Lets say in the last election, there were 0.5% questionable votes, or unusable/invalid votes. What percentage of the votes cast were questionable/unreadable/invalid this election? 0.4%? 1%?
That's how you can determine it.
If we were voting using black and white marbles in a box, it would be easy, there's no way to be ambiguous. Except that it's hella hard to count 120M marbles, and you'd need a huge box.:)
Accuracy in this case is probably best described as "signal to noise ratio", or the ratio of properly cast to improperly cast votes (with regards to the actual content of that vote).
No, I'm not "blaming" him. I think it's excellently performed manipulation of the voters to get votes. And I figure both sides are perfectly willing to play such games.
I didn't vote for either of them, and voted 3rd party.
I'm also not exactly sold, myself, on the idea that global warming is happening because of greenhouse gases. I look at the sheer amount of energy we use, and how thermodynamics says it must become waste heat, eventually, somewhere, and see that we're going to heat this place up by continuing to do work (in the physics sense of work).
I was following the exit polls throughout the day, and they weren't matching the voting results at all, until you applied the 3% or so error of margin that exit polls are supposed to have.
At which point it became obvious that in this close of a race, exit polls are meaningless, as they can be well within their error of margin, and still not accurately reflect the outcome, since the outcome is (essentially) a binary choice instead of an analog value.
I think these election results were well down into the statistical noise, and the results are not completely clear. It just proves that we need to further refine the voting system to make it more accurate, and to get out an even higher voter turnout.
The real question is if the anomolies are any more or any less than with paper ballots.
Are the new methods statistically any more or less accurate than the last election. That I haven't seen.
How many invalidated optical scan votes vs. hanging chad votes?
Although how you screw up an optical scan vote is beyond my comprehension. The ones we used in Santa Cruz county in Cali were as simple as could be, and you got a copy of the ballot in a voting guide nearly a month in advance, with which to familiarize yourself.
No, the hole is there, it's just that people choose not to see it. They get hurricanes, that in and of itself is par for the course in Florida. However, the frequency has been going up, possibly climatic change, and Bush isn't doing anything to stop that. What he IS doing is giving them federal aid.
Playing parent again. Which is the one thing he's done consistently well with his first term in office.
He's there to make it so the people don't need to worry or think, because he's strong and he'll take care of them. Or so they think, and he'd like them to think...
I think it's a great way to get votes from dumb voters who don't know how to take care of themselves.
In college I did a bit of 68000 assembly. Intel x86 code just bends my head in comparison. 8 general purpose registers, and 8 address registers, and the ability to math between any of them (using the address registers for ptrs), including array operations with the extended addressing modes, made the 68000 processor a dream to write for.
Now when debugging C++ on x86, and I drop to assembly to figure out what the hell the compiler has done, I look at the code and it's the ugliest stuff I've seen. I think the 68HC11 had a better assembly language (and was about as powerful as an 8086, IIRC)...
Was this the 125K model? It might be an option. Not sure exactly, we haven't dealt with one "in the field".
Our trainer on the tools has ran across a scene where the car slid on an icy road, into the median, and rolled on it's roof. Windshield wasn't a good exit path, but they couldn't shatter out the windows, so they pulled the doors instead.
The driver was not pleased with BMW for that feature (the car was essentially unhurt before they started cutting it apart).
They are. They compress, like a spring. Which means that once we start cutting with the jaws, they can spring back to their normal shape. I haven't heard of them trying to exit the vehicle until the cutting started, though.
But they tend to straighten out in a hurry.
Unfired airbags are evil. They need to be treated like time bombs with an unknown length fuse.
Also, if an occupant hits a windshield, they either weren't wearing a seatbelt, or the car has a serious problem with it's restraint system.
Modern cars (ie, airbags), make the occupant face-planting the windshield a thing of the past.
And the windshield shouldn't be transmitting energy, that's the job of the a-pillars, and then only in a rollover. In a standard front-end collission, ideally nothing rear of the firewall gets damaged.
And having seen a Mini Cooper go through a telephone pole and into a tree at about 45 mph, with no damage rear of the firewall, it can be done.
