It's Common Sense, I Guess
By Erick Posted in Elections — Comments (40) / Email this page » / Leave a comment »
I don't consider myself a crank or a kook, but the more I think about the more I don't have a problem requiring a paper print out of an electronic vote.
Here in Georgia our machines do not do that. This past election cycle, the state experimented by having a paper trail added. The paper results were counted and compared to the electronic results. The end result was a 2% deviation from the electronic results -- a deviation easily explained by human error (in fact, all the counters interviewed admitted to occasional mistakes in counting).
The left went out of its way before the election to build the case that Diebold was stealing the election for the Republicans. Once that did not happen, HBO scaled back the airing of "Hacking the Election" and the New York Times went silent on the issue.
But, the whole issue is filled with problems. Just like in Georgia, paper ballot counting is fraught with errors that the computers won't make. All errors will be made by the voter at the front end. And even if we print paper ballots, who's to say the paper ballot can't be rigged to reflect something different from the electronic record, etc., etc. etc.
The left shot itself in the foot over voter fraud allegations that did not pan out. At the same time, I think a paper record of the voter's intention would be worth having, if only to give peace of mind. Perhaps not to be counted unless there are sufficient allegations made in a court of law that something was rotten in the precincts where the vote was cast.
After all, paper ballots are more likely prone to fraud than electronic ballots. I guess it makes sense to have paper records. But the more I think about it, the more I think it only causes more problems and raises more questions.
What do you think?
« Question and answer time: the Wes Clark thing. — Comments (50) | The Democrats, Counting Their Chickens — Comments (8) »
It's Common Sense, I Guess 40 Comments (0 topical, 40 editorial, 0 hidden) Post a comment »
Fraud didn't happen (who could say, do you have any way to verify results?) doesn't mean we leave the door wide open for it. From a computer security standpoint, e voting as carried out in the last election is a complete joke. First off, you do not let poll workers keep the machines at home for days before the election. Second, you do not allow Diebold to use Access databases for tabulation. Third, you must have some sort of verification of machine results.
Without at least those first three fixed, I wouldn't consider evoting and would not trust election results no matter how they turned out.
Human beings, who are almost unique in having the ability to learn from the experience of others, are also remarkable for their apparent disinclination to do so. - Douglas Adams
Using Access is as much a security concern as it is a reliability one. If I'm going to store each vote as a record, and have more than a million, or hundred thousand + records, there is no way I'd use Access. Access is a great personal database, but horrible at large scale apps and is highly prone to corruption.
Access also implies a Windows OS. A Window OS attached to a network is never secure. To have a central tabulation and each precint relying on a VBA Access database tabulation program is absolutely disgraceful. Anyone who has access to the computers OS and has limited technical ability could do whatever they wish to the vote tabulation and edit the audit table at the same time leaving no trace of their activity.
Human beings, who are almost unique in having the ability to learn from the experience of others, are also remarkable for their apparent disinclination to do so. - Douglas Adams
That both an electronic AND paper record of each vote would be wise. We've seen the trauma that too-close elections has caused since 2000, and such a belt-and-suspenders approach goes well with the recordkeeping and evidentiary rules government is imposing on business today. What's good for the goose...
to claim the absence of one is proof of attempted fraud, unless, of course, your side wins.
I completely agree. Here in the blue-state northeast we've been using lever machines for nigh on 60 years (they've just started to phase them out this past election here in Connecticut).
You go in the booth, pull a lever to draw the curtain closed, and then flip little switches for all the candidates you want to vote for. When you're done, you pull the lever back to open the curtain and record the votes. No paper trail. No one knows for sure if the votes are really counted. No one knows whether the machines are delivered with pre-loaded votes in them (as has been alleged many times, particularly in places like Philadelphia).
Yet I've never heard a Democrat complain about the lack of a paper trail on these machines, that have been used in largely Democratic regions of the country for decades.
