I don't understand why this is a problem. Since candidates' names, being proper nouns, should not be translated in the first place. The only reason they would change is due to use of a different alphabet, in which case it is perfectly possible to print the name using Latin, Chinese, Arabic, etc characters on all ballot papers. There's also the solution used in India of including candidate photographs on the ballot paper which works even for the illiterate.
You also need to translate instructions and maybe referendums (I don't know if the latter is translated; if so, there's no way you could put the languages on the same ballot; even just adding English and Spanish would lead to a very big ballot).
But they can use a complex touch screen machine without assistance? With these being cheaper than ballot papers having suitably textured paper...
Touchscreens broaden the range of disabilities for which people can vote unassisted. For instance, someone who is paralyzed from the neck down would probably not be able to vote with a punch card or other paper ballot, but would certainly be able to vote using one of those mouth rods you see people use. Blind people can take in a pair of headphones and the computer would read the candidates to them and let them make a choice. (I don't know how actual systems work interfacewise, but one possible--I'd guess likely--idea is that the voter would just push a button after the candidate's name is read.)
So why bother with the machines in the first place?
Because they will improve the result proportionally to how much they are used. If half the votes are cast on machines, the overall error rate will be cut in half. If three quarters of the votes are cast on DREs, overall error rate will be quartered.
You also get all the other benefits such as much better ADA compatibility, etc.
Each machine would have a UPS, so that'd provide 15-20 minutes without power. If it were a standard system. I personally think a full-blown desktop would be overkill, and that a custom system could be built that would use less power. But the costs there might be too much. You wouldn't need an overly powerful system though, so grab an old Pentium or 486 from somewhere, which would draw even less power.
Just about the only thing that could screw with the system would be a prolonged electrical outage. (At which point the precinct resorts to the paper backup ballots on hand, which are also available for people who are intimidated by the e-voting system.)
You bring up interesting points. This is the one bit of my plan that I'm not exactly clear on. I'm not sure what is the best input method. I'm not convinced it's a touchscreen. When I started to give some thought to the e-voting problem (this was for an application essay in the couple months following the 2000 fiasco, which is what absolutely convinced me that a good e-voting system would be better than just about anything we have) I thought that a MAC machine style interface, with physical buttons alongside a display, would be a better interface than a touchscreen. I've had second thoughts about it since then and think that a touchscreen has significant advantages over that idea, but I don't trust them to work properly. I don't really want to given them a mouse, but I think maybe that's actually best. (Or a trackball.) You'd scare some people off from the electronic machines, but I think there should be paper ballots on hand anyway to satisfy those who don't want to use the electronic system.
First, I don't think communication is necessary. (If they are there they would be local; the system would absolutely not be connected to the internet.) Storing the data locally should work quite well. Though having links could allow the poll operators (I don't know what they are officially called) to have one machine to oversee the others to be able to check information such as ink levels for the printer, etc.
In either case, the physical computer would be locked up. In the same enclosure would be a UPS, so even if someone managed to unplug the cord on the outside, it would not interrupt service, but the poll operators could replace the plug and continue on.
demonlapin got it. The ballots stay at the polling place. It's actually illegal to give a receipt for who you voted for. Part of the reason is what you gave, part is outright vote selling, but the biggest one is your employer coming up to you and saying "vote for Bugs Bunny or I'll fire you; bring me the receipt to prove it". The latter is the main reason that votes were mandated to be anonymous.
Congratulations, you have found one of only two fundamental differences in our voting systems. Yes, for federal elections we only need one ballot. But there is no reason we could not have several. A grey ballot for the MP, green for Senator, etc.
Great, but now you need several times the manpower to count those votes. The "the US is bigger" argument doesn't work because number of vote counters goes up proportionately. But the number of vote counters won't go up with the number of races. We'd have the same number of people counting potentially over a dozen races as you do counting one race.
Disabled people don't want assistance; voting is a private matter and having someone mark down your choices for you feels like a violation. Electronic machines decrease the range of disabilities where assistance is necessary.
(Shoulda put this in my other response, but whatever...)
You seem to be confusing 'accuracy' and 'precision.' The precision of either a manual or an e-voting process is *always* one vote and the final vote tally would be 8,192 votes for either, to use your number.
Actually the chance that they'd both be the same is pretty small. Maybe I wasn't explicit enough earlier. I said 8,200 because it's close to 8,192 and it's a nice round number. Replace it with 8,189, or 8,196, or 8,199 if you'd like.
The *accuracy* of those two processes might be very different, however, for many of the reasons that people have provided here such as fraud, operator error, hardware failure, software bugs, or voter error.
