Well, I cannot respond to the issue of "equal importance" which is really a reference to equal protection. Every voter gets a chance to vote. But vote tallying is a different matter, mostly due to the statistical methods used.
IOW: The constitution (under current interpretation) enforces your right to cast a ballot. It also enforces a reasonably accurate count. That is, the constitution is not concerned with your vote so much as it is concerned with accurate race results. Which means that as long as the aggregate results are accurate, your specific vote actually doesn't "matter"!
That is not my position, but is standard procedure. And even if you dislike that policy, I am not in a position to change it.
Run several month long compute jobs on a plus-thousand node cluster (with the same data set) and get back to me when you discover exact reproducibility every time. There's always error. Always.
It's just that the error is usually so small nobody cares.
Let me put it to you this way: if I drop an egg a million times, will you argue that statistics mandates that there's no way the egg will fall every time?
As long as that dropped egg doesn't collapse back into itself to form a fully coherent entity - like Humpty put back together again - then perhaps we do live in the same universe.:p
OK. I have already responded to the issue of the differing state laws and election regulation which add a great deal of unnecessary complexity to the problem of accurate vote tallies.
On to your other point: that is, the issue of deterministic results from computing. I have a great deal of experience with large batch and MPI clustering; most of it running some variant of linux, but also SGIs and Suns. I have never once seen a very large job come back without some sort of error. Individual nodes crash during compute, memory (even good ECC RAM) flips bits on occasion, but most of the time the error is just not discernible. That is, there was an error somewhere but the input data set is so large that determining exactly which output was in error after concatenation of results is impossible. So, the scientists I work for just work out the margin of error and make certain their results fit within that margin. When you're dealing with hundreds of thousands of accumulated CPU time for a run job, it's just not possible to track down and verify every event.
These voting machines are the same deal. With a whole bunch of added complexity due to the state regs and such. Further, most everyone here arguing for deterministic results forgets the necessity for anonymous voting - which *really* complicates matters (I should have made that point earlier). Using a bank ATM as an analogy (which you did not do, but many others did) ignores the difference between how people use bank ATMs vs. how people vote. ATM machines are directly authenticated before a user is allowed access to their account. A track record is stored on paper within the ATM. The bank hires auditors to go over their books. And regulatory agencies hire auditors to audit the entire banking industry. And with all that - still money is lost by banks. A lot of money. But banks don't care - as long as the lost money is not directed toward a single (or set of) accounts indicating fraud. Because banks know that a certain statistical variance across all their accounts will happen no matter what they do.
Margin of error again.
But voting systems add the complexity of anonymity - which *really* throws a wrench into accurate vote tallies, because while one can determine how many registered voters actually voted, one can not determine a range of other questions about voter intent after the fact. It is not an easy problem. Just saying "computers are deterministic, just write correct software and we'll get an accurate vote count down to the individual voter" is impossible for all these reasons: thermodynamics of very large systems; anonymity of the US voting procedure; legal and regulatory differences between counties and states; and, of course, the most recent problem of certain voting machines that lack verifiable paper trails (not a problem in this case, thankfully).
Anyway, I'm gonna go eat my dinner. Thanks for the discussion. Glad it didn't degenerate into a pissed off flame war.:)
Are you taking a page from the BOFH and blaming neutrinos or something?
I give up. If you can't recognize that statistical variance is a fundamental property of nature, then we must live in two vastly different universes. Could our universe have some of your free energy?
Nope. It is impossible to create any physical system that exists outside the boundaries of probability. Which means that there is always a margin of error. Always. Thermodynamics and all that.
However, there are specific reasons why the US system is particularly prone to error. Especially when dealing with large state and national races where differing regional elections laws, differing voting systems, and tabulation rules, make *extremely* accurate counting absolutely impossible. Which is why elections officials care more about outcomes than individual vote counts.
IIRC, Avi Rubin talked about the probability issue during his interview on C-SPAN. Here is his web page, there are links to.mov files of that interview available there.
There are several reasons for this. One is just plain probability. The greater the numbers the more likely for error. This is a purely theoretical argument about physical systems and margin of error and does not take into account the specifics of voting systems.
