"What next? Seriously, why is voting any different from these other very important uses of computers? Doesn't it make more sense to fix the problem rather than ban the machines?"
There's a very good reason why it's different: because in a secret ballot recorded by computer there is no way to verify the integrity of the data. In all of the examples you list above (electronic trading, ATMs), it is possible to compare the data in the computer with the data you expect to be in the computer.
For example, if you take €200 out of an ATM and your bank balance is debited by €500, this error will be pretty obvious to you. If you cast a vote in an electronic election, you can never know if your vote was recorded correctly.
There is no way to get around this problem without a voter-verifiable audit trail (VVAT).
In any case, the election in question is in 3 weeks' time, which doesn't really allow adequate time to fix the machines, even if that were possible.
Are you certain there are seals on the EPROMs themselves? I thought there was just a seal on the back panel of the machine. The Commission found that it was a trivial matter to circumvent this seal with the aid of razorblade.
The key system chosen by Nedap for both the locks on the voting computer is the "C&K YL Series 4 Tumbler Camlock". This lock always comes with the same key (marked "A126"), which probably explains why the same key is used on all 8000 ES3B machines throughout The Netherlands. Spare keys can be ordered separately online for roughly a Euro each by searching for
the product number: 115140126. We ordered, payed for and were subsequently supplied with 100 of these keys without any problem. According to the product datasheet3, typical applications for this lock include "copy machines and office furniture". Even if spare keys were not so readily available:
this is quite literally the type of lock we can open with a bent paperclip.
The reader unit has, as stipulated by law, a lock with a different key for the slot marked
'programming' (it is marked "A154"), which is used to erase the ballot memory modules and to
write new candidate lists to the modules. The key is of the same insecure type and the we expect it
to also be the same all over the country.
As well as being used in Holland and France, thousands of these NEDAP machines were bought by the Irish government with a view to replacing our paper election system with electronic voting. They had been used in a few pilot constituencies, and were due to be rolled out nationwide for the 2004 local and European elections. Luckily, determined lobbying by computer professionals (Irish Citizens for Trustworthy E-Voting) and others forced the Government to set up an independent Commission on Electronic Voting, who decided that they couldn't stand over the use of the machines without further testing.
The most recent report of the Commission (July 2006) concluded that the machines needed some modification but were basically okay, but that the software used to manage an election was basically a joke and should be scrapped. The Government tried to use this as vindication of their actions in procuring the system, even though they had been perfectly willing to let a nationwide election go ahead with dodgy software.
Even that fig-leaf of respectibility has now been removed, and I expect that the Government will soon be moving the machines out of their costly storage facilities, and into the nearest recycling centre. As the Dutch hackers showed that they could be used to play chess, perhaps an amusement arcade will take them off their hands.
Lots of info at the Irish Citizens for Trustworthy E-Voting site linked above, including a discussion list archive which has covered every imaginable angle on E-Voting.
"In a world that... knights Tim Berners-Lee without an understanding of the pre-WWW background of stateless client/server document architectures... on which he built..."
The world didn't knight Tim Berners-Lee, the British Government did, presumably because he's a British Citizen who has made a distinguished contribution to technology and society. We will probably never know whether a deeper understanding of the pre-WWW background of stateless client/server document architecture on the part of Queen Elizabeth and Tony Blair would have had any impact on this decision.
Doing a feature comparison between the two browsers' implementation of tabbed browsing, all I can come up with is as follows:
Favicons Safari: Not supported Firefox: Site's favicon is displayed in tab Verdict: Win for Firefox
Close button Safari: Each tab has its own close button Firefox: Single close button to the extreme right of the tabs bar closes the active tab Verdict: Win for Safari
"What hypocracy! When Apple wanted to "prove" that Intel was slower, they used GCC. Now that they need to use Intel, they're using the manufacturer's recommended compiler and getting better results."
That word is usually spelled "hypocrisy" or, in this case, "marketing".
...in fact no mouse at all with the Mac Mini. Perhaps the following line from Apple's Mac Mini page was intended to answer the Slashdot crowd's inevitable response:
And yes, Mac mini will take advantage of your two-button USB mouse with scroll-wheel and your favorite USB keyboard. Just plug them in.
Why not just offer Itunes in English, with whatever languages they already have translated, and have the currency in Euros? That'd cover a whole bunch of countries. People all listen to a lot of American music anyway with a handful of local artists. Just do the American music and work with the local artists/labels after the site's up.