I'll address each of these:
operator error: could be minimized by making the systems easy to use and providing training. In any case, it shouldn't effect the vote total.
hardware failure: have several systems at each polling place. You'd want them anyway. If one goes down, you still have the others. Having more than one go down is very unlikely. And I would suggest having standby paper ballots anyway, for people would would prefer to not use the computer. Each system should have a RAID setup with redundancy. If one hard drive goes out, take the computer out of operation. If the data sources disagree, you have the paper ballots to trace.
software bugs: with adequate auditing and testing, shouldn't be a problem. I would reccomend releasing the source to at least select, independent security analysts, professors, etc. to audit. Bind them with an NDA if you'd like (that allows the disclosure of security holes) so that commercial companies will be more likely to get into the mix.
voter error: very unlikely with confirmation screens. see my other post.
fraud: here's where it gets tricky. I don't see an e-voting system necessarily being less secure than any other method, I really don't. Keep the computer boxes locked up to prevent tampering with the physical machine. The machines should never be connected to the internet, and I would suggest a plan that they aren't even connected together. Each computer maintains its own total. After polls close, the election worker goes around to the computers and collects both a plaintext and signed vote total, and calls the board of elections. Read them off, the board verifies the signatures are correct, and totals them there.
If there are suspicious events, you can always fall back to the paper trail. And doing a random screen of precincts also would help.
Leaving out fraud for a moment, there is no reason to believe that the accuracy of an e-voting machine will be better or worse than a typical manual process.
Sure there is! Except for the very unlikely* event that all of a precinct's computers are down at once, every vote will be recorded exactly correctly. There's no equivalent of a mechanical or optical reader misreading or miscounting a card. Computers don't have a 0.5% chance of accidentally incrementing the wrong memory address for example.
Second, go look at one of our punch cards. Now verify that the holes you punched are for the candidates you wanted. Kinda hard, huh? Since they aren't labeled, and it's virtually impossible to see while it's in the holder. (I've never used an optically read ballot so don't know what they look like.)
*With 'my' system; just about the only event that I could forsee would be a prolonged computer outage. Sure, you could have all the computers fail, but this would be incredibly unlikely.
Leaving out fraud for a moment, there is no reason to believe that the accuracy of an e-voting machine will be better or worse than a typical manual process. When fraud is also considered, however, the e-voting machines seem much more prone to it due to their lack of a paper trail for accountability (which means there is no way to check their accuracy) and due to the potential for votes to be easily and fraudulently changed on a widespread scale via software manipulation of the system. Therefore, the overall accuracy of an e-voting machine may be *much* worse than one of those old-fashioned manual methods.
Which is why there needs to be a voter-verified paper trail. Granted, I omitted it in my initial post and it didn't come up in the one you're responding to, but I have talked about my commitment to VVPTs many times in both this and other discussions. Said paper trail sholud be counted in, say, 10% of precincts and the totals compared with the electronic totals. If they differ by a significant margin (I suggest the margin of error of the counting method or if the overall outcome is different), then recount the state.
This is, I think, one of those things that seems like a good idea, but isn't quite fair.
It reminds me of, say, the objections put forth to the draft exemptions (Vietnam war memorial) if you were in college that were removed because blacks complained that they were being disproportionatly sent over because they were on average poorer and thus less likely to attend college.
Similar deal here. Poorer people would be less able to follow the vote updates, and thus would be less susceptable to the above effect. Overall result is to give a slight favor to the parties that are more favorable with the upperclass.
One of the major objections people had with the media's coverage of the 2000 election was that channels called Florida's vote before the polls closed in the panhandle region (which is on central time), and people weren't sure what effect that had on the outcome.
This is the difficulty: you are comparing an imaginary system to a real one.
However, my "imaginary system" is very doable. I've given this quite a bit of thought, and would be able to describe such a system in a fair amount of detail.
And the qualification is there just to disqualify criticisms of Diebold, et. al., because we're talking about the concept of e-voting generally rather than a specific implementation. An implementation could, without a terribly strenuous effort, conform to all of my points.
No real electronic (or mechanical) voting system is or can be error-free, which is what you are claiming for your system. When you say that it would be an advantage to be able to say, "So-and-so GOT *exactly* 8192 votes" what you actually mean is, "THE SYSTEM COUNTED *exactly* 8192 votes, AND THERE IS NO WAY TO DO A RECOUNT." This is nothing but a false sense of accuracy.
