Why would we spend billions of dollars on a system that is less secure or differently insecure than the current system? That's just stupid.
My understanding is that the major design objective here is to have a user-friendly device that correctly records the intent of the voter. There doesn't seem to be a massive election fraud issue in the US. We just need to ensure that in making the system easier to use (especially for those with disabilities) we don't make it less secure than today.
You can't print a unique id on the print outs because the votes have to be anonymous.
When I vote today, I am handed a ballot after I present my credentials. I trust that there is not a mark that ties me to the ballot. Same thing with a VOTE_ID, it would be generated sequencially. I would have to trust that there is no cross reference to my identity. Hopefully the voting process should make that obvious.
You also can't know that the end of day print out isn't altered in the same way as the voter print outs. They only way to be sure, would be to count _all_ the paper ballots and make sure the total sent to the master counting system jibes with the e-total.
I think a reporting / audit process could easily be designed but it would require a VOTE_ID to be able to use a sampling process.
Response: 4. The point here is not to offer more security than is available today. Today a judge could wait until the end of the day, bribe the other judges and replace current paper ballots. If we want a system more secure than today's systems, I agree we need a whole new plan.
9. The point of an audit is that you check a small (0.5% - 10% of the ballots against a report. If the vote report for a location listed all the VOTE_IDs above the total for the candidate. The audit compares the receipts to the reports. Even one mis-match (a VOTE_ID on the receipt in conflict with the report) reveals tampering. Much more reliable than today.
10b2. Of course, this is doomsday, this is a faulty report. The only choices are reschedule or accept a tampered vote. I think the choice is obvious.
1. User-Friendly Device receives user's vote choices. 2. Device records choices. 3. Device prints receipt reflecting votes for user. 4. User inspects receipt and (with the aid of Judge) can correct their vote. 5. User decides receipt properly reflects their desired vote. 6. User places receipt in sealed ballot box. 7. Device reports end-of-day results to Parent Device. 8. Parent Device reports detail and total results for district. 9. Judges audit detail against paper reciept according to local procedures. 10a. Results pass audit Judges certify the results. 10b. Results do not pass audit. Judges direct paper receipts must be counted entirely. 10b1. If discrepency would not effect outcome, election is certified. 10b2. If discrepency would effect outcome, election is declared null and void. A new vote is scheduled.
Above use case is copyrighted by author and hereby declared to usable by anyone without compensation or reference.
There was talk about raising $2 billion in an IPO, most of which will probably buy out VCs and founders. Some $ could pay down debt on the hardware.
But what would Google do with a $1 billion to invest in the business? Buy superbowl ads?
I think Google would work better as a private company. I don't see them needing a lot of capital.
Well said. Google is a great service, but I also think it's pretty much reached it's a mature level revenue potential. Growth will come, but slowly. And there are incremental costs as the web gets bigger.
My Mac is less than 12 months old and they want me to pay $129 for an upgrade? New features look cool, but as someone who recently "switched" from Intel/MS, I feel a little cheated. I'd pay less than $50 for the upgrade. One more of these and I'll give in and learn linux.
Recommendation for a linux distribution for me (one that lets me install the OS on a Intel box with a CD and instructions that would fit on one page)?
Look, there is nothing that makes a PHB's hole tighten up quicker than the word "lawsuit".
If you want OSS to be shunned to hobby land, then go ahead and cheer lawsuits against big companies.
"Under the license, if you distribute GPL software in a product, you must also distribute the software's source code. And not just the GPL code, but also the code for any 'derivative works' you've created". From the article
The threat to adoption here is that a "derivative" work is whatever a judge decides it is. A PHB would be reckless to mix commercial and OSS software together in a system if there is any ambiguity about whether the commercial part is seperate or a derivative. It could create a huge liability.
The GPL would be stronger if there was a way for PHB to be absolutely certain that the OSS authors have no claim on the commericial works.
Of course, if you are really committed to OSS, you should just put your work in the public domain.
Almost all the posts I read basically advocate for software diversity. Microsoft, rightly, has become the target of our ire because of their anti-competitive practices (which the current justice department gave up on changing).
However that does not mean that we should cheer any stone throne at MS. What is at stake here is the right to sell software "AS IS" (a condition of most software licenses).
1) if software makers are forced to warranty their software it will only benefit well capatalized companies in the long run and 2) the security problems with Microsoft are probably the best "selling point" of software diversity in general and more specifically OSS.
Good design would require that a trial system be designed with weights given to factors that are expected to be a problem. A "hold out" group of data that contains the targets should not be included in the system design. Only once the system is designed should the sample with the "hold out" data be included. This prevents "drawing the target around the bullet hole".
If you are interested...
The economic effect your are referring to is more frequently described as a network effect. Lock in is more frequently used to refer to switching costs.