Actually, what you're referring to is the side and rear windshield tempered safety-glass, which breaks into small squarish pieces with any sharp impact.
The windshield is actually layers of glass and a clear plastic, which holds the glass together as it shatters.
Safety glass breaks into cubes, but normal glass spiderwebs.
As a volunteer firefighter, and not only being trained on how to remove automotive glass expediently, but also having seen "forehead dents" in windshields (luckily no full-ejection of occupants out the windshield), I can attest to how it actually breaks.
BMW is putting Polycarbonate windows in the 740 series cars. You can't break those with a sledgehammer (I know someone who tried, our instructor on jaws of life tools).
My experience with Rain-X on three different vehicles is that you don't need wipers at anything over 45mph, and below that, it depends on how hard it's raining, and how recently you applied it.
A fresh coating, in light rain should collect into drops you can see around, and in heavy rain, it will streak down so fast it won't be a problem.
Misting is the only problem I've had, where it tends to form "microdrops" that aren't large enough and close enough together to clump into a larger drop and fall, but are difficult to see around, but it's not much worse than a problematic defogger.
In addition to problems like IE not clearing it's cache correctly (which I just confirmed with the latest version of IE), the local google cache of visited pages does not get emptied when the cache DOES get emptied.
- Emptied IE cache (control panel->Internet options->Delete Files) - Set IE to minimal cache (1MB), to destroy pages on close, and to not cache encrypted pages. - Visited a single page (my default page, a corporate intranet page) - Closed IE (confirmed in task manager) - Google Desktop searched for company name, and first item in results is page I viewed. - Went back to control panel, emptied IE cache, and researched in GDS, page still there in Google's cache (and the cached version viewable)
So, yes, this is distinctly a security issue. Not exactly Google's problem, because it does exactly what it says it does, like all good high-power tools. However, it should come with a large caveat for any multiple-user pcs (ie, don't use it).
Clearing the IE web history also doesn't get rid of it.
But then, it's also beta. Which means that there are still issues, features, and problems to work out. It's great for my purposes, so I use it.
But problems in BETA software shouldn't rank FUD from CNN.
Re:TCP is old...so what?
on
Replacing TCP?
·
· Score: 2, Insightful
But not from a computation standpoint. Yes, you lose some bandwidth due to FEC, but if you don't drop the packets, you also don't need to compute the missing data from the FEC data.
AFAIK, the point of FEC is that you need to send less data than you actually lose, but that's getting into lots of wierd math theory that I'm still coming up to speed on with FEC (looking at it for an application at work).
The other advantage of FEC over NACK that I can see is that on a high-latency connection, or a server dealing with lots of clients, the server doesn't need to get involved. The client can compute the missing piece, maybe in less time than the resend would take.
Re:Is it an open protocol?
on
Replacing TCP?
·
· Score: 1
Interesting, I wonder how this applies to devices which don't load a copy into memory (execute directly out of flash/nvram/rom/eeprom, etc).
Emergency services was ALL coordinated by ARES for that incident. And they had base stations running on-site that afternoon.
Cell-phones were useless on-scene, the system was overloaded.
FDNY's own radio system couldn't handle the traffic of the event itself while it was happening, just before the buildings came down. They didn't have enough frequencies allocated to deal with the problem.
But with BPL, all the houses using BPL in Jersey would have been interfering as well. The frequencies used, when radiated from the wires, bounce all over the place, and travel great distances, which means that even 100s of miles away, you get noise that you didn't use to have.
If the copyright owner loses a sale due to a violation of copyright law, then it is a real loss.
Ie, someone copied it instead of buying it. Potentially, i could buy any CD produced. It's just not likely, I have better taste than that.
But if a CD wasn't going to be bought new, at retail, then the copying of it doesn't constitute an actual loss. I've receive a ton of MP3s and copied CDs from friends which were of the "listen to this!" offer. Most have ended up in the trashcan (real or virtual). Those few I actually liked, I try to buy, although I usually end up buying used anyway, because I don't have the money for retail prices right now.
p2p allows someone to go browsing and finding interesting things, and decide if you want to own them or not. If you DON'T buy the CD, then yes, it's a loss of a sale, but if it was a CD you're weren't going to buy, as it costs too much, then it's not really a lost sale.