If only it were that simple! It's not the case that Apple can negotiate a deal in the USA to sell tunes by US artists all over the world. Deals have to be done in each country with the people who hold the licenses for the music in that country. So we're not talking about negotiations about carrying tracks from local artists - the negotiations are about carrying the whole iTunes music catalogue.
I based my statement on the national language of countries.
I thought Afrikaans was the language of South Africa, and I just plain forgot about lil' old Liberia!
You also forgot about Ireland, where English is by some considerable distance the most commonly spoken language, as well as being one of two official national languages (the other being Irish).
So, yes, you were being a little bit ignorant - in the sense of lacking knowledge - but then nobody knows everything, so you can usually get away with any amount of ignorance with a little bit of humility.
The lack of iTMS for Ireland is particularly galling as Apple were ready to roll it out, but have been stymied by IMRO, the people who collect music royalties here. As Ireland is a market of only 4 million people, I can't see Apple putting too much effort into overcoming the objections of these local princelings.
I'm not too worried about the syncing aspect of this, as Apple or other developers will step in to fill this gap. What is sad, however, is the end of development of Claris Organizer/Palm Desktop.
I've used Palm Desktop continuously since 1998, when it was still Claris Organizer. The application has hardly changed at all in the intervening years, but in my view it remains the most elegant PIM available. It's also remarkably feature-complete for such an old product.
"Palm Desktop 4.0" brought OS X compatibility and some terminology changes ("Contacts" became "Addresses", "Tasks" became "To dos" etc.), but beyond that it was the same app. It even retained the scripts to open URLs in Cyberdog, or create form letters in MacWrite.
Now that the product has been orphaned, I'll probably switch to Entourage, which I find nowhere near as elegant. What are the chances Palm could be convinced to open-source Palm Desktop and allow it to live on?
"Bob Geldof's knighthood is not honorary. He is addressed as Sir Bob Geldof."
Sorry, but Geldof's knighthood is honorary, as will be confirmed by any of the biographical material found here. As an Irish citizen Geldof cannot hold a full knighthood unless
a) He takes British citizenship (or subjectitude, or whatever) AND
b) He gets approval from the Irish government (if he also wants to retain his Irish citizenship).
A couple of Irish citizens who also hold British nationality have accepted full knighthoods with the permission of the government, notably former Heinz supremo and current newspaper magnate Tony O'Reilly (or Dr Sir Anthony J. F. O'Reilly as he probably prefers to be known).
It's true that Bob Geldof is often styled 'Sir Bob' in the media, but this is not an official title, and the status of his knighthood is no different from that of Spielberg or Greenspan.
"Regardless of how you dislike the embattled outcome of the last election it was *orderely*. There was no military coup, there was no mass unrest. It was orderly. A process was followed."
Ah yes. In fact, I think it was Kodos who once said, "I am looking forward to an orderly election tomorrow, which will eliminate the need for a violent blood bath."
Ever since we were tasked with leveraging mission-critical neologisms and jargoning, in order to grow units and upsize language weirding across all territories.
The winners were chosen by a combination of more than 20,000 votes from the public at the Design Museum's website, along with the votes of a four-strong jury. Mr Ive won both the public and the jury vote.
I don't know what weighting they gave to the internet vote, but in any case Ive was the choice of the jury also. Normally in these processes, if the organisers have any sense at all, they don't give the internet portion more weight than a single juror.
Well, my point was that if you have the required access to implement such a fix, there are plenty of other fixes you could also implement which don't rely on there being a random element to the counting. So the system would not be significantly more exploitable if that random element existed, which in any case it doesn't.
But the general point is that any of these exploits are made significantly more feasible by keeping the source closed, as you say. The actual voting system should not be a significant factor - PR-STV may be complicated for humans to count, but I would expect that any software written to perform a count under this system would be relatively simple and the sources thereof easily readable.
"Both of these ways of fixing such an ellection would be impractical in a paper system, but quite easy in an electronic black box."
The anomaly whereby the order of the ballots can (theoretically) affect the outcome of the count does not apply in the electronic system, as far as I'm aware. This is because the computer can perform operations which are considered too time-consuming for manual counters, e.g. allocating third and subsequent preferences in proportion to all a candidate's ballots, rather than simply re-allocating those ballots which make up the surplus. In a paper system, these anomalies are tolerated because they generally even out over the course of the count, and it would, as you say, be very difficult to fix.
What's more, the type of fix you describe is quite complicated - you'd have to be able to make the system perform sorts and multiple counts, and somehow cover this up. If somebody had enough access to implement this, they would be better off implementing a much simpler fix, such as generating extra votes for a particular candidate etc.
