Ultimately, perhaps even within the next few centuries,
We may not have a few centuries. If we continue our current 2.3% energy growth per year, then in a bit above 400 years the earth's surface temperature will reach the boiling point. Something has to give long before that, and that may put a wrench in our space dreams.
we're going to see a situation where the abundant resources in our solar system are harvested and processed by mostly automated engines,
It's far from clear that the EROEI (Energy Return On Energy Invested) of space resources is ever going to be greater than 1. Even for ores where the EROEI is less relevant, the energy cost still has to be compared to that of earth-based mining solutions and it's unlikely to be competitive. It looks like we will at most be able to capture a select few light asteroids (those coming close to the earth with just the right speed), but that won't provide abundant resources to everyone on earth. Note that even having a StarTram on one side won't help for the return trip.
Having a paper print out which can be checked by the voter works well enough.
So your solution is to do a regular election with hand counted paper ballots but add some computers that produce results that are thrown away because they cannot be trusted. And you say that's better than the same election without the useless computers?
And then nobody ever counts the paper ballots since the result is known anyway, and so rigging the electronic part of the election is all you need to get your candidate elected.
I agree. The technical problems are the lesser difficulty. They are all solvable problems.
No. The combination of anonymity, transparency, verifiability and robustness against insiders is unique to voting and there is no known DRE solution to this.
So why did Diebold refuse to release source code?
I think they just didn't want their competitors to steal their code, just like any other proprietary software company. They probably also (at least unconsciously) did not want the public to see how shoddy their code is and get called out for doing such a bad job.
I would suggest open-sourcing the software and the hardware design.
That would not help. What is needed is for any voter to be able to check, on election day, that the software really being used is the one that was specified. But there's obviously no way of doing that without also giving that voter all the access he needs to completely rig the election (or just DOS it as verifying everything would take days).
Nothing that an old-fashioned optical scan ballot couldn't handle.
In other words, using the machine was a solution looking for a problem (and causing numerous problems of its own).
While I agree with you on the whole, optical scan machines are electronic voting machines. They can malfunction and change 30% of the votes, so they could be rigged to steal an election too. And yes, two years later people may notice, but by then it's too late.
Um... USA and Canada, and I'd be surprised if Europe wasn't exactly the same way. What did you think that ATM/debit card was, a high-tech NFC system or something?
Well in France all our credit cards have had chips for the past twenty years. Of course the problem is that ATMs still have to accept credit cards that only have a magnetic stripe otherwise tourists would be unable to use them. So that reduces the security again.
You don't give the voters access to the whole machine, you only give them access to the touchscreen monitor, and maybe some kind of keypad. [...] When we talk about having "physical access" to a machine, that means the WHOLE machine, as in a desktop PC where you can put your hands on the tower case, plug in USB devices, open the side panel, unplug the hard drive and put it in another system to access it, etc.
Voting machines typically spend more than 99% of their time in storage (something close to 364 days out of 365). If you get access to the storage room for just one night you'll get full access to the machines. At 1 minute per machine to change a ROM that's enough to rig an election.
There obviously is and the Indian machines were designed with that in mind - for example, even if you steal a single machine and hack it to give 100% of it's votes to whoever is paying you that has a limited impact on the total vote. Stealing a lot of machines is likely to be noticed.
If you plan to rig the machines used in the election to favor your candidate, then obviously stealing them is a bad idea since the organising authority won't have them anymore. However they spend more than 99% of their time in storage (something like 364 days out of 365). So all you need is get access to them while they are in storage to rig them and then it much less likely anyone will notice. Since the machines have not been stolen, they will be used in the next election and get your candidate elected.
Oh, for a local election you will have access to all the machines at once so you'll be able to get your candidate elected with 51.5% of the votes so it does not raise any suspicion. For a national election you'll have to do it in a few more storage depots, or bribe a software developper working on the next 'security update'.
I pay my bills online.
I do my banking online.
I order my shopping online...
When you pay online, your the recipient knows who you are, how much you pay and why. The bank or an intermediary like PayPal also knows who you are and who you paid that money to, and how much you have left in your account. The stores you shop at know which items you looked at, for how long and obviously which ones you ordered, and provide you with proof that you ordered stuff.
A voting machine must provably not record who you voted for, must not provide you with a way to prove how you voted, must still be provably secure, and must be understandable and verifiable by anyone.
So no, shopping / paying bills online have nothing in common with voting online. The former is easy, but for the latter we simply don't know how to do it with DRE systems.
I read your second message, but in case you forgot, one can't go back on Slashdot and alter an earlier post to reflect new information.