The problem is that they see 1mil downloads of britney spears latest hit, and 1 mil downloads of the hit before that, and if off the same CD, claim that 2mil CDs worth of goods weren't sold, and were stolen, and that that's $36mil in revenue lost.
When in reality, they might only have lost a couple hundred thousand sales, say (250K), for a real loss of $4.5mil.
Which is yes a loss. On the flip side, they probably also gained some sales by people finding new artists, and buying thier CDs. The problem is that no one has any statistics for how many mp3s are kept instead of buying cds, which percentage of those would have resulted in sales had the person not already had the mp3s, how many cd sales resulted in mp3s/cds copied from friends, etc.
Actually, the stuff you're listing is exactly the kind of thing that people NEED to know about. If you're downwind of a plant during prevailing winds, that affects housing prices, and also affects emergency response planning.
The local FD sure as hell wants to know as much as possible about HazMat sources and how the wind blows/water flows in an area.
Things get especially nasty in valleys, where the wind blows up it during the day, and down it at night, changing mid-afternoon/early-morning or so, because then you need to plan out both scenarios AND how to deal with the change in wind.
I'm a volunteer FF in a small town (10K people) in a valley, we're an all-volunteer dept, and the county GIS data is crucial for us for planning purposes, but we also back it up with USGS, our own surveys of where hazmat is stored (both legally and illegally) and our own data on locations of hydrants, weight-limits on bridges, and widths of roads for getting fire engines up into the VERY narrow, tight turns around redwood trees.
This kind aerial photography, coupled with 10' accuracy USGS data (too bad it's only 40', which is laughable in our terrain, as it's much more convoluted), and with our own county GIS (property lines, zoning, roads) data, all in-cab and we'd have a HELL of a better time dealing with stuff.
Even better if we could get live sat imagery (even within last 30 minutes would be good) for location of fires, especially at night.
Trying to locate a wildland fire you can see, but it's off on some other ridge/valley is a major PITA, especially at night.
Why is it dismissed outright on slashdot (comprised of folks who apparently trust people more than they trust machines).
I have no idea, this seems like the kind of place where people would be considered much less secure than machines, especially how when computer security comes up, it's often pointed out the failings are social-engineering attacks instead of "hacking".
1-2: There's a known number of possible votes per polling center, and a known number of booths, therefore there should be more than enough paper available in advance for one day's voting. Same with ink.
Scantrons:
This is where a machine helps, black ink (especially magnetic ink, like financial institutions use(d?)) is much less error prone than a #2 pencil with a student erasing, especially since they aren't going to be erasing the mark. That's what the shredder is for, and a new ballot.
Also, a scantron is probably a bad example, as they read a series of dots, and I've seen them get off before on a read. The 2D barcode (a la postal service and UPS) are very accurate reads. UPS/FedEx/USPS send a LOT of mail daily relying on this sort of thing for tracking. And when things do go amis, you can know (embed CRC data into it), and then cause it to be flagged as "human countable", and with black ink, it shouldn't be hard to determine the right votes.
Then a manual entry station for the vote, using information off both the ballot AND the counter's own id, which needs to be validated, helping to deter ballot-stuffing of "unreadable ballots".
You have two sets of paper, one is a set of "votes", and another is a set of "tokens".
The votes are coded, as are the tokens. 1=1. They are kept separately, and counted separately, by separate people, preferrably separate locations.
Both contain a hashed version of the voter's "session id".
When the votes are counted, the votes are counted by machine (for speed), and can be counted by hand "for accuracy", but the machines should be near-perfect.
Then, the counting software can validate the sets of votes and the sets of tokens, electronically, and separately from the location of the counting.
Now, you provide an electronic file of votes, and an electronic file of tokens, as scanned by the machines. Each signed digitally (on the whole file), and verifiable.
Discrepancies are a cause of worry, but say below a certain threshold (a very, very low threshold) are statistical noise, and not to be worried about, at least no more than the hanging chad was.
Major discrepancies are a worry, and state that perhaps things were tampered with (or equipment was faulty).