Granted, I haven't mentioned a voter-verified paper trail on the path from my original post to here, but if you read the other responses I've put forth to objections over my original post, you'll see I think they are an absolute necessity.
Furthermore, a paper ballot contains far more information than an electronic ballot. To compare real systems, suppose a paper ballot contains a questionable mark that may or may not indicate a vote. The scrutineers can examine it for evidence of voter intent, and ask questions like, "How firm is it? Is it at all X-like, or just a dot or a single stroke? Does it centre on a circle or is it off to the edge?" and so on.
I see this as a very, very severe weakness. This is why the Florida recount took weeks and still was never complete; people staring at the ballots going "hmm, did the voter mean to punch out this chad, or not?" How is this a plus? I think there should be as little influence of the vote counters as possible. The more leeway you give them to determine if a vote was cast, the less accurate the final total.
The equivalent case in a touch-screen voting system is a voter whose sleeve brushes against the screen as they reach to make their vote. No one--not even the voter--will ever know that they didn't register the vote they intended.
Is it too hard to ask for verification at the end? Really... I'd be surprised if even current machines didn't do this. (And with a VVPT, there would even be a second opportunity to discard the ballot when looking at the paper version.)
Except that you have a degree of uncertainty with any measurement system; at least with paper ballots you have an opportunity to recount.
Hence the need for a voter-verified paper trail.
Even if the code is audited, you have no guarantee that the compiled code the machines are running is the same as the code that was audited. Diebold has been regularly caught running uncertified versions of their software on voting machines, for instance.
The provide a mechanism to ensure that. It wouldn't be hard. Have people take a MD5 hash or something to compare to an independently built version.
Sure it would. I said a voter-verifiable paper trail, remember. A machine prints out a ballot with your selections in plain text, which you then fold and put in the ballot box. Further, I trust the average election worker to be able to monitor and understand the security requirements to maintain a cardboard ballot box. I don'y trust those people to be able to manage an electronic voting system. You're quite right that a record that couldn't be verified by the voter is pointless. Further, a system that just recorded the votes in sequence on a paper tape could compromise the secrecy of the ballot.
Okay, I think I misunderstood you before. I thought you were referring to the electronic total. Which a VVPT doesn't help because the machine could print a different ballot than what it actually recorded, so you're putting your trust in the hands of the auditors just as much as you would without the VVPT.
On the other hand, you've gone to all the trouble to hire all those election workers to operate and monitor the polling stations. Why not have them stay for an extra hour to count ballots? It's not that difficult a task, and again it is something that is easily understood, monitored, and verified. Considering what it costs to buy, distribute, maintain, and repair the specialized hardware for an electronic voting system and the parallel paper handling to verify it, paying your election staff for an extra hour or two starts to look pretty cost-effective.
I never said that it'd be a cheap thing. The cost would end up being by far the biggest drawback of any otherwise decent e-voting plan. But as I said in another thread, I think the other benefits make e-voting worth going forward with; I don't want a "very good" election scheme, I want "as close to perfect as is possible to make it". And I think DREs are unquestionably closer to this ideal than any hand- or mechanically-counted system.
Accuracy: recount legislation is well developed to handle any close results where accuracy might be an issue
Again, do five recounts. Are you gonna have one result, or five?
As i said in another response: "I still think being able to say 'So-and-so got *exactly* 8,192 votes' is much better than 'So-and-so got 8,200 votes with a 95% confidance interval with a radius of 10', even if the next runner up is obviously behind."
Judgement: again, the legislation is well developed. Do some reading on the Canadian system. It's quite good.
It may be good enough, or even very good. But why not go to a system that is perfect? Enforced unambiguity.
Speed: are you joking? We had *very* fast results.
I'll concede the speed point. (Though electronic totals would probably be available within 30 minutes via the system I imagine, if not sooner.)
The question here is *should* it be done? Maybe not: you've got a hell's breakfast of disputable electonic and mechanical counting systems down south. Maybe you do need to go to a simple paper system that everyone can understand and thus *see* that it's working properly
I don't see why it shouldn't. It offers an improvement, and brings no real disadvantages besides cost. (And some challenges that I'm sure plenty of people would be more than willing to work on.) Elections are the absolute bedrock of a democracy, I see no reason to use anything but the absolute best system available. I think this is the bottom line. I don't want "good enough" or "very good"... I want "as close to perfect as we can possibly make it".
(Oh, and for the record, I'll take Canada's system before the mechanical systems we have now, and certainly before Diebold.)
I don't buy the speed argument at all. If you spend a year and a half campaigning - the current US federal campaign seems to have started last spring - surely you can wait a few hours for an official decision.