When speaking to the US voting system, the situation gets far more complex. There is a complex interaction between individual voters and the interface used to select candidates. Next, there is the interaction between anywhere from a few to potentially millions of voting machines deployed across disparate geographic boundaries. All these systems must then interlink to central tabulators within each state. For national elections the Associated Press then handles centralized tabulation for nonofficial election results reported to the press.
All of this introduces the potential for error well beyond the issue of a single voting machine tabulation error. You're not talking about one system, but a highly complex interaction of many systems, manufactured by different vendors, all under differing state authority and regulation. It's a nightmare.
1/36 is 3%. But holding up a 3% error rate because of one claimed error is just plain nutty. Someone made a mistake. Look into it. Stop freaking out over what is -- in reality -- the claim of a single candidate we cannot verify. Tear the machine apart and check the paper tally. That will help clear the dust more than the 300-400 uninformed yet angry comments this story will generate.
1) What caused this error, and could the problem be systemic?
Oh yeah. But you won't get the answer to that question from a newspaper article quoting the candidate. Which is why (as I said repeatedly) the elections commissioner must take apart the machine(s) and check the vote tally as recorded on the paper reel. Which is exactly how the particular brand of voting machine used within that county works.
I predict that will happen soon enough, and then we'll all learn it was due to some stupid mistake; not an example of some grand conspiracy.
You say it is is probably not an example of electronic voter fraud, but in fact you have zero data at this point to make that assertion.
That's because there is no further data beyond the statement of that candidate. As soon as election officials open the unit(s) up and check the paper tally, there will be real facts available. I just happen to think that in an election this small the error is most likely human error on the part of elections officials, or possibly the voter himself. It could also be a machine malfunction.
Please note that the machines used by this county actually have an internal paper trail. You're getting all riled up over nothing. Because there is a way to verify this election, and the county election board will do just that.
Fine point. I'm happy to have the machine count checked against the paper trail stored within these machines. But I don't think *this* example is indicative of rampant fraud through electronic voting machines. I think it's a minor local screw up that ought to be verified.
It's just slashdot (and reddit, and dailykos, etc etc etc) looking for reasons to get people riled up and draw ad revenue. If you want to freak out about voting machine problems, check out the Princeton Diebold study.
You didn't RTFA, did you? The machines contain a roll of paper similar to what is used in a cash register. These are used to verify vote totals and then signed by the county elections commissioner prior to the Secretary of State certifying the state vote. Other, fully electronic, systems record that vote tallies upon a smart card. It is this unverifiable system that computer security experts like Avi Ruben have been so concerned about.
Voting machines (or processes) will never reliably count one hundred million votes. Or one hundred thousand. There is always a margin of error. Which is why election officials care about race outcomes and not individual votes. However, in this case the vote totals are small and obviously in error. So, open the machine and count paper trail.
But IMO, this is not the instance to freak out over. I highly doubt this is a case of voter fraud. More likely it is a case of election official error, or possibly a machine malfunction.
So accuracy in counting votes only matters when it could change the outcome of an election? Bullshit.
Yup. That's how election officials do this sort of thing. In a case where so few votes are counted, one might reasonably argue: Why was there not an accurate tally? It's a good question, and one local election officials should investigate. But when scaling up an election to even just hundreds of thousands in a medium sized county election, it becomes impossible to count _every_ vote. It's just a statistical impossibility.
When confronted with such large numbers, it has become standard practice for election officials to concern themselves not with each individual vote, but with verifying outcomes for any particular race. That is, what matters is the outcome, not specific votes in the process.
Is that bad policy? I don't think so. Perhaps you disagree. *shrug*
We're talking about a local county election with a sum total of 36 votes cast. Clearly there was an error of some sort. Which brings up two fundamental questions all election officials must ask:
1) Did this error change the outcome of a race? That is the first consideration, because if it didn't then the severity of the error is vastly reduced.
2) If this error changed the outcome of a race, was it intentional? That is, was the outcome of democracy subverted, and done so with fraudulent intent?
I think (but don't know) that the answer to those two questions will ultimately be "NO". That is, the error did not affect the outcome of his loss - though the error might have impacted the necessity for a runoff. And, further, it is highly unlikely that for a race this small anyone would have been actively engaged in voter fraud. Certainly, if these results are the result of fraud, it is almost certainly not due to party involvement.