A colleague of mine who works for Kryptonite says in response to every smart ass (who has the great lock breaking solution) is that, with security, money is only buying you "time and noise". In other words, any detirmined thief will get in. The price we pay is to delay him and make it noisy to get in.
First, thanks for the first economic discussion of the issue.
I'm still wrestling with the effect on consumers of price descrimination. I understand the benefit that some products get made because of price descrimination (as illustrated in page 5 of the paper). But I am not ready to accept that price discrimination is on the whole economically efficient (or at least that the negative effects have been properly identified).
...what is best for the consumer is to pay as close to the production value as possible, whereas what is best for the vendor is for them to pay as close to the utility value as possible
The difference between what we pay and what we are willing to pay is the reward we get for buying the product (our consumer surplus, see graphs).
If we pay the maximum we are willing to pay, we get no benefit from the product, we gave all the benefit to the supplier. In fact we are indifferent to buying it or not buying it (i.e. we no longer WANT it).
Perfect price discrimination imposed by a monopoly would turn us into drones who have to pay huge prices for essentials like food, clothing, and shelter.
Thankfully competition comes to the rescue, but by how much is difficult to understand.
If we live in a world where everyone nickels and dimes each other to the point where the total profit through the supply chain is on the order of pennies on the dollar, then it drags down the whole system.
That does not quite apply because the costs that dictate the supply curve are usually said to include "normal" profits. You only earn abnormal profits when you have a monopoly (i.e. a patent). No worries about lack of innovation.
So firm A discovers I am willing to pay $400 for a DVR. They come to my door and offer it to me. (I stand there forever because I can't decide whether or not to buy it, at $400 I don't care.) Since it only costs $200 to make one more, firm B is willing to sell me one at $390. (I would buy it if I thought these were the only two DVRs in the world.) But now I've realized that if I do a little bit of work, I can save some money (i.e. capture a little more surplus).
I continue to look around, maybe wait for sales, and as I wait I find that I can save more money. So with companies trying to price discriminate (based even on perfect knowledge of my utility value), I had to work harder to capture that surplus. The suppliers have added SEARCH COSTS.
The supplier who eventually sold me the product caputered the surplus that I was willing to surrender because the savings I was finding was no longer enough to keep searching.
Search costs don't seem like a very productive use of my time. I can't buy something else with the money that I gave up to stop looking. How can added search costs be efficient?
Isn't it a bad thing to have consumer surplus disappear? How can we predict how much prices in general will fall?
Slashdot Mirror
Why would we spend billions of dollars on a system that is less secure or differently insecure than the current system? That's just stupid.
My understanding is that the major design objective here is to have a user-friendly device that correctly records the intent of the voter. There doesn't seem to be a massive election fraud issue in the US. We just need to ensure that in making the system easier to use (especially for those with disabilities) we don't make it less secure than today.
You can't print a unique id on the print outs because the votes have to be anonymous.
When I vote today, I am handed a ballot after I present my credentials. I trust that there is not a mark that ties me to the ballot. Same thing with a VOTE_ID, it would be generated sequencially. I would have to trust that there is no cross reference to my identity. Hopefully the voting process should make that obvious.
You also can't know that the end of day print out isn't altered in the same way as the voter print outs. They only way to be sure, would be to count _all_ the paper ballots and make sure the total sent to the master counting system jibes with the e-total.
I think a reporting / audit process could easily be designed but it would require a VOTE_ID to be able to use a sampling process.
Response:
4. The point here is not to offer more security than is available today. Today a judge could wait until the end of the day, bribe the other judges and replace current paper ballots. If we want a system more secure than today's systems, I agree we need a whole new plan.
9. The point of an audit is that you check a small (0.5% - 10% of the ballots against a report. If the vote report for a location listed all the VOTE_IDs above the total for the candidate. The audit compares the receipts to the reports. Even one mis-match (a VOTE_ID on the receipt in conflict with the report) reveals tampering. Much more reliable than today.
10b2. Of course, this is doomsday, this is a faulty report. The only choices are reschedule or accept a tampered vote. I think the choice is obvious.
Does this answer the objections raised above?
The use case seems so obvious to me:
1. User-Friendly Device receives user's vote choices.
2. Device records choices.
3. Device prints receipt reflecting votes for user.
4. User inspects receipt and (with the aid of Judge) can correct their vote.
5. User decides receipt properly reflects their desired vote.
6. User places receipt in sealed ballot box.
7. Device reports end-of-day results to Parent Device.
8. Parent Device reports detail and total results for district.
9. Judges audit detail against paper reciept according to local procedures.
10a. Results pass audit Judges certify the results.
10b. Results do not pass audit. Judges direct paper receipts must be counted entirely.
10b1. If discrepency would not effect outcome, election is certified.
10b2. If discrepency would effect outcome, election is declared null and void. A new vote is scheduled.