Ok when I said "SD is actually either a 3.5Mb/s MPEG2 stream or around 1.8Mb/s in MPEG4" I forgot to then mention that HD is always broadcast as MPEG4. I though that was obvious and I still blame you for not wondering why I had brought it up, not doing any research; and for ignoring this new data in your next reply which was "All I can say is that is that 9 MB/s isn't that much of an improvement over 5Mb/s;)" (so no need to bring up the impossibility of time travel here).
This has gone on long enough anyway so you can now go on and on about my many faults without fear of further corrections on my part. I won't be replying to your messages anymore.
I was simply going with the numbers that you gave me. You said "5Mb/s is enough for a standard definition channel and 9Mb/s is enough for full HD". Can't criticize me for using your numbers!:)
I guess my first post should not have assumed that everyone knows SD is usually broadcast in MPEG2 while HD always uses MPEG4. However I criticize you for not reading my second message where I clarified things:
SD is actually either a 3.5Mb/s MPEG2 stream or around 1.8Mb/s in MPEG4
I added some margin to be comfortable. SD is actually either a 3.5Mb/s MPEG2 stream or around 1.8Mb/s in MPEG4 (that latter stream is mostly to extend the availability of TV to long ADSL lines so not geared towards top quality). The HD streams start at 6Mb/s which is already three times more and, as far as I know, they never go above 9.5Mb/s which is for over-the-air HD. So yeah, it's not the 30Mb/s you get from a Blu-Ray but by that standard there's no HD broadcast TV (DVB-T tops out at 30Mb/s but no broadcaster is going to allocate that much to a single TV channel).
Can you confirm that you're actually getting significantly more from your provider?
Here in Iceland I get my TV service through something called "myndlykill"; I don't know the English word for it, but it's a box with a Cat5 on the back that plugs into your hub and downloads channels from the net and yeilds an HDMI signal. Most people here have 50Mb/s or 100Mb/s optical fiber net connections so there's enough bandwidth for a good picture.
If you're watching a single channel at a time there's no need for 50+Mb/s or fiber. 5Mb/s is enough for a standard definition channel and 9Mb/s is enough for full HD. So this can go through a regular ADSL2+ connection with no problem as long as the length of your phone line is less than 2 km (~1.25 miles). That how all French ISPs offer Triple-Play: Internet+Phone+TV.
I'll grant you that things are different if you want to record one channel while watching another, or if you want to watch one channel while your kids watch another. It's also likely to change in the coming years with 3D, 4K, high refresh rates, etc.
From that description it sounds like Sony has reinvented screens with matte finish. Surely there is more to this.
Yes. The answer comes from Wikipedia's description of the nanoscale structure found in Moth Eyes:
Moths' eyes have an unusual property: their surfaces are covered with a natural nanostructured film which eliminates reflections. [...] The structure consists of a hexagonal pattern of bumps, each roughly 200 nm high and spaced on 300 nm centers. This kind of antireflective coating works because the bumps are smaller than the wavelength of visible light, so the light sees the surface as having a continuous refractive index gradient between the air and the medium, which decreases reflection by effectively removing the air-lens interface.
How efficient is the production of natural gas and pumping it to houses?
That's irrelevant since you incur the same costs whether you directly burn the gas for heating or use it in the fuel cell and then use electric heating. Now in the absolute conventional gas has an EROI somewhere between 20 and 100. I doubt distribution uses much energy but I could not find any figure so if you find any please post them.
You're forgetting that power plants have efficiencies of 30% or less. So add a heat pump with a typical COP of 3 and the overall cycle is no better than burning the gas directly. Now if the 50% efficiency figure quoted for this fuel cell is really just for the electricity generation side (i.e. does not take into account heat generation), then that may be more interesting. It would actually be a step up from standard power plants too so if it can be scaled up it should be.
Why not move the ballot boxes to a secure location like, I dunno, where they count them or where the ballots are stored in the case of a recount.
How do you verify the ballot boxes are not tampered with while they are being moved? And, by the way, the only reasonable place for counting the ballots is at the polling place right when the election closes. anything else means the voters lose control of the election.
You can even count them while they're there that night.
You could even publish partial results. No wait, that's probably not allowed!
I have a feeling that securing ballot boxes is a solved problem. (Or at least solved enough that this doesn't introduce undue risk.)
It's clear we're just not ready for electronic voting. Let's stick to paper ballots and re-visit this idea in twenty years or so.
There's no point revisiting the issue in 20 years. A voter won't be any more able to verify that the computer counting the votes runs clean software on clean hardware.
No one is going to count every ballot by hand. Why? Because hand-counting is far more inaccurate than machine counting.
