Florida Voting Machine Logs Reveal Anomalies

Florida Voting Machine Logs Reveal Anomalies

Please think before posting please! don't want this thread shut down.
The right for accurate voting results is an American issue not a republican/democrat one.

"Having 'successfully sued former Palm Beach County (FL) Supervisor of Elections Theresa LePore to get the audit records for the 2004 presidential election,' Black Box Voting reports that the 'internal logs of at least 40 Sequoia touch-screen voting machines reveal that votes were time and date-stamped as cast two weeks before the election, sometimes in the middle of the night.' Besides the date discrepancies, they claim to have discovered countless other errors and anomalies, including a case of one voting machine being 'powered down 128 times during the election'."

http://it.slashdot.org/it/06/02/24/1326217.shtml

Just another example of why I prefer the old mechanical voting booths to our new 'better' electronic ones. It is FAR to easy to monkey with computers to make them give the response that you want. This is also why I refuse to play on video poker or electronic slot machines.

Unless I missed it in the article there is no reference to who the votes were actually cast for. It would be interesting to see if they were mostly/all for one cantidate (suggesting an attempt to rig the elections) or if there is some mix of votes for both parties/cantidates (suggesting bad hardware/software in the voting computers).

The authority on voting machine fraud is Bev Harris, who has been thoroughly researching the subject for years.

Here's an interview with Bev Harris:

http://www.buzzflash.com/interviews/03/12/int03323.html

Here's additional info:

http://www.seattleweekly.com/news/0410/040310_news_blackbox.php
http://www.commondreams.org/headlines04/1106-30.htm

I am generally Republican in nature, actually more constitutionalist with a tinge of libertarian but I digress.

If proof is ever found of actual vote tampering with the intent of illegally electing a particular person (I am talking about malice here not bugs) then my personal take on this, is that it is treason and all involved should be shot. No matter what party they belong to.

I am not a huge fan of som of the extremely liberal left and being a gun owner some of the more vocal proponents of gun control but we should be electing are leadership through our constitutionally guaranteed means, anything else is just a coup de tat and I don't care which party is at fault.

Chris

I have to wonder at what point the voting process went from fool proof to fraud friendly? In the early days the voting was for representatives, and every one knew who their representative was supposed to vote for. It was easy to see fraud, since a small, local population pretty much insured that you'd notice if the wrong party won.

My grand dad once said that when he was young, they could tell there was hanky-panky if the ratio of votes differed from the ratio of the registered party affiliations.

At some point, the volunteers at the polls no longer recognize their neighbors. They have to rely on paperwork, yet there is no authentication to prove the paperwork is accurate nor to prevent a person from
visiting multiple pooling sites.

Daniel

No election system is ever going to be 100% perfect. There will always be errors no matter what system. These newer systems just keep more data so you can see every error. After all, what proof would there be if one of the old mechanical machines occasional didn't raise the counter by one when somebody pulled the lever. That being said, I think it was stupid to spend all that money switching over to new systems after Florida 2000. Whatever errors were made by the machines had no way to favor one particular candidate, and the overvotes and undervotes probably balanced out when you looked at a large enough number of votes. Once they started hand counting votes in only certain areas, they introduced the human bias in favor of particular candidates.

However, I do think it's proper to investigate if you have indications of vote fraud. Since Florida 2000 these stories have gotten a lot of press. The cases where there were more votes logged than registered voters, early voting ballots sent out with votes already filled in, or political office buildings being used as registered voter addresses need to be investigated. And if there is sufficient evidence of a pattern of fraud, even if you cannot prove who benefitted from it, then the courts should get involved and action to restore the integfrity of the election process must be taken. Someday it might even be necessary to throw out the results of an election and do a limited revote with all the TV cameras running. It would be a circus but at least it would restore most peoples confidence in the system.

For whatever reason, the 2004 Gubernatorial (I love that word!) election here in Washington never got a lot of air time on the news, but it makes Florida's problems pale.

There were more votes cast in King County (Seattle) than registered voters! Yet, the courts found no "voter fraud". Keeping this politically neutral, one candidated was declared the winner. Then they had a recount and the candidate was affirmed. Then the other candidate sued for another recount, and that one "discovered" all these votes in King County, and the candidate who lost the first two counts was declared the winner! (I guess we have a 'worst two out of three' policy here) What a mess.

The biggest problem here and in the Florida 2000 presidential results was that the winning margin was LESS than the error rate of the voting process. I.e. statistically, you could not determine a winner, but there is no law to account for that.