It may be a justifiable fear that someone might perpetrate a nationwide fraud using unverifiable electronic voting machines. But this example does not support that fear. Mostly because there is a verifiable paper trail within the machines, and the race is too small for organized fraud to be worth the trouble.
Which is why the country election commissioner should take the machine(s) apart and check the vote tape. It is only 36 votes, so it shouldn't be hard to do. I am simply pointing out that this example is more likely error than fraud.
Had his vote, and the votes he assumes had been cast for him (because his friends said they did), he still wouldn't have received enough votes to win the election. Further, it's not clear he would have received even enough votes to change the *outcome* of the election (there will be a runoff due to two other candidates having won the same vote count).
Yes, the county should take the voting machines apart and verify the printed tape count. No, this incident is probably not an example of electronic voter fraud.
This is different because in all of your examples you are theorizing some sort of service which is provisioned in a manner that is tightly controlled by specific companies and highly regulated by government.
Cable service is not regulated by the government? Last I checked, the government regulated cable television prices, service, and are even debating the regulation of content. Cable television and Internet companies lay cable straight to their customers' homes across public ways just as is electricity, water, and telephone service (or metered gas, for that matter). And, just like WAP, cellular telephone service is transmitted using radiowaves - yet no one argues that it wouldn't be theft to use another person's cell phone without their consent.
So, I'll ask a different question: If it is improper to use someone else's cell phone service without their knowledge or consent, but proper to use someone else's WAP point - also without their knowledge or consent - then what is the distinction between the two. According to you that separation would appear to be whether the service is "controlled by specific companies and highly regulated by the government." I fail to see how having a corporate charter makes any difference - a private individual has the same property rights as does a corporation. And the government IS enforcing regulation through the criminal court system.
The topic at hand is a person purchasing and implementing a device aimed at making it more convenient to use the internet service he or she has agreed to pay for, be it cable/dsl/[whatever-controlled-form-of-high-speed- access].
That's funny. I thought the topic was about a person in Singapore being convicted and jailed due to having used a neighbor's wireless internet without permission. I am arguing that the judge's decision was proper.
Finally, WRT: ease of use - or the "making it more convenient to use the internet service" arguement. might I point out that WAP routers are intended to make Internet access easier for the *owners* of the service (as you say). But it does not exist to make it easier for *strangers* to use a service they have neither paid to use nor have permission to use by those who did pay.
If your neighbor has an apple tree that has branches that are growing over your fence into your backyard, are you allowed to pick the apples? Or is that theft as well?
That's a good question. Courts would probably break it down thusly, though I'm just guessing:
a) Apples on or near adjoining property could not be picked from the tree by person A, if it is deemed owned by person B.
b) Person A's apples which have fallen by no over action onto person B's property, have likely _transferred ownership_ to person B. But Apples still attached to Person A's apple tree are still owned by him.
This would be my guess. But suppose we assume that apple's hanging over into Person B's property are somehow now owned by Person B. Using this analogy, does that mean that it is proper to tap a buried electricity cable which happens to cross Person B's property (or adjoins the property) to Person A's main junction? Siphoning off someone else's metered service is still theft, is it not?
[...]which is the notion that it is deemed wrong to use an access point in its apparently intended manner.
By this logic it is "intended" for a non-owner to use someone else's electricity. Or water. Or POTS service. Or even cell phone service.
wireless internet is simply a physical carrier. Issues of open access and whether security restrictions against unauthorized use were imposed (a lock installed on a front door) do not negate the principle of a private resource. That is, just because I left my door unlocked does not give you right to enter my home and take something of mine as if it were yours.
This is no different. The technology may have changed, but the principle of private ownership remains the same.
I'm trying to imagine an electricity hookup capable of transmitting its availability and automagically hooking up to provide electricity to a nearby house/car/man-in-the-street. That's the difference you asked for.
Tesla did it seventy or so years ago. But the point behind the analogy is that all of them are metered services sold to a specific individual. That the physical transmission medium for wireless Internet is different from electricity, eater, or telephone service does not matter. Do you argue the that you should have free reign to use a neighbor's cell phone service simply because the transmission medium is wireless?
Slashdot Mirror
OK, so I will respond to this on two levels:
1) Error creeps in and you might not even notice it. Corollary: Shit Happens.
2) It sounds like you're doing some seriously cool research. !!! Are you involved with Blue Brain or something?
