincidence). Of course, that would be another thread, because all we are doing here is evaluating whether the rBr effect is valid or Mitovsky should release their data...

Why do you engage in such silly speculation? You haven't the faintest idea of what will satisfy me. You could reasonably infer from the case I have presented so far that the three most telling issues, in my mind, are the exit poll discrepancy, the Diebold op-scan machine hook, and the machine type discrepancies in Florida. I would also like to see investigation of the anomalies in Cuyahoga County, Ohio. If those issues were all resolved, then I wouldn't see much of a case for further investigation. Why are you opposed to any investigation under any circumstances?


Feel free to debate those issues on their sites. This topic is not about those issues.


If you wish to start a topic on that topic, feel free. And if in that topic you presented decent evidence to support suspicions that the election there was fraudulent, I'd be all for an investigation. That's because I'm consistent and not motivated by partisan considerations. Can you say the same thing?

Best of luck on satisfying your criteria. I acquiesce. You win.

carlitoswhey, falling back on vague insinuation when logic fails you does not constitute a logical response. Please explain the apparent discrepancy between your suggestion that an investigation is called for when there might be electoral fraud in favor of Democrats, and your absolute refusal to consider an investigation when there might be electoral fraud in favor of Republicans. Is this consistent?

As I have noted periodically in this thread, there have been many, many investigations of our election in 2000 and 2004. My theory is that there will never be enough investigations to satisfy those who do not believe the election results. It's become the myth of our time.

Here is a summary of the latest investigation into Ohio, conducted by (highly partisan) Donna Brazile, chairwoman of the Democratic National Committee's Voting Rights Institute. The report shows that, yes, there was confusion and long lines, but no, there was not a "pattern of vote shifting" from Kerry to Bush. It recommended a bunch of changes to voting methods, etc.



If this does not satisfy your Diebolt / Mitovsky jones, will you now at least cross Coyahoga County off your list?

1, What the members of the right seem to forget is we under the Bush administration (the USA) paid for polls in Uzbekistan to check if there was voter fraud. Hmmm that tells me that polls are important. That is standard in countries where it is not a banana republic election.

2, the US poll cost was in the millions of dollars it was not a casual poll.

3, a poll can of course be off, but what it really means is if the results do not match significant effort should be made to determine what happened. This investigation should always be run by the party that lost. All manufacturers and related of the machinery and election should be deposed.

4, In my professional area we call polling AQL, this is what makes products you buy out of the package better and better, if you think we do a complete test of the entire product you would be mistaken, we take a random sample all the time and if it hits a certain level for critical failure the entire lot must be inspected. As we know many years ago before AQL was used we had to return so many brand new products because they just did not work out of the box. I have also heard accounting firms use a version of this to catch accounting fraud. So statistical analysis is used as a trigger to investigate in detail. That is what we should do.

5, Most if not all the companies that make voting related machines are owned and run by partisan republicans. This should never ever be the case.

6, Also the republicans over history MO have proved to be involved in conspiracies, Nixon’s boys, and under Reagan Ollie. Sure there were many more never uncovered. Remember it took us many years before Nixons surfaced, if not for deep throat he would have gotten away with it.

7, One of the major manufacturers Diebold had significant documented failures, including settling with California for several million dollars for machine issues. Co-incidentally shortly after the election. As we know Diebold also makes ATMs. When was the last time a ATM had a electronic failure? If we had failures like the election our banking system would collapse. Thus if they can be at 9 or so sigma for their cash machines they certainly can for voting machines.

8, A person (former republican) programmer testified under oath that a republican who is now a congressman asked the republican company he(programmer) was working at to write a program that could change the votes. This representative was asked to testify under oath and he would not. What was his motive, once the program was written did he go to the FBI, the news media, to congress to stop any potential fraud, NO he did not so we have to assume he had questionable motives.

9, I am an Engineer and have been VP of Engineering of a company and Director of R&D and making these machines would be well within my capabilities to design. I can tell you under no certain terms there is no way to stop vote fraud the way the elections have been run; and I could NEVER be caught except by a poll from cheating. I am dead serious there is just no way you could catch me. Again I could put in a chip IC in the board that has code burned in it and you would never know the chip is there and what its purpose is. You do not have the money to pull every chip and get the code out of it. It just is not practical to go do this to every machine. So reading the source code would do absolutely no good. My chip would circumvent the operating system on a certain day at a certain time after 200 votes were cast at a max rate of say 10 votes per hour. Sorry no simulation/test you could ever perform could catch it… Or it would only activate when a certain combination is pressed or a Comm link is hooked up after the preceding. Again impossible to stop.. One person could do this. Also could be triggered by the memory card after the previous were done. Some of the memory cards could also have hooks implanted in a small percentage of them.