The elections process in this country definitely needs fixing. I would like to see a law that if the winning margin is less than 2x the error rate of the voting process, an automatic runoff is triggered.

just read Rudi's link above.

duh... this makes sense!

"There are three types of activities that fraud-prone and require auditing designed to deter the fraud: financial transactions, gambling, and elections. Yet we have not sought the counsel of the very people who understand this type of accounting: Accountants, bookkeepers and auditors! As a result, we have legislation in many states, and in this case, in HR 2239, that uses an inappropriate and flawed auditing model which will not work.

The very first thing we need to do is get solid input from auditors who are experienced in fraud detection. When it comes to setting up practical, effective auditing for these systems, bookkeepers from Las Vegas probably have better expertise than computer scientists from Princeton."

I think no matter if it is digital, mechanical, or paper there will always be problems.

Some people seem to not care how many dead people vote.

Here in Indiana there was someone following around some vans full of voters. The all went into a voting area then came back oiut after some time had passed, then the van went to another precinct and they filed out and went to another polling place, then another, and finally another.
At work when I said that was outrageous but my fellow workers thought I was silly and why should anyone care.

When the people asked the van full of people why they were voting so many places there was an uproar that the people were trying to deny them the right to vote by following them around. It seemed the people that discovered this got more of a bashing than the people doing mulitiple voting.

When we see problems in this world, it is easy to look up at the top and say there is a big problem but I wonder if we should also look around our selves and wonder if the problem is all around us. If we act in an unethical manner and those around us act in an unethical manner should it be so shocking that some at the top of the food chain do the same thing?

I personally think we should go 100% towards voting on the Internet. With proper registration procedures it wouldn't be possible for a person to vote more than once, and it'll increase voter participation. All the other systems seem fraught with intractible problems, plus they cost way more than an Internet-based system would. The only drawbacks of voting on the Internet I can think of would be that those who don't have or know how to use a computer couldn't vote. IMHO this would be a good thing since this would mostly constitute the poorer/uneducated portion of the population who is more likely to vote in those who give them handouts. In fact, I really feel that if you're getting some sort of unfunded handout (not Social Security or unemployment since those are funded via payroll taxes) then you shouldn't be allowed to vote as long as you're receiving it. Internet voting would accomplish that in a way without explicitly doing so.

The problem with a valid audit is that it violates the "secret ballot" concept. There's no way for the auditor to tell what the person actually voted for unless someone can contact that voter and ask. WIthout that starting point the rest is just validating part of the counting.

The internet voting has the same problem. You have to require each person to pre-register and then require that they authenicate themselves before voting. Anything else is trivial to defraud.

There are many, many other ways to influence the election legally. You can set voting districts to favor one party over another. You can make it difficult to find polling places in neighborhoods that favor the opposition. You can send out voting materials far, far in advance to areas where you want a low turn-out and closer to the election where you want more votes.

Long ago, it took a conspiracy to create election fraud simply because so many people were looking closely. Now it can be done with only a couple well placed people. The company that provides most of the electonic voting systems even went so far as to require that there is no independent validation of their machines here in Calif.

I think the best solution is simple. It's also better for the country. Require a minimum of a 3/5 majority to win any election. That's beyond the margin of error AND it would assure we have someone that most of us like. I refuse to believe that we have such a polarized society that we have to settle for a 1% margin electing someone who represents all of us.

Daniel

gadget_lover said:

Require a minimum of a 3/5 majority to win any election. That's beyond the margin of error AND it would assure we have someone that most of us like. I refuse to believe that we have such a polarized society that we have to settle for a 1% margin electing someone who represents all of us.

Click to expand...

Not a bad idea at all. I'll go along with it regardless of voting method. I'll also add that in our legislature chambers any tax/spending increases should require a four-fifths majority and any tax decreases a simple majority (I personally think a one-third vote for tax/spending decreases is even better). Fact is the size of government has gotten out of control with our representatives using government benefits/pork to buy themselves votes.

It's not correct that under and overvotes etc. necessarily cancel out. Suppose Alice and Bob are running for mayor of LEDville, population 2 million. All the voting machines in LEDville have a 3% error rate (this is about the situation for the type of punch card voting machines that became famous in southern Florida in 2000 when the machines are well maintained). In this case the errors cancel out. If it's a close election, there's about 60,000 votes scored incorrectly by the machines, which are unlikely to skew the result by more than sqrt(60,000) or about 250 votes (1 standard deviation). If the actual votes are 50.1% to 49.9%, that's a 4000 vote difference, more than plenty to swamp any likely error due to the machines. No problem.