Above use case is copyrighted by author and hereby declared to usable by anyone without compensation or reference.
There was talk about raising $2 billion in an IPO, most of which will probably buy out VCs and founders. Some $ could pay down debt on the hardware. But what would Google do with a $1 billion to invest in the business? Buy superbowl ads? I think Google would work better as a private company. I don't see them needing a lot of capital.
Well said. Google is a great service, but I also think it's pretty much reached it's a mature level revenue potential. Growth will come, but slowly. And there are incremental costs as the web gets bigger.
My Mac is less than 12 months old and they want me to pay $129 for an upgrade?
New features look cool, but as someone who recently "switched" from Intel/MS, I feel a little cheated.
I'd pay less than $50 for the upgrade. One more of these and I'll give in and learn linux.
Recommendation for a linux distribution for me (one that lets me install the OS on a Intel box with a CD and instructions that would fit on one page)?
Look, there is nothing that makes a PHB's hole tighten up quicker than the word "lawsuit".
If you want OSS to be shunned to hobby land, then go ahead and cheer lawsuits against big companies.
"Under the license, if you distribute GPL software in a product, you must also distribute the software's source code. And not just the GPL code, but also the code for any 'derivative works' you've created". From the article
The threat to adoption here is that a "derivative" work is whatever a judge decides it is. A PHB would be reckless to mix commercial and OSS software together in a system if there is any ambiguity about whether the commercial part is seperate or a derivative. It could create a huge liability.
The GPL would be stronger if there was a way for PHB to be absolutely certain that the OSS authors have no claim on the commericial works.
Of course, if you are really committed to OSS, you should just put your work in the public domain.
Almost all the posts I read basically advocate for software diversity. Microsoft, rightly, has become the target of our ire because of their anti-competitive practices (which the current justice department gave up on changing).
However that does not mean that we should cheer any stone throne at MS. What is at stake here is the right to sell software "AS IS" (a condition of most software licenses).
1) if software makers are forced to warranty their software it will only benefit well capatalized companies in the long run and 2) the security problems with Microsoft are probably the best "selling point" of software diversity in general and more specifically OSS.
Good design would require that a trial system be designed with weights given to factors that are expected to be a problem. A "hold out" group of data that contains the targets should not be included in the system design. Only once the system is designed should the sample with the "hold out" data be included. This prevents "drawing the target around the bullet hole".
If you are interested...
The economic effect your are referring to is more frequently described as a network effect. Lock in is more frequently used to refer to switching costs.
A colleague of mine who works for Kryptonite says in response to every smart ass (who has the great lock breaking solution) is that, with security, money is only buying you "time and noise". In other words, any detirmined thief will get in. The price we pay is to delay him and make it noisy to get in.
First, thanks for the first economic discussion of the issue.
I'm still wrestling with the effect on consumers of price descrimination. I understand the benefit that some products get made because of price descrimination (as illustrated in page 5 of the paper). But I am not ready to accept that price discrimination is on the whole economically efficient (or at least that the negative effects have been properly identified).
The difference between what we pay and what we are willing to pay is the reward we get for buying the product (our consumer surplus, see graphs).
If we pay the maximum we are willing to pay, we get no benefit from the product, we gave all the benefit to the supplier. In fact we are indifferent to buying it or not buying it (i.e. we no longer WANT it).Perfect price discrimination imposed by a monopoly would turn us into drones who have to pay huge prices for essentials like food, clothing, and shelter.
Thankfully competition comes to the rescue, but by how much is difficult to understand.
If we live in a world where everyone nickels and dimes each other to the point where the total profit through the supply chain is on the order of pennies on the dollar, then it drags down the whole system.
That does not quite apply because the costs that dictate the supply curve are usually said to include "normal" profits. You only earn abnormal profits when you have a monopoly (i.e. a patent). No worries about lack of innovation.
So firm A discovers I am willing to pay $400 for a DVR. They come to my door and offer it to me. (I stand there forever because I can't decide whether or not to buy it, at $400 I don't care.) Since it only costs $200 to make one more, firm B is willing to sell me one at $390. (I would buy it if I thought these were the only two DVRs in the world.) But now I've realized that if I do a little bit of work, I can save some money (i.e. capture a little more surplus).
I continue to look around, maybe wait for sales, and as I wait I find that I can save more money. So with companies trying to price discriminate (based even on perfect knowledge of my utility value), I had to work harder to capture that surplus. The suppliers have added SEARCH COSTS.
The supplier who eventually sold me the product caputered the surplus that I was willing to surrender because the savings I was finding was no longer enough to keep searching.
Search costs don't seem like a very productive use of my time. I can't buy something else with the money that I gave up to stop looking. How can added search costs be efficient?Isn't it a bad thing to have consumer surplus disappear? How can we predict how much prices in general will fall?