I have participated in the counting of many elections in France. I'm not an election official or anything. In France the ballots are counted by volunteers under the watch of the ballot place officials, of party representatives and of the general public. Here is how it works.
Volunteers arrive a bit before the election closes. As soon as the election closes the officials open the transparent ballot box, count the envelopes and make sure that count matches the counter on the ballot box and in the registration book. Then they make groups of a hundred envelopes. Volunteers sit at (typically 3) tables in groups of four and receive the groups of a hundred envelopes one at a time. One volunteer opens the envelope, unfolds the ballot and reads it. He hands it to the volunteer sitting diagonally accross the table who reads the ballot aloud. The other two volunteers (also sitting diagonally from each other) have a sheet of paper and make a tick for the right candidate for every ballot. All the while officials, party representatives and general public hover around making sure there are no shenanigans. When the counting is done the election results for that polling place are then proclaimed right there on the spot.
I've never seen or heard of a discrepancy of more than one envelope in a thousand. In the few polling places that use e-voting machines however there's been discrepancies of up to 13 ballots in a polling place! So no, e-voting machines don't seem any more reliable.
Have you never been to a magician's stage show? He gets 500 people to all look at the wrong thing at the same time with close to 100% accuracy. And you are claiming that a well timed car backfire won't make people look. Really? Really?
Yes, really. A magic trick is one thing. Substituting hundreds of ballots / enveloppes without adding or removing even one is quite another. And even if your genius magician was able to tamper with the votes at a polling place, he'd only be able to impact a tiny fraction of the votes whereas a single programmer working on the e-voting software can tamper with every single vote.
Not only is there massive interest in openness and transparency in the voting process, but there also a need for extremely thorough vetting of the software, its design, and its update lineage.
Open-source is totally useless for e-voting. The reason is that on election day you cannot verify that the code that runs is the one that you reviewed.
Slashdot Mirror
Such optimism!
Ultimately, perhaps even within the next few centuries,
We may not have a few centuries. If we continue our current 2.3% energy growth per year, then in a bit above 400 years the earth's surface temperature will reach the boiling point. Something has to give long before that, and that may put a wrench in our space dreams.
we're going to see a situation where the abundant resources in our solar system are harvested and processed by mostly automated engines,
It's far from clear that the EROEI (Energy Return On Energy Invested) of space resources is ever going to be greater than 1. Even for ores where the EROEI is less relevant, the energy cost still has to be compared to that of earth-based mining solutions and it's unlikely to be competitive. It looks like we will at most be able to capture a select few light asteroids (those coming close to the earth with just the right speed), but that won't provide abundant resources to everyone on earth. Note that even having a StarTram on one side won't help for the return trip.
One should be able to assume that the hash would be also published in say a newspaper
Irrelevant.
(also in the event of a problem somebody could open the machine and then run the hashes as a check)
Can only happen after the election, in other words, too late.
besides i said OUTLINE details would need to be sorted out.
The devil is in the details. There are researchers working on this and they have not yet found even so much as a theoretical solution.
the voting screen should have a hash of both images on a "rail" at the bottom so that both can be verified at random
That's like asking a suspect whether he is guilty and believing him when he says no.
Having a paper print out which can be checked by the voter works well enough.
So your solution is to do a regular election with hand counted paper ballots but add some computers that produce results that are thrown away because they cannot be trusted. And you say that's better than the same election without the useless computers?
And then nobody ever counts the paper ballots since the result is known anyway, and so rigging the electronic part of the election is all you need to get your candidate elected.
I agree. The technical problems are the lesser difficulty. They are all solvable problems.
No. The combination of anonymity, transparency, verifiability and robustness against insiders is unique to voting and there is no known DRE solution to this.
So why did Diebold refuse to release source code?
I think they just didn't want their competitors to steal their code, just like any other proprietary software company. They probably also (at least unconsciously) did not want the public to see how shoddy their code is and get called out for doing such a bad job.
I would suggest open-sourcing the software and the hardware design.
That would not help. What is needed is for any voter to be able to check, on election day, that the software really being used is the one that was specified. But there's obviously no way of doing that without also giving that voter all the access he needs to completely rig the election (or just DOS it as verifying everything would take days).
Nothing that an old-fashioned optical scan ballot couldn't handle.
In other words, using the machine was a solution looking for a problem (and causing numerous problems of its own).
While I agree with you on the whole, optical scan machines are electronic voting machines. They can malfunction and change 30% of the votes, so they could be rigged to steal an election too. And yes, two years later people may notice, but by then it's too late.
Um... USA and Canada, and I'd be surprised if Europe wasn't exactly the same way. What did you think that ATM/debit card was, a high-tech NFC system or something?