The files are digitally signed, if modified, they'll fail their signature checks (if THAT's handled correctly), and then that set of files can be thrown out, and the votes ordered to be rescanned at the two counting locations by the central tabulating office.
This is just off the top of my head, I'm sure there are flaws, but things can be layered, made redundant, and that's when security really starts to show up.
I think some people have been, but the problem is that they keep getting ignored by press/those in power.
We've so screwed ourselves over here. The govm't is so much larger than the elected officials, to the point that I'm not even certain if we replaced every incumbant in an election, how much of a percentage of the policy-making people we'd have actually replaced. DoD/EPA/DoE/TSA/NHSTA/etc. is a WHOLE lot of people, and they tend to make their own rules for the most part, and they're all appointed.
1-2: Handled by millions of point-of-sale terminals already. This is no large feat of engineering that needs to be reinvented.
3: Scantrons are ancient, and work well, with a very low error rate, at least, lower than hanging chads when you've got machines to properly mark the cards in the first place.
4: He walks out of the booth with it, and right up to the ballot box, just like we do currently. No big deal, and after that, he can have proof he voted, but the card with the actual votes on it is in the box.
=====
I wouldn't be amiss to a mis-vote called whenever the election was indeterminate with a known (low) level of error. Like, 0.01% or less (or some other number, that one was pulled out of thin air). To cover error in the system.
But with proper security you can have an audit trail on the system that's rather non-trivial to hack. This is a system with no redundancy, with no way of knowing if it's been tampered with after-the-fact.
"Did windows just eat the votes, or was it malicious?"
Just what I want to deal with. There are MANY security schemes that could make this bullet-proof, but it's obvious that Diebold should have stuck to ATMs. (Actually, makes me wonder what software THEY run inside... But then, the finance industry is apparently a LOT more uptight than voting districts/boards are).
Slashdot Mirror
Margin of error.
:)
Lets say in the last election, there were 0.5% questionable votes, or unusable/invalid votes. What percentage of the votes cast were questionable/unreadable/invalid this election? 0.4%? 1%?
That's how you can determine it.
If we were voting using black and white marbles in a box, it would be easy, there's no way to be ambiguous. Except that it's hella hard to count 120M marbles, and you'd need a huge box.
Accuracy in this case is probably best described as "signal to noise ratio", or the ratio of properly cast to improperly cast votes (with regards to the actual content of that vote).
No, I'm not "blaming" him. I think it's excellently performed manipulation of the voters to get votes. And I figure both sides are perfectly willing to play such games.
I didn't vote for either of them, and voted 3rd party.
I'm also not exactly sold, myself, on the idea that global warming is happening because of greenhouse gases. I look at the sheer amount of energy we use, and how thermodynamics says it must become waste heat, eventually, somewhere, and see that we're going to heat this place up by continuing to do work (in the physics sense of work).
C/C++ doesn't specify the size of an int/signed int in bits. That's platform dependent.
On an 8-bit micro, it's 8 bits, on a 16-bit micro, it's 16, 32 on a 32, etc.
Ints can be either shorts or longs, depending on the platform. (ints on longs on Intel, and IIRC were shorts on Motorola 68K).
but yes, uint64 is the way to go.
I was following the exit polls throughout the day, and they weren't matching the voting results at all, until you applied the 3% or so error of margin that exit polls are supposed to have.
At which point it became obvious that in this close of a race, exit polls are meaningless, as they can be well within their error of margin, and still not accurately reflect the outcome, since the outcome is (essentially) a binary choice instead of an analog value.
I think these election results were well down into the statistical noise, and the results are not completely clear. It just proves that we need to further refine the voting system to make it more accurate, and to get out an even higher voter turnout.
The real question is if the anomolies are any more or any less than with paper ballots.
Are the new methods statistically any more or less accurate than the last election. That I haven't seen.
How many invalidated optical scan votes vs. hanging chad votes?
Although how you screw up an optical scan vote is beyond my comprehension. The ones we used in Santa Cruz county in Cali were as simple as could be, and you got a copy of the ballot in a voting guide nearly a month in advance, with which to familiarize yourself.