I'm not necessarily opposed to paper trails. If people want them, then fine. I just think the argument the left has put forward for them has been a bit disingenuous.
...fraud committed by the manufacturers of voting machines (which they assume will benefit Republicans), then the generated paper trail can be just as fraudulent as the electronically-generated results, because it comes from the same raw data.
If they're concerned about hacking committed by bad guys in the network, then I'd like to know what voting machines are doing on the open network in the first place.
If they're concerned that electronic voting will actually succeed, then it makes sense to insist on a paper trail, because the paper trail attenuates the whole point of electronic voting (which is to reduce costs and also to cut error rates by an order of magnitude). Paper-based elections and elections run on unauditable machines are far more amenable to the kind of retail fraud that Democrats specialize in.
One presumes that the matter is largely academic because truly random errors will benefit neither party, whether they're in the range of 0.4% (electronic) or 4% (conventional). I think the Left is just being paranoid. As you said, allegations of fraud receive wide attention only when Democrats lose.
Has anyone examined the fact that retired people are overrepresented among election workers, in red districts as well as blue ones? And older people skew Democratic?
2004: Voters 60 and older favored Bush 54-46 -- the best he did in any age grouo -- and the 65+ went 52-47 for him.
2006: Republicans did better among the old than any other group.
In Ohio we have new machines where you can vote electronically. Once you are done and check your answers, the voting machine then prints your answers. You can see part of the printed copy through a window on the machine. I made sure the printed copy (as much of it as I could see) matched what I voted, and it did.
Without printed copy, the chances of fraud go up astronomically. And I'm sorry. Democrats have stolen more than one election. We have to be on guard.
The mechanism for detecting fraud is entirely dependent on the real-time visual "audit" performed by each voter, who presumably will demand to re-vote if she spots an error. I'm very confident that the quality of this "audit" will be low, and also will vary with the average education level of the district. And in any case, the number of people who actually do it will probably be vanishingly small.
The paper records will thus correlate with the electronic tabulation to some unknown degree. But then you have a pile of paper records, and recounting paper receipts is guaranteed to have far higher error rates than the electronic count. So what's the point? Unless you're Democratic and you want to change the count after the fact.
That's exactly where I'm coming from. Unless there is some credible proof that the machines were tampered with, why inject human error into the process?
Does anyone know if electronic machines store transaction journals with time-keyed HMACs? Or do they just store raw vote totals?
I would trust an electronic audit trail far more than a paper one.
You mean you'de like to be able to examine the source code of a machine that our democratic elections rely on? No dice, although a great idea.
Human beings, who are almost unique in having the ability to learn from the experience of others, are also remarkable for their apparent disinclination to do so. - Douglas Adams
Why wouldn't it matter? It would matter greatly to me if experts (without any stake invested) could verify the software. Now, I am left to trust the same people who decided that all votes cast by these machines should be tabulated by an Access VBA program. No thank you.
Human beings, who are almost unique in having the ability to learn from the experience of others, are also remarkable for their apparent disinclination to do so. - Douglas Adams
They're just tools. The tools CAN be make reliable applications or they can produce pure garbage. It depends more on the programmer(s) than the tools.
I'm no fan of Microsoft and would not use their products except they're "standardized" and supported by my IT groups. In fact I MUCH prefer Linux and open source tools. But open source doesn't automatically mean better, just different and usually less costly.
Socialism doesn't work. It looks nice on paper, but it's been tried and it's failed miserably every time (usually accompanied by widespread death and suffering).
Proud member of the V.R.W.C.
Enough to know that Access has fundamental flaws that make it both unsecure and unstable for such large scale operations. That doesn't have to be Open, use Oracle, MS SQL, anything you can give a degree of security to. A personal DB such as Access cannot be reasonably sucured unless sitting in a locked room on a computer with no network connection.
Although I will say the last App I developed for Access was Access 2000. Maybe they have fixed some of the problems, like being able to open the db in notepad and extract the password if the developer used linked tables or specified the password in code, data corruption, and record locking errors, but I doubt it. It is intended as a small scale personal db program, why would they change the scope that it works well in?