Well in France all our credit cards have had chips for the past twenty years. Of course the problem is that ATMs still have to accept credit cards that only have a magnetic stripe otherwise tourists would be unable to use them. So that reduces the security again.
You don't give the voters access to the whole machine, you only give them access to the touchscreen monitor, and maybe some kind of keypad. [...] When we talk about having "physical access" to a machine, that means the WHOLE machine, as in a desktop PC where you can put your hands on the tower case, plug in USB devices, open the side panel, unplug the hard drive and put it in another system to access it, etc.
Voting machines typically spend more than 99% of their time in storage (something close to 364 days out of 365). If you get access to the storage room for just one night you'll get full access to the machines. At 1 minute per machine to change a ROM that's enough to rig an election.
There obviously is and the Indian machines were designed with that in mind - for example, even if you steal a single machine and hack it to give 100% of it's votes to whoever is paying you that has a limited impact on the total vote. Stealing a lot of machines is likely to be noticed.
If you plan to rig the machines used in the election to favor your candidate, then obviously stealing them is a bad idea since the organising authority won't have them anymore. However they spend more than 99% of their time in storage (something like 364 days out of 365). So all you need is get access to them while they are in storage to rig them and then it much less likely anyone will notice. Since the machines have not been stolen, they will be used in the next election and get your candidate elected.
Oh, for a local election you will have access to all the machines at once so you'll be able to get your candidate elected with 51.5% of the votes so it does not raise any suspicion. For a national election you'll have to do it in a few more storage depots, or bribe a software developper working on the next 'security update'.
I pay my bills online. I do my banking online. I order my shopping online...
When you pay online, your the recipient knows who you are, how much you pay and why. The bank or an intermediary like PayPal also knows who you are and who you paid that money to, and how much you have left in your account. The stores you shop at know which items you looked at, for how long and obviously which ones you ordered, and provide you with proof that you ordered stuff.
A voting machine must provably not record who you voted for, must not provide you with a way to prove how you voted, must still be provably secure, and must be understandable and verifiable by anyone.
So no, shopping / paying bills online have nothing in common with voting online. The former is easy, but for the latter we simply don't know how to do it with DRE systems.
I read your second message, but in case you forgot, one can't go back on Slashdot and alter an earlier post to reflect new information.
Ok when I said "SD is actually either a 3.5Mb/s MPEG2 stream or around 1.8Mb/s in MPEG4" I forgot to then mention that HD is always broadcast as MPEG4. I though that was obvious and I still blame you for not wondering why I had brought it up, not doing any research; and for ignoring this new data in your next reply which was "All I can say is that is that 9 MB/s isn't that much of an improvement over 5Mb/s ;)" (so no need to bring up the impossibility of time travel here).
This has gone on long enough anyway so you can now go on and on about my many faults without fear of further corrections on my part. I won't be replying to your messages anymore.
I was simply going with the numbers that you gave me. You said "5Mb/s is enough for a standard definition channel and 9Mb/s is enough for full HD". Can't criticize me for using your numbers! :)
I guess my first post should not have assumed that everyone knows SD is usually broadcast in MPEG2 while HD always uses MPEG4. However I criticize you for not reading my second message where I clarified things:
SD is actually either a 3.5Mb/s MPEG2 stream or around 1.8Mb/s in MPEG4
Oh, I have no clue how much bandwidth the device uses. All I can say is that is that 9 MB/s isn't that much of an improvement over 5Mb/s ;)
You cannot compare an MPEG4 bitrate with an MPEG2 one. The right comparison is 1.8Mb/s to 9Mb/s which is over a 4x increase.
I added some margin to be comfortable. SD is actually either a 3.5Mb/s MPEG2 stream or around 1.8Mb/s in MPEG4 (that latter stream is mostly to extend the availability of TV to long ADSL lines so not geared towards top quality). The HD streams start at 6Mb/s which is already three times more and, as far as I know, they never go above 9.5Mb/s which is for over-the-air HD. So yeah, it's not the 30Mb/s you get from a Blu-Ray but by that standard there's no HD broadcast TV (DVB-T tops out at 30Mb/s but no broadcaster is going to allocate that much to a single TV channel).
Can you confirm that you're actually getting significantly more from your provider?
Here in Iceland I get my TV service through something called "myndlykill"; I don't know the English word for it, but it's a box with a Cat5 on the back that plugs into your hub and downloads channels from the net and yeilds an HDMI signal. Most people here have 50Mb/s or 100Mb/s optical fiber net connections so there's enough bandwidth for a good picture.