No, the hole is there, it's just that people choose not to see it. They get hurricanes, that in and of itself is par for the course in Florida. However, the frequency has been going up, possibly climatic change, and Bush isn't doing anything to stop that. What he IS doing is giving them federal aid.
Playing parent again. Which is the one thing he's done consistently well with his first term in office.
He's there to make it so the people don't need to worry or think, because he's strong and he'll take care of them. Or so they think, and he'd like them to think...
I think it's a great way to get votes from dumb voters who don't know how to take care of themselves.
NIN (Trent Reznor), who's from Pennsylvania, has released numerous albums only available in the UK/Europe, and get imported to the States/Canada.
They seem to like more interesting music over there.
In college I did a bit of 68000 assembly. Intel x86 code just bends my head in comparison. 8 general purpose registers, and 8 address registers, and the ability to math between any of them (using the address registers for ptrs), including array operations with the extended addressing modes, made the 68000 processor a dream to write for.
Now when debugging C++ on x86, and I drop to assembly to figure out what the hell the compiler has done, I look at the code and it's the ugliest stuff I've seen. I think the 68HC11 had a better assembly language (and was about as powerful as an 8086, IIRC)...
Was this the 125K model? It might be an option. Not sure exactly, we haven't dealt with one "in the field".
Our trainer on the tools has ran across a scene where the car slid on an icy road, into the median, and rolled on it's roof. Windshield wasn't a good exit path, but they couldn't shatter out the windows, so they pulled the doors instead.
The driver was not pleased with BMW for that feature (the car was essentially unhurt before they started cutting it apart).
They are. They compress, like a spring. Which means that once we start cutting with the jaws, they can spring back to their normal shape. I haven't heard of them trying to exit the vehicle until the cutting started, though.
But they tend to straighten out in a hurry.
Unfired airbags are evil. They need to be treated like time bombs with an unknown length fuse.
Also, if an occupant hits a windshield, they either weren't wearing a seatbelt, or the car has a serious problem with it's restraint system.
Modern cars (ie, airbags), make the occupant face-planting the windshield a thing of the past.
And the windshield shouldn't be transmitting energy, that's the job of the a-pillars, and then only in a rollover. In a standard front-end collission, ideally nothing rear of the firewall gets damaged.
And having seen a Mini Cooper go through a telephone pole and into a tree at about 45 mph, with no damage rear of the firewall, it can be done.
Actually, what you're referring to is the side and rear windshield tempered safety-glass, which breaks into small squarish pieces with any sharp impact.
The windshield is actually layers of glass and a clear plastic, which holds the glass together as it shatters.
Safety glass breaks into cubes, but normal glass spiderwebs.
As a volunteer firefighter, and not only being trained on how to remove automotive glass expediently, but also having seen "forehead dents" in windshields (luckily no full-ejection of occupants out the windshield), I can attest to how it actually breaks.
BMW is putting Polycarbonate windows in the 740 series cars. You can't break those with a sledgehammer (I know someone who tried, our instructor on jaws of life tools).
My experience with Rain-X on three different vehicles is that you don't need wipers at anything over 45mph, and below that, it depends on how hard it's raining, and how recently you applied it.
A fresh coating, in light rain should collect into drops you can see around, and in heavy rain, it will streak down so fast it won't be a problem.
Misting is the only problem I've had, where it tends to form "microdrops" that aren't large enough and close enough together to clump into a larger drop and fall, but are difficult to see around, but it's not much worse than a problematic defogger.
In addition to problems like IE not clearing it's cache correctly (which I just confirmed with the latest version of IE), the local google cache of visited pages does not get emptied when the cache DOES get emptied.
- Emptied IE cache (control panel->Internet options->Delete Files)
- Set IE to minimal cache (1MB), to destroy pages on close, and to not cache encrypted pages.
- Visited a single page (my default page, a corporate intranet page)
- Closed IE (confirmed in task manager)
- Google Desktop searched for company name, and first item in results is page I viewed.