Look though at a slightly more complex case: Alice comes from North LEDville (population 1 million) and Bob comes from South LEDville (population 1 million) and each gets 80% of the vote in their home region. Again, the totals are very close. North LEDville uses expensive new machines with 0.1% error rate (like the optical scan machines in northern Florida) while South LEDville has old, poorly-maintained machines with 5% error rate. That means in North LEDville, of the 800k votes cast for Alice, 792k actually get credited to her and 8k (the 0.1% errors) get credited to Bob. And of the 200k cast for Bob, 198k get credited to him and 2k to Alice. Result is: 794k Alice, 206k Bob, looks nice for Bob. But now look at South LEDville: of Bob's 800k votes, he's credited with only 760k while Alice gets 40k, and of Alice's 200k votes, she's credited with 190k with Bob getting 10k. South LEDville totals are reported as Bob 770k, Alice 230k. Adding up the totals for North and South LEDville, we get 1022k Alice, 978k Bob; the difference in equipment has systematically biased the error so Alice has gotten an extra 4%, enough to swamp out the real difference in a close election.

The notion that the errors cancel is only true if every polling station uses equipment with the exact same error rate. If machines with higher error rates are predominantly in precincts of one candidate, there's a bad problem. The distribution in FL wasn't as extreme as 80%-20% anywhere, but it's definitely the case that just like in any state, some regions favored one candidate and some favored the other.

US elections have a lot of problems and I remember former president Carter (who now runs an organization that monitors elections in countries that are recovering from dictatorship etc.) saying US elections are too screwed up for his organization to certify. There's much less uniformity of equipment and election laws than in other countries. To some extent this is difficult to fix here, because the stuff is left up to individual states and often decided by people who are ignorant or corrupt. There's just no way for some nationwide standard to get approved with public input and then deployed everywhere in the country.

(insert tounge in cheek here)

I think we should have voting done by dial in.

Dial 1-800-vote-now

Enter your national voter Id number
I'm sorry I did not understand, please enter your national voter number.

for national election press or say one
to vote for president press or say one
to vote for candidate x press or say one
to vote for candidate y press or say two
to vote for candidate zzz press or say nnn...
to vote for Pat Paulsen say "Chicago dead voter precinct x"

(those of you under 40 ask an adult )

for governor press or...... you get the idea.

The poor shlubs at the bottom of the ticket won't ever get a vote.

Are voter machines Windows powered?

(remove tounge from cheek here)

We have been using the electronic paper ballot, the large newspaper format ballot with real switches next to the choices. Our precinct workers tell everyone in line when there are two pages many times before they step up to the ballot. It seems easy enough but they have been glitchy well before 2000. Probably can't help but be a little buggy since they are used so infrequently and then when used are quite heavily loaded. There is a little light that lights up on the outside to tell the precinct worker if the vote was logged. I am partial to the notion of internet voting if it could be secure.

I don't have proof, but logic suggests that the error will not cancel out. My reasoning is:

1) the error is caused by something going wrong, and broken software or hardware tends to make the same mistake over and over. A memory bit that is going bad will seldom change from set to unset and back again erroneously. I dirty optical scanner will read poorly over and over. A mechanical feeler that's a few thousandths off to the side will read the same section wrong when it makes it's mistakes.

2) Human factors cause errors too. A ballot that's confusing will confuse most people into making the same mistake. If you have three check boxes one above the other, but off center, it's more likely that a mistake will end up credited to the middle guy, and almost never will the top guys' vote get accidently tallied in the bottom guy's box.

3) Truely random errors are not evenly spread about. An expert once pointed out that a graph of random results should always show groupings somewhere, otherwise it's actually representing an ordered result.

Yes, the dibold voting machines are powered by Windows. Well, they were last year.

Daniel

paulr said:

It's not correct that under and overvotes etc. necessarily cancel out.......

Click to expand...

Your theoretical statistical analysis assumes that the errors in different systems are going to go to a specific candidate. My point is that the truely random errors, that are going to occur no matter what system is used, will even out in the long run. Unless there is fraud involved, there is no way that the "type" of system used can favor one candidate over another in terms of these random errors. Although different systems might favor one type of error over another, the system errors are equally likely to occur for one candidate as another. In the end, with a large amount of votes cast, all candidates should gain or loose votes due to error equally. In the end, the net results will be unchanged. The only exceptions are mechanical breakage on the old machines or a programming error with the newer electronic systems which could selectively misrepresent vote results and misapply the results consistently. But in this case, the error should be apparent by wildly erroneous results.