Human beings, who are almost unique in having the ability to learn from the experience of others, are also remarkable for their apparent disinclination to do so. - Douglas Adams
security of whatever server is used for the database platform. ANY system can be compromised if proper procedures aren't followed. Such a server SHOULD be in a locked room behind a firewalled network (or better with no external network access at all).
By the way, I've been taking your word that Deibold uses Access and VB as their vote tallying application, I assume you've checked that. Oracle may be better, more secure, more flexible, etc but it's also more $$$$$$$$$$ (plus $$$$$$ for support). County election boards DO have a budget to live within and cost benefit tradeoffs will be made.
I wouldn't call vote tallying a "large scale" application. In fact it's dead simple and not all that large on the precinct or county level. It gets a bit more complex as you move up to a state level if you were trying to track all the votes from all the precints, but the counties usually only feed totals to the state.
The user interface for the voting machines is a bit more complex and you have to make certain that a voter can't crash the system with any input, but that can be done with proper testing.
But all of this just supports my preference for optical scanned paper ballots.....
Socialism doesn't work. It looks nice on paper, but it's been tried and it's failed miserably every time (usually accompanied by widespread death and suffering).
Proud member of the V.R.W.C.
When I say large scale, I mean large table with high record counts. In my experience, large tables in Access are more prone to mysterious corruption. Then again, I started moving away from Access after Office 2K. For an Access table, I start getting weary at around 100K records, depending on the size and table design. Especially if the DB is split into a front-end back-end with linked tables or more than one machine connects to it.
Since we don't get to know such specifics, I can't say what the true scenario is. How exactly the actual machines connect to the tabulation program would interest me greatly. If they are networked and multiple machine are connecting to the machine for each vote, I'd be worried. That's just something no one but the manufacturers know.
That's why most security/tech people really want to see more info on the source code/system used. I have no doubt e voting can be made reliable and have accountability, but we have been shown so far that cost cutting at every corner has ruled the design of these machines. When they won't even add a cheap direct thermal printer for verification, what assurance do we have they went the extra mile to secure what little we do know about their systems?
Human beings, who are almost unique in having the ability to learn from the experience of others, are also remarkable for their apparent disinclination to do so. - Douglas Adams
I certainly would like to run open-source voting machines, but that's a different issue. What I asked was:
Do current voting machines keep a transaction journal as they go through the day? Or do they just record raw vote totals?
I'd want an electronic file with one time-stamped and HMAC'ed XML record for each voter transaction. Identical to the paper trail. Any reason not to do this?
They have a seperate table in the db where they write "Audit Records"
Human beings, who are almost unique in having the ability to learn from the experience of others, are also remarkable for their apparent disinclination to do so. - Douglas Adams
First let me say I really dislike both extreme sides on this issue. The far lefts assertion that voter fraud happened without any evidence or proof was criminal. It damaged our democracy and sowed division and discord in a time when we needed unity. The far right that asserted the machines were just fine and nothing could go wrong was equally bad. Their assertion flew in the face of facts and demonstrations to the contrary. It made the right look bad and lent credibility to the lefts claims.
Any consideration of the problem of electronic voting needs to hit the more general problem of how to insure that votes are properly recorded. If you look at this problem you rapidly come to the conclusion that any system that does not produce a tangible token allowing the voter to verify their vote is flawed.
Consider a voting machine, it can be based on any mechanism you prefer. The machine allows a voter to vote in privacy, making their choices and recording them in a manner inaccessible to voter after the ballot has been cast. The mechanics of the machine are unknown to the voter. The recording medium the machine uses is not available to the voter. The net effect is that the voter has to rely on the certification of the supervisor of elections that this machine records votes properly. Where I live not long ago we had our supervisor of elections removed by the governor for gross incompetence. Its at best rather tenuous to place that level of trust in a political office.