If you're watching a single channel at a time there's no need for 50+Mb/s or fiber. 5Mb/s is enough for a standard definition channel and 9Mb/s is enough for full HD. So this can go through a regular ADSL2+ connection with no problem as long as the length of your phone line is less than 2 km (~1.25 miles). That how all French ISPs offer Triple-Play: Internet+Phone+TV.
I'll grant you that things are different if you want to record one channel while watching another, or if you want to watch one channel while your kids watch another. It's also likely to change in the coming years with 3D, 4K, high refresh rates, etc.
From that description it sounds like Sony has reinvented screens with matte finish. Surely there is more to this.
Yes. The answer comes from Wikipedia's description of the nanoscale structure found in Moth Eyes:
Moths' eyes have an unusual property: their surfaces are covered with a natural nanostructured film which eliminates reflections. [...] The structure consists of a hexagonal pattern of bumps, each roughly 200 nm high and spaced on 300 nm centers. This kind of antireflective coating works because the bumps are smaller than the wavelength of visible light, so the light sees the surface as having a continuous refractive index gradient between the air and the medium, which decreases reflection by effectively removing the air-lens interface.
How efficient is the production of natural gas and pumping it to houses?
That's irrelevant since you incur the same costs whether you directly burn the gas for heating or use it in the fuel cell and then use electric heating. Now in the absolute conventional gas has an EROI somewhere between 20 and 100. I doubt distribution uses much energy but I could not find any figure so if you find any please post them.
Electricity can be more than 100 % efficient if you use a heat pump ( http://en.wikipedia.org/wiki/Heat_pump ).
You're forgetting that power plants have efficiencies of 30% or less. So add a heat pump with a typical COP of 3 and the overall cycle is no better than burning the gas directly. Now if the 50% efficiency figure quoted for this fuel cell is really just for the electricity generation side (i.e. does not take into account heat generation), then that may be more interesting. It would actually be a step up from standard power plants too so if it can be scaled up it should be.
Why not move the ballot boxes to a secure location like, I dunno, where they count them or where the ballots are stored in the case of a recount.
How do you verify the ballot boxes are not tampered with while they are being moved? And, by the way, the only reasonable place for counting the ballots is at the polling place right when the election closes. anything else means the voters lose control of the election.
You can even count them while they're there that night.
You could even publish partial results. No wait, that's probably not allowed!
I have a feeling that securing ballot boxes is a solved problem. (Or at least solved enough that this doesn't introduce undue risk.)
It's not solved at all, just ignored.
It's clear we're just not ready for electronic voting. Let's stick to paper ballots and re-visit this idea in twenty years or so.
There's no point revisiting the issue in 20 years. A voter won't be any more able to verify that the computer counting the votes runs clean software on clean hardware.
No one is going to count every ballot by hand. Why? Because hand-counting is far more inaccurate than machine counting.
I have participated in the counting of many elections in France. I'm not an election official or anything. In France the ballots are counted by volunteers under the watch of the ballot place officials, of party representatives and of the general public. Here is how it works.
Volunteers arrive a bit before the election closes. As soon as the election closes the officials open the transparent ballot box, count the envelopes and make sure that count matches the counter on the ballot box and in the registration book. Then they make groups of a hundred envelopes. Volunteers sit at (typically 3) tables in groups of four and receive the groups of a hundred envelopes one at a time. One volunteer opens the envelope, unfolds the ballot and reads it. He hands it to the volunteer sitting diagonally accross the table who reads the ballot aloud. The other two volunteers (also sitting diagonally from each other) have a sheet of paper and make a tick for the right candidate for every ballot. All the while officials, party representatives and general public hover around making sure there are no shenanigans. When the counting is done the election results for that polling place are then proclaimed right there on the spot.
I've never seen or heard of a discrepancy of more than one envelope in a thousand. In the few polling places that use e-voting machines however there's been discrepancies of up to 13 ballots in a polling place! So no, e-voting machines don't seem any more reliable.
Have you never been to a magician's stage show? He gets 500 people to all look at the wrong thing at the same time with close to 100% accuracy. And you are claiming that a well timed car backfire won't make people look. Really? Really?
Yes, really. A magic trick is one thing. Substituting hundreds of ballots / enveloppes without adding or removing even one is quite another. And even if your genius magician was able to tamper with the votes at a polling place, he'd only be able to impact a tiny fraction of the votes whereas a single programmer working on the e-voting software can tamper with every single vote.
Not only is there massive interest in openness and transparency in the voting process, but there also a need for extremely thorough vetting of the software, its design, and its update lineage.
Open-source is totally useless for e-voting. The reason is that on election day you cannot verify that the code that runs is the one that you reviewed.