Well, I cannot respond to the issue of "equal importance" which is really a reference to equal protection. Every voter gets a chance to vote. But vote tallying is a different matter, mostly due to the statistical methods used.
IOW: The constitution (under current interpretation) enforces your right to cast a ballot. It also enforces a reasonably accurate count. That is, the constitution is not concerned with your vote so much as it is concerned with accurate race results. Which means that as long as the aggregate results are accurate, your specific vote actually doesn't "matter"!
That is not my position, but is standard procedure. And even if you dislike that policy, I am not in a position to change it.
Run several month long compute jobs on a plus-thousand node cluster (with the same data set) and get back to me when you discover exact reproducibility every time. There's always error. Always.
It's just that the error is usually so small nobody cares.
Let me put it to you this way: if I drop an egg a million times, will you argue that statistics mandates that there's no way the egg will fall every time?
:p
:)
As long as that dropped egg doesn't collapse back into itself to form a fully coherent entity - like Humpty put back together again - then perhaps we do live in the same universe.
OK. I have already responded to the issue of the differing state laws and election regulation which add a great deal of unnecessary complexity to the problem of accurate vote tallies.
On to your other point: that is, the issue of deterministic results from computing. I have a great deal of experience with large batch and MPI clustering; most of it running some variant of linux, but also SGIs and Suns. I have never once seen a very large job come back without some sort of error. Individual nodes crash during compute, memory (even good ECC RAM) flips bits on occasion, but most of the time the error is just not discernible. That is, there was an error somewhere but the input data set is so large that determining exactly which output was in error after concatenation of results is impossible. So, the scientists I work for just work out the margin of error and make certain their results fit within that margin. When you're dealing with hundreds of thousands of accumulated CPU time for a run job, it's just not possible to track down and verify every event.
These voting machines are the same deal. With a whole bunch of added complexity due to the state regs and such. Further, most everyone here arguing for deterministic results forgets the necessity for anonymous voting - which *really* complicates matters (I should have made that point earlier). Using a bank ATM as an analogy (which you did not do, but many others did) ignores the difference between how people use bank ATMs vs. how people vote. ATM machines are directly authenticated before a user is allowed access to their account. A track record is stored on paper within the ATM. The bank hires auditors to go over their books. And regulatory agencies hire auditors to audit the entire banking industry. And with all that - still money is lost by banks. A lot of money. But banks don't care - as long as the lost money is not directed toward a single (or set of) accounts indicating fraud. Because banks know that a certain statistical variance across all their accounts will happen no matter what they do.
Margin of error again.
But voting systems add the complexity of anonymity - which *really* throws a wrench into accurate vote tallies, because while one can determine how many registered voters actually voted, one can not determine a range of other questions about voter intent after the fact. It is not an easy problem. Just saying "computers are deterministic, just write correct software and we'll get an accurate vote count down to the individual voter" is impossible for all these reasons: thermodynamics of very large systems; anonymity of the US voting procedure; legal and regulatory differences between counties and states; and, of course, the most recent problem of certain voting machines that lack verifiable paper trails (not a problem in this case, thankfully).
Anyway, I'm gonna go eat my dinner. Thanks for the discussion. Glad it didn't degenerate into a pissed off flame war.
Are you taking a page from the BOFH and blaming neutrinos or something?
I give up. If you can't recognize that statistical variance is a fundamental property of nature, then we must live in two vastly different universes. Could our universe have some of your free energy?
Nope. It is impossible to create any physical system that exists outside the boundaries of probability. Which means that there is always a margin of error. Always. Thermodynamics and all that.
.mov files of that interview available there.
However, there are specific reasons why the US system is particularly prone to error. Especially when dealing with large state and national races where differing regional elections laws, differing voting systems, and tabulation rules, make *extremely* accurate counting absolutely impossible. Which is why elections officials care more about outcomes than individual vote counts.
IIRC, Avi Rubin talked about the probability issue during his interview on C-SPAN. Here is his web page, there are links to
He's been reading up on the issue.
There are several reasons for this. One is just plain probability. The greater the numbers the more likely for error. This is a purely theoretical argument about physical systems and margin of error and does not take into account the specifics of voting systems.