The voter is in effect reduced to taking the word of an appointee or politician that their votes are being correctly counted. Now even assuming the person in charge is honest the machines may just be better liars. The certification process for machines usually involves feeding them votes and making sure the count correctly. There is also an evaluation of the mechanism when the machines are initially purchased. Neither of these are adequate in any machine of reasonable complexity be it electronic or mechanical.
Voting machines in general are part of a class of devices called state machines. State machines in simple, are devices devices that assume discrete states and transition from one state to another based on the current state and a given input. It is very easy to include extra states that need a special input to reach and either transition out after a number of times of need a special input to transition out of.
The fact that the workings, the current state, and the output of means the voter has no way of knowing that they are being recorded accurately and in principle it becomes nearly impossible to post hoc verify that they were. A good example of the problems involved here would be a mechanical slot machine that was rigged to adjust its payout by how fill its hopper is. You have no idea how full the hopper is, and you have no way to check that the payout odds are as advertised.
The above applies to all closed system voting machines. It does not matter if they are pull the lever mechanical machines, Electronic touch screens, or if they are arrangements of Incan quipu.
Paper of course has its problems. There is the usual ballot stuffing and the reverse ballot loss, unreadable ballots and tampered ballots.
Having both creates a check system for the process.The question is if the goal is to make elections tamper proof or to just ignore the problems.
...at the point where votes are recorded. The only real answer to that is what you suggested, which is to directly count the untranslated gestures actually made by the voters (most simply, their handmade marks on pieces of paper).
Now in proposing this you're implicitly accepting a vote count that can't actually be "correct" or free from bias, except in a statistical sense. The margin of error with hand-tabulated paper ballots will be high (maybe 4 or 5 percent), variable by district, and possibly subject to systemic biases.
Wouldn't it then be better to randomly choose a small sample of voters from each district, hand-count their paper ballots, and then decide the election statistically, with a 95% degree of confidence? In a future Presidential election unburdened by the Electoral College, you could get a highly dependable result from perhaps 3500 voters, so long as the random function that chooses them is of extremely high quality.
My concern is over fraud, but by voting machine tampering. There is a secondary consideration that having a black box using complicated software thats proprietary opens up a can of worms on the litigation side.
The paper is there to provide a check. In 2000 the complaint was that people couldn't figure out how to fill in the ballot. This escalated to trying to count what normally would have been undervotes.
If an electronic machine records the votes and produces a paper record that the voter gets to see you have a higher threshold for ridiculous litigation and fraud. If the machine is operating properly the ballots will be able to be tallied by an optical reader for the recount. No hanging chads, no making an x in the oval instead of filling it in. The voter gets to see their ballot so if it isn't correct they can shout about it.
Hacking a voting machine has been shown to be ridiculously easy. This way you have to hack the voting machine the ballot box and the voter. Its still not out of the realm of possibility, but its much harder.
No matter how good sampling is would you want to tell people that the government isn't going to bother to count their vote ?
Personally, I think if we're going to use paper records, we shouldn't bother with the machines at all. We should just have big pieces of paper, with big boxes, where you get to mark the candidate you want with a big red marker. Then we should hand-count every last vote.
Either trust the machines or don't bother with them at all. I've always been fine either way, though I'd have a preference for making the machines work. That's only because I know enough about software engineering to know they can be made to work though.
--
It is much more important to kill bad bills than to pass good ones. -- Calvin Coolidge
Its a malicious people issue. If you had the machines tallying your vote on a big screen that kept a running count it would solve problem at the expense of a secret ballot.
We use the optical scan technology with a local secure scanner where I live and it is the way to go. We had one poor lady ahead of me that mis-voted and the scanner spit out here ballot and she had to have it voided and revote.
The thing is, not only is the system auditable, in that the paper record can be compared to the electronic record, but there is a date timestamp and a location pointer in the electronic record associated with each vote. Making it much harder to voting outside of legitimate voting hours.