- Went back to control panel, emptied IE cache, and researched in GDS, page still there in Google's cache (and the cached version viewable)
So, yes, this is distinctly a security issue. Not exactly Google's problem, because it does exactly what it says it does, like all good high-power tools. However, it should come with a large caveat for any multiple-user pcs (ie, don't use it).
Clearing the IE web history also doesn't get rid of it.
But then, it's also beta. Which means that there are still issues, features, and problems to work out. It's great for my purposes, so I use it.
But problems in BETA software shouldn't rank FUD from CNN.
But not from a computation standpoint. Yes, you lose some bandwidth due to FEC, but if you don't drop the packets, you also don't need to compute the missing data from the FEC data.
AFAIK, the point of FEC is that you need to send less data than you actually lose, but that's getting into lots of wierd math theory that I'm still coming up to speed on with FEC (looking at it for an application at work).
The other advantage of FEC over NACK that I can see is that on a high-latency connection, or a server dealing with lots of clients, the server doesn't need to get involved. The client can compute the missing piece, maybe in less time than the resend would take.
Interesting, I wonder how this applies to devices which don't load a copy into memory (execute directly out of flash/nvram/rom/eeprom, etc).
Emergency services was ALL coordinated by ARES for that incident. And they had base stations running on-site that afternoon.
Cell-phones were useless on-scene, the system was overloaded.
FDNY's own radio system couldn't handle the traffic of the event itself while it was happening, just before the buildings came down. They didn't have enough frequencies allocated to deal with the problem.
But with BPL, all the houses using BPL in Jersey would have been interfering as well. The frequencies used, when radiated from the wires, bounce all over the place, and travel great distances, which means that even 100s of miles away, you get noise that you didn't use to have.
I'd rephrase that to:
If the copyright owner loses a sale due to a violation of copyright law, then it is a real loss.
Ie, someone copied it instead of buying it. Potentially, i could buy any CD produced. It's just not likely, I have better taste than that.
But if a CD wasn't going to be bought new, at retail, then the copying of it doesn't constitute an actual loss. I've receive a ton of MP3s and copied CDs from friends which were of the "listen to this!" offer. Most have ended up in the trashcan (real or virtual). Those few I actually liked, I try to buy, although I usually end up buying used anyway, because I don't have the money for retail prices right now.
p2p allows someone to go browsing and finding interesting things, and decide if you want to own them or not. If you DON'T buy the CD, then yes, it's a loss of a sale, but if it was a CD you're weren't going to buy, as it costs too much, then it's not really a lost sale.
The problem is that they see 1mil downloads of britney spears latest hit, and 1 mil downloads of the hit before that, and if off the same CD, claim that 2mil CDs worth of goods weren't sold, and were stolen, and that that's $36mil in revenue lost.
When in reality, they might only have lost a couple hundred thousand sales, say (250K), for a real loss of $4.5mil.
Which is yes a loss. On the flip side, they probably also gained some sales by people finding new artists, and buying thier CDs. The problem is that no one has any statistics for how many mp3s are kept instead of buying cds, which percentage of those would have resulted in sales had the person not already had the mp3s, how many cd sales resulted in mp3s/cds copied from friends, etc.
Actually, the stuff you're listing is exactly the kind of thing that people NEED to know about. If you're downwind of a plant during prevailing winds, that affects housing prices, and also affects emergency response planning.
The local FD sure as hell wants to know as much as possible about HazMat sources and how the wind blows/water flows in an area.
Things get especially nasty in valleys, where the wind blows up it during the day, and down it at night, changing mid-afternoon/early-morning or so, because then you need to plan out both scenarios AND how to deal with the change in wind.
I'm a volunteer FF in a small town (10K people) in a valley, we're an all-volunteer dept, and the county GIS data is crucial for us for planning purposes, but we also back it up with USGS, our own surveys of where hazmat is stored (both legally and illegally) and our own data on locations of hydrants, weight-limits on bridges, and widths of roads for getting fire engines up into the VERY narrow, tight turns around redwood trees.
This kind aerial photography, coupled with 10' accuracy USGS data (too bad it's only 40', which is laughable in our terrain, as it's much more convoluted), and with our own county GIS (property lines, zoning, roads) data, all in-cab and we'd have a HELL of a better time dealing with stuff.