Because people are impatient. That's why TV stations rushed to scoop others with calling Florida, which is a small part of the mess that happened in that state. That's why news stations carry live updates with *predictions* of who will carry the state, and why they have "too close to call" in there when their exit polls aren't conclusive.
Remember, the election of President is probably the biggest regular, predictable political event that occurs in this country.
And how final is an instant result anyway? The vote in Florida took WEEKS to settle in 2000
Assuming that there was no fraud, the initial totals would be final. Fraud would be detected the following day or earlier, based on a random sample of comparing the electronic total to a total from paper ballots.
Florida is a perfect example too of one of the benefits of DREs. There aren't any hanging chads, there's no trying to guess the intent of the voter. It's recorded directly, without any ambiguity in both the electronic and paper versions. No holding ballots up to the light and going "did the voter punch this hole out?" Even if a paper recount was necessary, it'd just be "ok, this says "Gore, Al" so it goes in this pile, this one says "Bush, George" so it goes in this other pile.
In every voting system I know of except DREs, it'd be possible to arrive at this situation. Partly punch a chad out, put a small mark in an oval but not completely fill it, etc. But with a DRE, having a clear, unambiguous vote is enforced by the voting machine.
On the other hand, if the five are tightly distributed around a mean, then we're still in good shape.
I still think being able to say "So-and-so got *exactly* 8,192 votes" is much better than "So-and-so got 8,200 votes with a 95% confidance interval with a radius of 10", even if the next runner up is obviously behind.
Presuming the system generates a paper trail. Otherwise, all bets are off. Of course, to maintain anonymity and voter verifiability, the paper trail would probably take the form of a human-readable paper ballot printed and deposited in a sealed box...
First, I think that DREs *should* be required to have a paper trail. It'll give an assurance in the case of hacking or other fraud. Outside of fraud, I think that you can rest assured that the vote would be recorded correctly, especially if the code was audited by professors and other independent sources. It'd take a really really bad bug or a deliberate fraudulant attempt from within the company to do it otherwise. And anyway, a paper trail wouldn't add any assurances in that situation anyway....which could then be hand-counted to verify...wait, that sounds awfully familiar...
Except you wouldn't handcount all votes. Take a random selection of maybe 10% of the precincts to count. If the count differs from the electronic totals by more than the margin of error of the counting system, or if the two methods yield different outcomes of the elections, recount the state.
My 100% accuracy was in regards to the system counting the votes that it got correctly. Like if you have an optical system or a punch card system, the machine will occasionally misread the card going through and register the vote for the wrong person or not at all. By contrast, an e-voting system would never accidentally increment the wrong variable assuming the person writing the program has the least bit of knowledge. This is how the companies that make voting systems count their accuracy (like optical scanners are 99.something% accurate, punch card readers a little less) AFAIK, so that's how I counted evoting machines.
And BTW, I do think that we need to have a paper trail. After each election, take a random selection of precincts and do a count of the paper ballots. If they differ by an amount greater than the margin of error of the ballot count, or if the paper ballots and electronic totals tell different winners, then do a full count of the state.
4) Along with #2, the electronic machine can verify your votes and tell you if you overvoted (in which case you can't submit) or undervoted (in which case it just asks for confirmation).
1) Accuracy. I secure evoting system should be 100% accurate. Unless you happen to have more than 2^32-1 voters in your district all voting for the same person. Now look at Canada. Count the votes 5 times. Do you think you'll get one result, or five? I'm betting on the five. Humans make mistakes. Granted, they will probably be close, but there have been elections in the US (not presidential, but the point stands) decided by literally 12 votes in a large populated area. A couple states in the US in 2000 were, IIRC, decided by under 100 votes.
2) Along with that idea: judgement calls. Maybe the person made a stray mark and didn't notice; was it intended as a vote? You have to decide. With electronic voting, the system says "ok, here's who you voted for" and you can rest assured that the machine recorded it correctly. (We're talking a good system here, not a Diebold system.)
3) Speed. We're an impatient country. If we can be told the vote totals right after elections close, we're happier.
Ah, okay, I get it. I know the shirt and I know there's a Radiohead song called 2+2=5, and thought that it was based off of that. Didn't realize there was a programmatical (is that a word? if not, it is now...) reason. Thanks.
Slashdot Mirror
I don't understand why this is a problem. Since candidates' names, being proper nouns, should not be translated in the first place. The only reason they would change is due to use of a different alphabet, in which case it is perfectly possible to print the name using Latin, Chinese, Arabic, etc characters on all ballot papers. There's also the solution used in India of including candidate photographs on the ballot paper which works even for the illiterate.