As something with 20 years experience in Information Technology, I will say that the Diebold system is utter crap and should have never EVER been allowed to be used for conducting elections. The number of flaws in the Diebold system is unbelievable.
While HBO's Hacking Democracy has a nauseatingly pro-Dem bias, that does not undercut its central message, which is that it can be very easy to "hack" and election.
And there are obvious problems, one only has to look at FL-13 with the 13,000 under vote. The more you did into the story the worse it becomes.
It is essential that we have a secure voting system in this country and I see that goal a one that should be absolutely non-partisan. Hopefully the Dems will push for meaningful e-voting reform and hopefully the Repubs will keep the Dems honest and then back the reforms as well.
Here in Cook County/Chicago Illinois ,where I am a volunteer election judge, we use a dual system: optical scan verified as each person votes, and one Sequoia touchscreen machine mostly to be used by the blind. If you mess up the paper ballot the machine rejects it while you are there and we judges can give you a fresh ballot. The touchscreen produces a paper tape in a sealed cassete which you can read before it spools up into the box.
This solves a couple of problems, but not all. It works great if you buy an optical scanner for each of 2000 polling places. Expensive. If you bring the ballots to a central scanning station, you still have the messed up vote problem, just like the hanging chads. The electronic system is still iffy, but its geting better since many problems in the primary back in March.
Is to vote
We should all go the distance to make sure that there is no problems. Seriously who here would trust Microsoft with there vote?
have been very effective here in Alaska. They give the best of both worlds in that the voter hand marks a paper ballot which is then read, tabulated, and transmitted by the machine. That is especially helpful here where many polling places are in remote villages accessible only by air. The counting process that once took days now takes at most a few hours and there is a paper ballot trail for recounts and audits.
Of course, nothwithstanding the fact that they were bought by Democrat Lt. Gov. Fran Ulmer (the Lt. Gov's office runs elections here), the Democrats challenged them trying to cast doubts on Lisa Murkowski's defeat of Tony Knowles in the '04 Senate race.
On a side note, you'd think the Democrats would want voters to have some sort of receipt to keep their operatives from being defrauded in handing out the "walking around money."
In Vino Veritas
Is that it doesn't cost billions to implement. Coming up with the most expensive and least practical solution to a simple problem seemed to be the driving force behind electronic voting.
---
Underlying most arguments against the free market is a lack of belief in freedom itself. - Milton Friedman
marshal mcluhan, that is.
Having tasted a life wasted, I ain't ever going back again.
-E.V.
I do think a paper trail is always a good idea.
What I think the "voter suppression" people don't realize that there is no fool proof method of voting, all methods come with an error rate, and some votes are going to get miscounted. Humans are involved at some point in the proccess, and last time I checked humans weren't perfect.
Our system should seek to make sure only those eligible to vote are voting, that they are given a fair opportunity to cast ballots, and I would prefer some kind of paper record of their vote, even if something computerized or electronic was used.
Could you describe to me a "paper trail" that you think would work?
John
--------
Ethnic humor is part of human nature. The Dutch tell Belgian jokes. The Belgians tell French jokes. The French tell English jokes. The English tell Irish jokes. The Irish tell Irish jokes.
you have a Diebold type machine that is (theoretically) easy for the voter to use and understand, on which they initially answer the ballot questions. When they've finished voting, the press the button and print out an optical scanner record that clearly indicates their votes. They can review the optical scanner ballot to make sure it corresponds to what they thought they selected. Once they are certain of that they take it to the optical scanner that actually tallies the votes. Once every hour the system updates the electronic billboard at the far end of the room so everybody can see what the current vote tally is.
Problem solved.
I'm slowly coming to the conclusion that the biggest part of the answer is to make it illegal to report poll results the day of the election; maybe the day of the election plus the week before.