When speaking to the US voting system, the situation gets far more complex. There is a complex interaction between individual voters and the interface used to select candidates. Next, there is the interaction between anywhere from a few to potentially millions of voting machines deployed across disparate geographic boundaries. All these systems must then interlink to central tabulators within each state. For national elections the Associated Press then handles centralized tabulation for nonofficial election results reported to the press.
All of this introduces the potential for error well beyond the issue of a single voting machine tabulation error. You're not talking about one system, but a highly complex interaction of many systems, manufactured by different vendors, all under differing state authority and regulation. It's a nightmare.
. ..
1/36 is 3%. But holding up a 3% error rate because of one claimed error is just plain nutty. Someone made a mistake. Look into it. Stop freaking out over what is -- in reality -- the claim of a single candidate we cannot verify. Tear the machine apart and check the paper tally. That will help clear the dust more than the 300-400 uninformed yet angry comments this story will generate.
1) What caused this error, and could the problem be systemic?
Oh yeah. But you won't get the answer to that question from a newspaper article quoting the candidate. Which is why (as I said repeatedly) the elections commissioner must take apart the machine(s) and check the vote tally as recorded on the paper reel. Which is exactly how the particular brand of voting machine used within that county works.
I predict that will happen soon enough, and then we'll all learn it was due to some stupid mistake; not an example of some grand conspiracy.
Yes. And, in fact, I *did* say in my top post that verifying the paper count is appropriate, and within every comment reply.
You say it is is probably not an example of electronic voter fraud, but in fact you have zero data at this point to make that assertion.
That's because there is no further data beyond the statement of that candidate. As soon as election officials open the unit(s) up and check the paper tally, there will be real facts available. I just happen to think that in an election this small the error is most likely human error on the part of elections officials, or possibly the voter himself. It could also be a machine malfunction.
But organized fraud? Highly doubtful.
Please note that the machines used by this county actually have an internal paper trail. You're getting all riled up over nothing. Because there is a way to verify this election, and the county election board will do just that.
Fine point. I'm happy to have the machine count checked against the paper trail stored within these machines. But I don't think *this* example is indicative of rampant fraud through electronic voting machines. I think it's a minor local screw up that ought to be verified.
It's just slashdot (and reddit, and dailykos, etc etc etc) looking for reasons to get people riled up and draw ad revenue. If you want to freak out about voting machine problems, check out the Princeton Diebold study.
This story, by comparison, is a big fat nothing.
You didn't RTFA, did you? The machines contain a roll of paper similar to what is used in a cash register. These are used to verify vote totals and then signed by the county elections commissioner prior to the Secretary of State certifying the state vote. Other, fully electronic, systems record that vote tallies upon a smart card. It is this unverifiable system that computer security experts like Avi Ruben have been so concerned about.
Voting machines (or processes) will never reliably count one hundred million votes. Or one hundred thousand. There is always a margin of error. Which is why election officials care about race outcomes and not individual votes. However, in this case the vote totals are small and obviously in error. So, open the machine and count paper trail.
But IMO, this is not the instance to freak out over. I highly doubt this is a case of voter fraud. More likely it is a case of election official error, or possibly a machine malfunction.
So accuracy in counting votes only matters when it could change the outcome of an election? Bullshit.
Yup. That's how election officials do this sort of thing. In a case where so few votes are counted, one might reasonably argue: Why was there not an accurate tally? It's a good question, and one local election officials should investigate. But when scaling up an election to even just hundreds of thousands in a medium sized county election, it becomes impossible to count _every_ vote. It's just a statistical impossibility.
When confronted with such large numbers, it has become standard practice for election officials to concern themselves not with each individual vote, but with verifying outcomes for any particular race. That is, what matters is the outcome, not specific votes in the process.
Is that bad policy? I don't think so. Perhaps you disagree. *shrug*
We're talking about a local county election with a sum total of 36 votes cast. Clearly there was an error of some sort. Which brings up two fundamental questions all election officials must ask:
1) Did this error change the outcome of a race? That is the first consideration, because if it didn't then the severity of the error is vastly reduced.
2) If this error changed the outcome of a race, was it intentional? That is, was the outcome of democracy subverted, and done so with fraudulent intent?
I think (but don't know) that the answer to those two questions will ultimately be "NO". That is, the error did not affect the outcome of his loss - though the error might have impacted the necessity for a runoff. And, further, it is highly unlikely that for a race this small anyone would have been actively engaged in voter fraud. Certainly, if these results are the result of fraud, it is almost certainly not due to party involvement.