You also need to translate instructions and maybe referendums (I don't know if the latter is translated; if so, there's no way you could put the languages on the same ballot; even just adding English and Spanish would lead to a very big ballot).
But they can use a complex touch screen machine without assistance? With these being cheaper than ballot papers having suitably textured paper...
Touchscreens broaden the range of disabilities for which people can vote unassisted. For instance, someone who is paralyzed from the neck down would probably not be able to vote with a punch card or other paper ballot, but would certainly be able to vote using one of those mouth rods you see people use. Blind people can take in a pair of headphones and the computer would read the candidates to them and let them make a choice. (I don't know how actual systems work interfacewise, but one possible--I'd guess likely--idea is that the voter would just push a button after the candidate's name is read.)
So why bother with the machines in the first place?
Because they will improve the result proportionally to how much they are used. If half the votes are cast on machines, the overall error rate will be cut in half. If three quarters of the votes are cast on DREs, overall error rate will be quartered.
You also get all the other benefits such as much better ADA compatibility, etc.
Each machine would have a UPS, so that'd provide 15-20 minutes without power. If it were a standard system. I personally think a full-blown desktop would be overkill, and that a custom system could be built that would use less power. But the costs there might be too much. You wouldn't need an overly powerful system though, so grab an old Pentium or 486 from somewhere, which would draw even less power.
Just about the only thing that could screw with the system would be a prolonged electrical outage. (At which point the precinct resorts to the paper backup ballots on hand, which are also available for people who are intimidated by the e-voting system.)
You bring up interesting points. This is the one bit of my plan that I'm not exactly clear on. I'm not sure what is the best input method. I'm not convinced it's a touchscreen. When I started to give some thought to the e-voting problem (this was for an application essay in the couple months following the 2000 fiasco, which is what absolutely convinced me that a good e-voting system would be better than just about anything we have) I thought that a MAC machine style interface, with physical buttons alongside a display, would be a better interface than a touchscreen. I've had second thoughts about it since then and think that a touchscreen has significant advantages over that idea, but I don't trust them to work properly. I don't really want to given them a mouse, but I think maybe that's actually best. (Or a trackball.) You'd scare some people off from the electronic machines, but I think there should be paper ballots on hand anyway to satisfy those who don't want to use the electronic system.
I also agree with you regarding the oversight.
First, I don't think communication is necessary. (If they are there they would be local; the system would absolutely not be connected to the internet.) Storing the data locally should work quite well. Though having links could allow the poll operators (I don't know what they are officially called) to have one machine to oversee the others to be able to check information such as ink levels for the printer, etc.
In either case, the physical computer would be locked up. In the same enclosure would be a UPS, so even if someone managed to unplug the cord on the outside, it would not interrupt service, but the poll operators could replace the plug and continue on.
demonlapin got it. The ballots stay at the polling place. It's actually illegal to give a receipt for who you voted for. Part of the reason is what you gave, part is outright vote selling, but the biggest one is your employer coming up to you and saying "vote for Bugs Bunny or I'll fire you; bring me the receipt to prove it". The latter is the main reason that votes were mandated to be anonymous.
Congratulations, you have found one of only two fundamental differences in our voting systems. Yes, for federal elections we only need one ballot. But there is no reason we could not have several. A grey ballot for the MP, green for Senator, etc.
Great, but now you need several times the manpower to count those votes. The "the US is bigger" argument doesn't work because number of vote counters goes up proportionately. But the number of vote counters won't go up with the number of races. We'd have the same number of people counting potentially over a dozen races as you do counting one race.
Disabled people don't want assistance; voting is a private matter and having someone mark down your choices for you feels like a violation. Electronic machines decrease the range of disabilities where assistance is necessary.
(Shoulda put this in my other response, but whatever...)
You seem to be confusing 'accuracy' and 'precision.' The precision of either a manual or an e-voting process is *always* one vote and the final vote tally would be 8,192 votes for either, to use your number.
Actually the chance that they'd both be the same is pretty small. Maybe I wasn't explicit enough earlier. I said 8,200 because it's close to 8,192 and it's a nice round number. Replace it with 8,189, or 8,196, or 8,199 if you'd like.
The *accuracy* of those two processes might be very different, however, for many of the reasons that people have provided here such as fraud, operator error, hardware failure, software bugs, or voter error.