Even better if we could get live sat imagery (even within last 30 minutes would be good) for location of fires, especially at night.
Trying to locate a wildland fire you can see, but it's off on some other ridge/valley is a major PITA, especially at night.
Why is it dismissed outright on slashdot (comprised of folks who apparently trust people more than they trust machines).
I have no idea, this seems like the kind of place where people would be considered much less secure than machines, especially how when computer security comes up, it's often pointed out the failings are social-engineering attacks instead of "hacking".
1-2: There's a known number of possible votes per polling center, and a known number of booths, therefore there should be more than enough paper available in advance for one day's voting. Same with ink.
Scantrons:
This is where a machine helps, black ink (especially magnetic ink, like financial institutions use(d?)) is much less error prone than a #2 pencil with a student erasing, especially since they aren't going to be erasing the mark. That's what the shredder is for, and a new ballot.
Also, a scantron is probably a bad example, as they read a series of dots, and I've seen them get off before on a read. The 2D barcode (a la postal service and UPS) are very accurate reads. UPS/FedEx/USPS send a LOT of mail daily relying on this sort of thing for tracking. And when things do go amis, you can know (embed CRC data into it), and then cause it to be flagged as "human countable", and with black ink, it shouldn't be hard to determine the right votes.
Then a manual entry station for the vote, using information off both the ballot AND the counter's own id, which needs to be validated, helping to deter ballot-stuffing of "unreadable ballots".
You have two sets of paper, one is a set of "votes", and another is a set of "tokens".
The votes are coded, as are the tokens. 1=1. They are kept separately, and counted separately, by separate people, preferrably separate locations.
Both contain a hashed version of the voter's "session id".
When the votes are counted, the votes are counted by machine (for speed), and can be counted by hand "for accuracy", but the machines should be near-perfect.
Then, the counting software can validate the sets of votes and the sets of tokens, electronically, and separately from the location of the counting.
Now, you provide an electronic file of votes, and an electronic file of tokens, as scanned by the machines. Each signed digitally (on the whole file), and verifiable.
Discrepancies are a cause of worry, but say below a certain threshold (a very, very low threshold) are statistical noise, and not to be worried about, at least no more than the hanging chad was.
Major discrepancies are a worry, and state that perhaps things were tampered with (or equipment was faulty).
The files are digitally signed, if modified, they'll fail their signature checks (if THAT's handled correctly), and then that set of files can be thrown out, and the votes ordered to be rescanned at the two counting locations by the central tabulating office.
This is just off the top of my head, I'm sure there are flaws, but things can be layered, made redundant, and that's when security really starts to show up.
I think some people have been, but the problem is that they keep getting ignored by press/those in power.
We've so screwed ourselves over here. The govm't is so much larger than the elected officials, to the point that I'm not even certain if we replaced every incumbant in an election, how much of a percentage of the policy-making people we'd have actually replaced. DoD/EPA/DoE/TSA/NHSTA/etc. is a WHOLE lot of people, and they tend to make their own rules for the most part, and they're all appointed.
1-2: Handled by millions of point-of-sale terminals already. This is no large feat of engineering that needs to be reinvented.
3: Scantrons are ancient, and work well, with a very low error rate, at least, lower than hanging chads when you've got machines to properly mark the cards in the first place.
4: He walks out of the booth with it, and right up to the ballot box, just like we do currently. No big deal, and after that, he can have proof he voted, but the card with the actual votes on it is in the box.
=====
I wouldn't be amiss to a mis-vote called whenever the election was indeterminate with a known (low) level of error. Like, 0.01% or less (or some other number, that one was pulled out of thin air). To cover error in the system.
Automatic revote.
But with proper security you can have an audit trail on the system that's rather non-trivial to hack. This is a system with no redundancy, with no way of knowing if it's been tampered with after-the-fact.
"Did windows just eat the votes, or was it malicious?"
Just what I want to deal with. There are MANY security schemes that could make this bullet-proof, but it's obvious that Diebold should have stuck to ATMs. (Actually, makes me wonder what software THEY run inside... But then, the finance industry is apparently a LOT more uptight than voting districts/boards are).