It may be a justifiable fear that someone might perpetrate a nationwide fraud using unverifiable electronic voting machines. But this example does not support that fear. Mostly because there is a verifiable paper trail within the machines, and the race is too small for organized fraud to be worth the trouble.
Which is why the country election commissioner should take the machine(s) apart and check the vote tape. It is only 36 votes, so it shouldn't be hard to do. I am simply pointing out that this example is more likely error than fraud.
Had his vote, and the votes he assumes had been cast for him (because his friends said they did), he still wouldn't have received enough votes to win the election. Further, it's not clear he would have received even enough votes to change the *outcome* of the election (there will be a runoff due to two other candidates having won the same vote count).
Yes, the county should take the voting machines apart and verify the printed tape count. No, this incident is probably not an example of electronic voter fraud.
This is different because in all of your examples you are theorizing some sort of service which is provisioned in a manner that is tightly controlled by specific companies and highly regulated by government.
- access].
Cable service is not regulated by the government? Last I checked, the government regulated cable television prices, service, and are even debating the regulation of content. Cable television and Internet companies lay cable straight to their customers' homes across public ways just as is electricity, water, and telephone service (or metered gas, for that matter). And, just like WAP, cellular telephone service is transmitted using radiowaves - yet no one argues that it wouldn't be theft to use another person's cell phone without their consent.
So, I'll ask a different question: If it is improper to use someone else's cell phone service without their knowledge or consent, but proper to use someone else's WAP point - also without their knowledge or consent - then what is the distinction between the two. According to you that separation would appear to be whether the service is "controlled by specific companies and highly regulated by the government." I fail to see how having a corporate charter makes any difference - a private individual has the same property rights as does a corporation. And the government IS enforcing regulation through the criminal court system.
The topic at hand is a person purchasing and implementing a device aimed at making it more convenient to use the internet service he or she has agreed to pay for, be it cable/dsl/[whatever-controlled-form-of-high-speed
That's funny. I thought the topic was about a person in Singapore being convicted and jailed due to having used a neighbor's wireless internet without permission. I am arguing that the judge's decision was proper.
Finally, WRT: ease of use - or the "making it more convenient to use the internet service" arguement. might I point out that WAP routers are intended to make Internet access easier for the *owners* of the service (as you say). But it does not exist to make it easier for *strangers* to use a service they have neither paid to use nor have permission to use by those who did pay.
How again, is this not theft?
If your neighbor has an apple tree that has branches that are growing over your fence into your backyard, are you allowed to pick the apples? Or is that theft as well?
That's a good question. Courts would probably break it down thusly, though I'm just guessing:
a) Apples on or near adjoining property could not be picked from the tree by person A, if it is deemed owned by person B.
b) Person A's apples which have fallen by no over action onto person B's property, have likely _transferred ownership_ to person B. But Apples still attached to Person A's apple tree are still owned by him.
This would be my guess. But suppose we assume that apple's hanging over into Person B's property are somehow now owned by Person B. Using this analogy, does that mean that it is proper to tap a buried electricity cable which happens to cross Person B's property (or adjoins the property) to Person A's main junction? Siphoning off someone else's metered service is still theft, is it not?
[...]which is the notion that it is deemed wrong to use an access point in its apparently intended manner.
By this logic it is "intended" for a non-owner to use someone else's electricity. Or water. Or POTS service. Or even cell phone service.
wireless internet is simply a physical carrier. Issues of open access and whether security restrictions against unauthorized use were imposed (a lock installed on a front door) do not negate the principle of a private resource. That is, just because I left my door unlocked does not give you right to enter my home and take something of mine as if it were yours.
This is no different. The technology may have changed, but the principle of private ownership remains the same.
I'm trying to imagine an electricity hookup capable of transmitting its availability and automagically hooking up to provide electricity to a nearby house/car/man-in-the-street. That's the difference you asked for.
Tesla did it seventy or so years ago. But the point behind the analogy is that all of them are metered services sold to a specific individual. That the physical transmission medium for wireless Internet is different from electricity, eater, or telephone service does not matter. Do you argue the that you should have free reign to use a neighbor's cell phone service simply because the transmission medium is wireless?