I'll address each of these:
operator error: could be minimized by making the systems easy to use and providing training. In any case, it shouldn't effect the vote total.
hardware failure: have several systems at each polling place. You'd want them anyway. If one goes down, you still have the others. Having more than one go down is very unlikely. And I would suggest having standby paper ballots anyway, for people would would prefer to not use the computer. Each system should have a RAID setup with redundancy. If one hard drive goes out, take the computer out of operation. If the data sources disagree, you have the paper ballots to trace.
software bugs: with adequate auditing and testing, shouldn't be a problem. I would reccomend releasing the source to at least select, independent security analysts, professors, etc. to audit. Bind them with an NDA if you'd like (that allows the disclosure of security holes) so that commercial companies will be more likely to get into the mix.
voter error: very unlikely with confirmation screens. see my other post.
fraud: here's where it gets tricky. I don't see an e-voting system necessarily being less secure than any other method, I really don't. Keep the computer boxes locked up to prevent tampering with the physical machine. The machines should never be connected to the internet, and I would suggest a plan that they aren't even connected together. Each computer maintains its own total. After polls close, the election worker goes around to the computers and collects both a plaintext and signed vote total, and calls the board of elections. Read them off, the board verifies the signatures are correct, and totals them there.
If there are suspicious events, you can always fall back to the paper trail. And doing a random screen of precincts also would help.
Leaving out fraud for a moment, there is no reason to believe that the accuracy of an e-voting machine will be better or worse than a typical manual process.
Sure there is! Except for the very unlikely* event that all of a precinct's computers are down at once, every vote will be recorded exactly correctly. There's no equivalent of a mechanical or optical reader misreading or miscounting a card. Computers don't have a 0.5% chance of accidentally incrementing the wrong memory address for example.
Second, go look at one of our punch cards. Now verify that the holes you punched are for the candidates you wanted. Kinda hard, huh? Since they aren't labeled, and it's virtually impossible to see while it's in the holder. (I've never used an optically read ballot so don't know what they look like.)
*With 'my' system; just about the only event that I could forsee would be a prolonged computer outage. Sure, you could have all the computers fail, but this would be incredibly unlikely.
Leaving out fraud for a moment, there is no reason to believe that the accuracy of an e-voting machine will be better or worse than a typical manual process. When fraud is also considered, however, the e-voting machines seem much more prone to it due to their lack of a paper trail for accountability (which means there is no way to check their accuracy) and due to the potential for votes to be easily and fraudulently changed on a widespread scale via software manipulation of the system. Therefore, the overall accuracy of an e-voting machine may be *much* worse than one of those old-fashioned manual methods.
Which is why there needs to be a voter-verified paper trail. Granted, I omitted it in my initial post and it didn't come up in the one you're responding to, but I have talked about my commitment to VVPTs many times in both this and other discussions. Said paper trail sholud be counted in, say, 10% of precincts and the totals compared with the electronic totals. If they differ by a significant margin (I suggest the margin of error of the counting method or if the overall outcome is different), then recount the state.
This is, I think, one of those things that seems like a good idea, but isn't quite fair.
It reminds me of, say, the objections put forth to the draft exemptions (Vietnam war memorial) if you were in college that were removed because blacks complained that they were being disproportionatly sent over because they were on average poorer and thus less likely to attend college.
Similar deal here. Poorer people would be less able to follow the vote updates, and thus would be less susceptable to the above effect. Overall result is to give a slight favor to the parties that are more favorable with the upperclass.
One of the major objections people had with the media's coverage of the 2000 election was that channels called Florida's vote before the polls closed in the panhandle region (which is on central time), and people weren't sure what effect that had on the outcome.
This is the difficulty: you are comparing an imaginary system to a real one.
However, my "imaginary system" is very doable. I've given this quite a bit of thought, and would be able to describe such a system in a fair amount of detail.
And the qualification is there just to disqualify criticisms of Diebold, et. al., because we're talking about the concept of e-voting generally rather than a specific implementation. An implementation could, without a terribly strenuous effort, conform to all of my points.
No real electronic (or mechanical) voting system is or can be error-free, which is what you are claiming for your system. When you say that it would be an advantage to be able to say, "So-and-so GOT *exactly* 8192 votes" what you actually mean is, "THE SYSTEM COUNTED *exactly* 8192 votes, AND THERE IS NO WAY TO DO A RECOUNT." This is nothing but a false sense of accuracy.
Granted, I haven't mentioned a voter-verified paper trail on the path from my original post to here, but if you read the other responses I've put forth to objections over my original post, you'll see I think they are an absolute necessity.