If it is illegal to report the poll results, the news media can't try to muck around with voter turnout. They also don't report absurd results like Casey wins with less than 1% of the PA vote counted and Santorum in the lead by 7%. I don't really care that their projection did accurately reflect the final vote, it is insane to make that kind of projection with so littel actual reported data.
Casey led in every single poll taken in the race, usually by 10-12 points. The exit polls on 11/7 confirmed a beat-down. Why eviscerate the First Amendment just to put a balm on Santorum's feelings?
There should be no projected winners, polls, early results on races on election day. What possible good can it do? Zero. How does any of this information give a person good information on making a choice?
The goal on election day is to allow anyone to vote based on their own thoughts based on whatever information they have. Early poll results offer no information on what may be the best choice for anyone.
If you argue early results do in fact give a voter good information or should just plain be allowed for free speech reasons, then early voters are still discriminated against because they are not privy to that information when they voted.
Everyone should go to the poll booth w/ the same opportunity to know the same information to make their decisions.
To address voter fraud couldn't a biotmetric (electric hand/fingerprint scan or retinal scan) system be used to guarantee one vote per person? The system wouldn't have to be tied to any database to identify any voter. It could just check against a new database populated w/ voters to assure one vote per person. It could also be able to be checked against criminal database to prevent convicted felons from voting.
If you most often find yourself arguing the exceptions rather than the rule you just might be on your own slippery slope to irrelevance. -CommonCents
To address voter fraud couldn't a biotmetric (electric hand/fingerprint scan or retinal scan) system be used to guarantee one vote per person? The system wouldn't have to be tied to any database to identify any voter. It could just check against a new database populated w/ voters to assure one vote per person. It could also be able to be checked against criminal database to prevent convicted felons from voting.
Biometric are good at proving who you are, but not so good at proving you haven't done something before. A fingerprint scanner is just a scanner. It's very easy to scan your finger in such a way that it doesn't match the last time you did it... as anyone who uses biometric security for anything knows. You also have 10 of those fingerprints just on your hands, and that doesn't include the ability to scan other stuff that isn't even fingerprints, or other non-matching parts of your fingers. There would also be a major voter intimidation factor here. It also wouldn't prevent ineligible people from voting, such as either legal or illegal aliens, under age people, or felons.
---
Underlying most arguments against the free market is a lack of belief in freedom itself. - Milton Friedman
Just found this
http://www.internetnews.com/bus-news/article.php/3646231
Summary is that the NIST(National Institute for Standards) will recommend decertifying all electronic voting machines that don't leave a paper trail.
Amongst the reasons they cite, are problems with recent elections and primaries and the very real possibility that single hacker could rig an election
I'm in the computer business so I know the technological issues involved and am not afraid of the electronic machines. The chance of "rigging" a vote is very low on a properly run electronic system. The chance of losing the vote count due to a catastrophic failure of a few components (like some nut smashing the machine with a hammer) is a little higher. Leaving a paper trail is a reasonable method of avoiding both potential problems.
For a model of what the receipt should look like, look at the lottery machines at the local gas stations. They are able to print secure verifiable tickets for millions of people on a daily basis. A similar system with text and bar codes for verification could be used for voting machines. It's not new technology, it's secure and it's available off the shelf.
They do need to use a bit of sense though. Thermal printers are NOT a good idea. The paper is more expensive, tends to tear easier, and is easily damaged to the point of being unreadable. Lottery machines use a durable paper and ink for a reason.
The procedures for using the printed vote have to make sense too. There have to be proper steps, IE: Vote, print the record, have the voter verifiy the printed copy matches the electronic display, accept the electronic vote and deposit the paper copy into a secure ballot box.
It's a bit complex (which means more prone to procedural errors) so I actually prefer the optical scan ballots. They are more prone to spoilage with overvotes, etc. but they eliminate several steps needed for the electronic machine with receipts.
Socialism doesn't work. It looks nice on paper, but it's been tried and it's failed miserably every time (usually accompanied by widespread death and suffering).
Proud member of the V.R.W.C.