10, the issue of not having printed receipts causing voter fraud is well discussed for at least 20 years, and I can tell you a printer like on gas pumps costs about $30.00. Thus again why do these machines not have them when the issue has been raised a long time ago, even the ATMs give receipts so Diebold has no excuse and it can be assumed there was not a good motive for not having them on what $4000.00 machines (my guess they were at least 4K-6K no idea), but a printer at $30.00 is nothing. http://www.cbma.com/Products.asp?Departmen...n=P&ShowIntro=0

11, anecdotal, I have several friends that for the first time in their life did not vote republican. And know of no liberals that changed and all independents that also voted for Kerry. I know no liberal that changed their vote and some that voted for Nader the last time and Kerry this time. Again this is anecdotal. Even on the main board I post on many of the major debaters from the conservative side did not vote for Bush. So something is rotten in Denmark…

12, The question I ask of the people on the right is if you think that these companies making the machines are on the up and up when they are owned and run by partisan republicans why don’t we switch every other election, so the next two presidential should be run by machines I make or how about if the company was owned by Michael Moore? Would you ever trust an election run by machines made by Moore? Or even better yet, would you ever trust an election that was run with machines made by the Chinese?

So again people on the right would you trust any election where the machines was made by Michael Moore or the Chinese??????????? END OF DEBATE………… no republican has said yes to that yet.

He's baaaack! Yes, the scandal that just refuses to stay under the rug is back with even more grist for conservative bruxing. Actually, this is old news, but I was too lazy to bring it back to life until now. You can find the complete technical report on the security of the Diebold opscan machines right here. Now, when Bev Harris first reported on this, I noted it here but also warned that Bev Harris is prone to overstating her case. However, this time, she wasn't overstating the case. The detailed technical report demonstrates that the security situation with these machines is much, much worse than I could have imagined. To put it bluntly, the Diebold opscan voting machines have no security. None, zero, zilch, nada. If you can gain physical access to one of the memory cards -- not even the machine, just the card -- then you can do anything you wish with the machine. You can produce any vote count you want.

Here's a quick summary of their results.

1. The machines are programmed in BASIC. Now, some of you may know about BASIC. It's an ancient computer language invented in 1968 to help students learn programming. The name is an ancronym, the first word of which is "Beginner's". BASIC is a beginner's language. Nobody does professional programming in BASIC. There was a brief period of time, around 1980 and 1981, where there were commericial programs for personal computers written in BASIC. But that was back in the days when people dismissed computers as toys. Nowadays, BASIC is considered a toy language. Telling a group of programmers that you program in BASIC is rather like walking into a bar in the Old West and asking for a glass of milk. Yet here we have this commercial, supposedly secure product written in a kiddie-toy language. This is ludicrous beyond imagination.

2. There are no software security protocols whatsoever. Not even checksums. There is absolutely nothing in that machine to protect it from being taken over by just about anybody. There were computer games written in the 1980s that had better security than this. Your average home PC has better security than these machines.

3. The machine refers to an external device for its code. Do you know the little "thumbdrives" you can buy for maybe $50 that have 256 MB of storage? You can plug them into any USB port and they look like a hard drive to your computer. This is a quick way to transfer data from one machine to another. Well, the Diebold machines have something like this, except that it's older, clunkier, and more expensive, but it's basically the same thing. They use it to transfer data from the machine at the polling place to the central machine at county headquarters. What they never told anybody was that they also program those little memory cards with the programming instructions for each election. It's rather like this: your friend Joe gives you a thumbdrive with some pretty pictures on it. You plug it into your machine. The instant it's plugged in, your machine runs out to the thumbdrive and asks it, do you have any programs you'd like me to load and run? Joe has loaded a little Trojan horse into the thumbdrive that will download all your financial information into his thumbdrive. So you plug it in, it gets your financial information, and you download the pretty pictures. You return the thumbdrive to Joe and guess what? You're screwed. That's pretty much how these things work.

4. It gets worse. You might think that, well, at least the technical information that a hacker would need to use is secret. But no! That technical information has been publicly available on the Internet for years. Everything you need to know to hack the system is there. I believe it has since been taken down from the Diebold source, but the hackers all have it.

5. It gets even worse. You might think that, well, at least the cards that they use are custom-designed and so nobody could actually get their hands on a card unless they stole it from the county. But no! The cards have been on the open market for years.

6. It gets even worse. You might think that, well, at least the devices needed to program the cards are custom-built hardware that only Diebold has. But no! They're available for under $500 from a vendor in Minnesota (or perhaps Wisconsin, my memory slips here).

Let's summarize all this. The security on these machines is basically similar to a bank vault consisting of four posts holding up a roof, with a table underneath loaded with cash and jewels, with some little old ladies working at tables nearby to keep an eye on the vault. Any high school kid with a few hundred bucks to spend and some way to swap cards (not too difficult in the pandemonium at many polling places) could alter election results to obtain any results he desired. For all we know, the 2004 election was hacked by BOTH the Democrats and the Republicans, and the Republicans won only because they were more aggressive in their hacking than the Democrats. Right now, there is no way to have any confidence in any of the vote counts produced by the Diebold opscan machines.