Furthermore, a paper ballot contains far more information than an electronic ballot. To compare real systems, suppose a paper ballot contains a questionable mark that may or may not indicate a vote. The scrutineers can examine it for evidence of voter intent, and ask questions like, "How firm is it? Is it at all X-like, or just a dot or a single stroke? Does it centre on a circle or is it off to the edge?" and so on.
I see this as a very, very severe weakness. This is why the Florida recount took weeks and still was never complete; people staring at the ballots going "hmm, did the voter mean to punch out this chad, or not?" How is this a plus? I think there should be as little influence of the vote counters as possible. The more leeway you give them to determine if a vote was cast, the less accurate the final total.
The equivalent case in a touch-screen voting system is a voter whose sleeve brushes against the screen as they reach to make their vote. No one--not even the voter--will ever know that they didn't register the vote they intended.
Is it too hard to ask for verification at the end? Really... I'd be surprised if even current machines didn't do this. (And with a VVPT, there would even be a second opportunity to discard the ballot when looking at the paper version.)
Except that you have a degree of uncertainty with any measurement system; at least with paper ballots you have an opportunity to recount.
Hence the need for a voter-verified paper trail.
Even if the code is audited, you have no guarantee that the compiled code the machines are running is the same as the code that was audited. Diebold has been regularly caught running uncertified versions of their software on voting machines, for instance.
The provide a mechanism to ensure that. It wouldn't be hard. Have people take a MD5 hash or something to compare to an independently built version.
Sure it would. I said a voter-verifiable paper trail, remember. A machine prints out a ballot with your selections in plain text, which you then fold and put in the ballot box. Further, I trust the average election worker to be able to monitor and understand the security requirements to maintain a cardboard ballot box. I don'y trust those people to be able to manage an electronic voting system. You're quite right that a record that couldn't be verified by the voter is pointless. Further, a system that just recorded the votes in sequence on a paper tape could compromise the secrecy of the ballot.
Okay, I think I misunderstood you before. I thought you were referring to the electronic total. Which a VVPT doesn't help because the machine could print a different ballot than what it actually recorded, so you're putting your trust in the hands of the auditors just as much as you would without the VVPT.
On the other hand, you've gone to all the trouble to hire all those election workers to operate and monitor the polling stations. Why not have them stay for an extra hour to count ballots? It's not that difficult a task, and again it is something that is easily understood, monitored, and verified. Considering what it costs to buy, distribute, maintain, and repair the specialized hardware for an electronic voting system and the parallel paper handling to verify it, paying your election staff for an extra hour or two starts to look pretty cost-effective.
I never said that it'd be a cheap thing. The cost would end up being by far the biggest drawback of any otherwise decent e-voting plan. But as I said in another thread, I think the other benefits make e-voting worth going forward with; I don't want a "very good" election scheme, I want "as close to perfect as is possible to make it". And I think DREs are unquestionably closer to this ideal than any hand- or mechanically-counted system.
Accuracy: recount legislation is well developed to handle any close results where accuracy might be an issue
Again, do five recounts. Are you gonna have one result, or five?
As i said in another response: "I still think being able to say 'So-and-so got *exactly* 8,192 votes' is much better than 'So-and-so got 8,200 votes with a 95% confidance interval with a radius of 10', even if the next runner up is obviously behind."
Judgement: again, the legislation is well developed. Do some reading on the Canadian system. It's quite good.
It may be good enough, or even very good. But why not go to a system that is perfect? Enforced unambiguity.
Speed: are you joking? We had *very* fast results.
I'll concede the speed point. (Though electronic totals would probably be available within 30 minutes via the system I imagine, if not sooner.)
The question here is *should* it be done? Maybe not: you've got a hell's breakfast of disputable electonic and mechanical counting systems down south. Maybe you do need to go to a simple paper system that everyone can understand and thus *see* that it's working properly
I don't see why it shouldn't. It offers an improvement, and brings no real disadvantages besides cost. (And some challenges that I'm sure plenty of people would be more than willing to work on.) Elections are the absolute bedrock of a democracy, I see no reason to use anything but the absolute best system available. I think this is the bottom line. I don't want "good enough" or "very good"... I want "as close to perfect as we can possibly make it".
(Oh, and for the record, I'll take Canada's system before the mechanical systems we have now, and certainly before Diebold.)
I don't buy the speed argument at all. If you spend a year and a half campaigning - the current US federal campaign seems to have started last spring - surely you can wait a few hours for an official decision.
Because people are impatient. That's why TV stations rushed to scoop others with calling Florida, which is a small part of the mess that happened in that state. That's why news stations carry live updates with *predictions* of who will carry the state, and why they have "too close to call" in there when their exit polls aren't conclusive.