What I hated about the whole Florida 2000 mess was trying to turn the random errors, a hanging chad or a partially filled in optical ballot mark that might be misread by an impartial machine, into an excuse to introduce natural human bias into the results process. It was even worse later on when it was used to demonize systems that had worked fine for years. As we are starting to see, there is no perfect system and it has been a complete waste of people's hard earned tax money to replace all the old systems. The best we can do is try for the highest accuracy achievable at a reasonable cost. And we need to keep the human biases out of the system. At least the machines don't belong to any political party and their errors won't favor anyone.

One of my favorite quotes (as near as I can recall): "There are three types of lies, in order of severity: Lies, Damned Lies, and Statistics". Depending on the initial conditions and assumptions used, the exact same statistical analysis can be used to produce two exactly opposite results.

Whether errors cancel out when they're distributed randomly is irrelevant to real-world situations where they're distributed non-randomly. Florida did not assign equipment to precincts randomly. Just as with North and South LEDville, the type of equipment was correlated with the voting patterns. That difference wasn't politically motivated or anything like that. But the economically poorer districts of FL had the older and less accurate equipment, and also tended to have higher proportions of voters of a certain ethnic persuasion that has traditionally leaned heavily towards one party, while voters in other parts of FL favored the other party, giving overall a (less extreme) version of LEDville. One partial solution is to have the state buy all the equipment for every district instead of letting/making the districts buy their own stuff.

donshock said:

My point is that the truely random errors, that are going to occur no matter what system is used, will even out in the long run.

Click to expand...

That statement is only true if you have an infinite number of ocurances AND you stop it when it is even.

If you test random occurances you will see that there are points where the deviation is quite marked.

Take a simple example. Flip a coin. How many times do you have to flip it before you have twice as many heads as tails? Flip it long enough and it will happen. In an election the end point is not arbitrary. The accumulated error freezes whenever the last vote is counted. If it happens to favor one canidate, well, that's the definition of random.

The method that we use in the US makes minor errors more important when the race is close. A miscount in justa few key districts can allocate the delegates for the whole state to the wrong contestant. The trick is to validate those votes when the only audit trail is that X number of people showed up to vote.

Daniel

gadget_lover said:

Take a simple example. Flip a coin. How many times do you have to flip it before you have twice as many heads as tails? Flip it long enough and it will happen. In an election the end point is not arbitrary. The accumulated error freezes whenever the last vote is counted. If it happens to favor one canidate, well, that's the definition of random.

Click to expand...

I don't think the coin flip example is true. For any N no matter how large, the chance of getting 2x as many heads as tails after N flips is extremely low if the coin is fair. We can say if you flip long enough you will eventually get 20 heads in a row if the coin is fair. But the difference between num(heads) and num(tails) will almost always be within a small multiple of sqrt(N). You can compute the probability of the difference being larger than, say, 3*sqrt(N) and it's very low (less than .001). If the machines have error rate E (like E=3%) and the difference in reported totals is larger than 3*sqrt(E*N) then it's 99.9% certain that the totals reflect the correct winner--IF--very big IF--IF the errors are uncorrelated. The problem in this situation is that the errors are NOT uncorrelated.

Lets take an even simpler example: the machines never assign a vote to the wrong candidate. They have a success rate S (say S is 99.9% for a good machine and 95% for a bad one). When you cast your vote, with probability S for whatever your machine is, your candidate gets credited for it. With probability 1-S, your vote doesn't get counted at all.

Say you live in a district that heavily favors Alice and your district uses bad machines. That means that every voter in your district effectively gets only 0.95 votes, while the voters in Bob's district (with good machines) each effectively get 0.999 votes. You can see how this doesn't cancel. It's just a less extreme case of letting your opponent's supporters vote twice while you only get to vote once.

Of course in presidential elections there's an even larger disparity because of the electoral college, but that's an anomaly of history rather than anything having to do with machines.

Ahhh. Paul said what I was trying to say, only more clearly. Thanks Paul.

The point about the coin toss is that it is NOT certain that it will be balanced, and a non zero chance that it will be drastically unbalanced at SOME POINT. If you happen to stop at that point the result is unbalanced. Only .001 chance that it's way off? That's a non zero probabliity and (in my mind) that negates the assertion that it will balance out all the time.

I was once told (don't know of it's true) that when you flip a coin, your chances are always 50/50, even if it's just come up heads 100 times in a row.

Daniel