Remember, the election of President is probably the biggest regular, predictable political event that occurs in this country.
And how final is an instant result anyway? The vote in Florida took WEEKS to settle in 2000
Assuming that there was no fraud, the initial totals would be final. Fraud would be detected the following day or earlier, based on a random sample of comparing the electronic total to a total from paper ballots.
Florida is a perfect example too of one of the benefits of DREs. There aren't any hanging chads, there's no trying to guess the intent of the voter. It's recorded directly, without any ambiguity in both the electronic and paper versions. No holding ballots up to the light and going "did the voter punch this hole out?" Even if a paper recount was necessary, it'd just be "ok, this says "Gore, Al" so it goes in this pile, this one says "Bush, George" so it goes in this other pile.
In every voting system I know of except DREs, it'd be possible to arrive at this situation. Partly punch a chad out, put a small mark in an oval but not completely fill it, etc. But with a DRE, having a clear, unambiguous vote is enforced by the voting machine.
On the other hand, if the five are tightly distributed around a mean, then we're still in good shape.
...which could then be hand-counted to verify...wait, that sounds awfully familiar...
I still think being able to say "So-and-so got *exactly* 8,192 votes" is much better than "So-and-so got 8,200 votes with a 95% confidance interval with a radius of 10", even if the next runner up is obviously behind.
Presuming the system generates a paper trail. Otherwise, all bets are off. Of course, to maintain anonymity and voter verifiability, the paper trail would probably take the form of a human-readable paper ballot printed and deposited in a sealed box...
First, I think that DREs *should* be required to have a paper trail. It'll give an assurance in the case of hacking or other fraud. Outside of fraud, I think that you can rest assured that the vote would be recorded correctly, especially if the code was audited by professors and other independent sources. It'd take a really really bad bug or a deliberate fraudulant attempt from within the company to do it otherwise. And anyway, a paper trail wouldn't add any assurances in that situation anyway.
Except you wouldn't handcount all votes. Take a random selection of maybe 10% of the precincts to count. If the count differs from the electronic totals by more than the margin of error of the counting system, or if the two methods yield different outcomes of the elections, recount the state.
My 100% accuracy was in regards to the system counting the votes that it got correctly. Like if you have an optical system or a punch card system, the machine will occasionally misread the card going through and register the vote for the wrong person or not at all. By contrast, an e-voting system would never accidentally increment the wrong variable assuming the person writing the program has the least bit of knowledge. This is how the companies that make voting systems count their accuracy (like optical scanners are 99.something% accurate, punch card readers a little less) AFAIK, so that's how I counted evoting machines.
And BTW, I do think that we need to have a paper trail. After each election, take a random selection of precincts and do a count of the paper ballots. If they differ by an amount greater than the margin of error of the ballot count, or if the paper ballots and electronic totals tell different winners, then do a full count of the state.
I agree. Accuracy comes before speed. But if you can have both, I can see how this would be a desirable outcome.
Oh, and one more I forgot:
4) Along with #2, the electronic machine can verify your votes and tell you if you overvoted (in which case you can't submit) or undervoted (in which case it just asks for confirmation).
I can give you a few reasons.
1) Accuracy. I secure evoting system should be 100% accurate. Unless you happen to have more than 2^32-1 voters in your district all voting for the same person. Now look at Canada. Count the votes 5 times. Do you think you'll get one result, or five? I'm betting on the five. Humans make mistakes. Granted, they will probably be close, but there have been elections in the US (not presidential, but the point stands) decided by literally 12 votes in a large populated area. A couple states in the US in 2000 were, IIRC, decided by under 100 votes.
2) Along with that idea: judgement calls. Maybe the person made a stray mark and didn't notice; was it intended as a vote? You have to decide. With electronic voting, the system says "ok, here's who you voted for" and you can rest assured that the machine recorded it correctly. (We're talking a good system here, not a Diebold system.)
3) Speed. We're an impatient country. If we can be told the vote totals right after elections close, we're happier.
So they don't overbook the auditorium they are in? I don't see anything that says that you need to send in even an address. (Though maybe I missed it)
Santa Claus: Kindly old elf, or CIA spook?
(mirror here and maybe here))
Ah, okay, I get it. I know the shirt and I know there's a Radiohead song called 2+2=5, and thought that it was based off of that. Didn't realize there was a programmatical (is that a word? if not, it is now...) reason. Thanks.
No, 2+2=5 for large values of 2. 2+2=3 for small values.
"You can make 2+2 output 5 without using a messed system."
What? Are you serious? If so, how?