6.3 Source code
The program source can be found at Nedap on the Specials_nts1 server in the/klanten/stemcomp/ierland directory. The bottom layer, the routines that operate directly on the hardware, are called the drivers and can be found in the drv/h and drv/src directories. The other layers called the application are found in the h and src directories. In the directory lang_h the language dependant header is stored.
Additionally only functions from the standard ANSI C libraries have been used, which are delivered with the compiler itself.
If this is not a publicly accessible server, why bother telling us where to find the source?
The other thing revealed by the document linked above is that the whole process makes heavy and totally unnecessary use of Microsoft software. We'll never have the source code to Microsoft Access, but that's the DB application which stores the ballots. Also, counting requires a bank of computers running Microsoft Window - surely this can not represent value for money for the State?
"Electronic Voting (while at the same time, eliminating "exit polls" which might have shown a different picture) allowed the Irish Government to obtain large "YES" votes in heavily populated areas that typically vote the party line, though not usually in numbers large enough to outnumber the rural population"
In any case, yours doesn't hold up to even the most rudimentary checking of facts. You're suggesting that the Yes vote was artificially inflated in urban constituencies, in order to make it outweigh the No vote in rural constituencies. If you took some time to examine the actual results, you'd see that this makes absolutely no sense:
In the first Nice referendum, every constituency in the country returned a 'No' majority, apart from Dublin South and Dun Laoghaire.
In the second referendum, every constituency in the country, including Dublin South and Dun Laoghaire, returned a 'Yes' majority.
In the second referendum, electronic voting was used in 7 constituencies. In these seven constituencies, 67% voted Yes - this is not significantly higher than the national total of 63%.
Not all of the 7 constituencies were 'heavily-populated' areas (by which I take it you mean 'densely-populated' - all constituencies have roughly the same population), and certainly not all heavily populated areas were included in the e-voting trial. The two constituencies mentioned above, which voted Yes the first time out, are densely-populated and were not included.
The suggestion that densely-populated areas follow the 'party line' doesn't hold up. The 'party' in question, Fianna Fail, the senior party in the government, is consistently stronger in rural areas than urban areas. While there has been an urban-rural split in previous referenda on social issues (such as divorce or abortion), this is not the case with EU referenda.
Exit polls were not eliminated. There was a proposal to prevent the holding of polls within seven days of an election (not sure if it was supposed to apply to referenda), but this did not affect exit polls. In any case, it was not introduced.
I'm not saying that fixing of the e-voting system couldn't or wouldn't happen, but your suggestion that this is what did happen in the 2002 referendum stretches credulity.
Mary is no longer Minister for Public Enterprise - she lost her seat in the last election and has been kicked upstairs to the Senate. In fact the Department of Public Enterprise no longer exists: it has been replaced with the Department of Transport under Seamus Brennan.
Slashdot Mirror
"What next? Seriously, why is voting any different from these other very important uses of computers? Doesn't it make more sense to fix the problem rather than ban the machines?"
There's a very good reason why it's different: because in a secret ballot recorded by computer there is no way to verify the integrity of the data. In all of the examples you list above (electronic trading, ATMs), it is possible to compare the data in the computer with the data you expect to be in the computer.
For example, if you take €200 out of an ATM and your bank balance is debited by €500, this error will be pretty obvious to you. If you cast a vote in an electronic election, you can never know if your vote was recorded correctly.
There is no way to get around this problem without a voter-verifiable audit trail (VVAT).
In any case, the election in question is in 3 weeks' time, which doesn't really allow adequate time to fix the machines, even if that were possible.
Are you certain there are seals on the EPROMs themselves? I thought there was just a seal on the back panel of the machine. The Commission found that it was a trivial matter to circumvent this seal with the aid of razorblade.
"they can't open it with a minibar key.."
Don't be so sure...
From their report (the PDF linked above):
As well as being used in Holland and France, thousands of these NEDAP machines were bought by the Irish government with a view to replacing our paper election system with electronic voting. They had been used in a few pilot constituencies, and were due to be rolled out nationwide for the 2004 local and European elections. Luckily, determined lobbying by computer professionals (Irish Citizens for Trustworthy E-Voting) and others forced the Government to set up an independent Commission on Electronic Voting, who decided that they couldn't stand over the use of the machines without further testing.
Interestingly enough, these Dutch hackers used the First Report of the Commission on Electronic Voting to glean a lot of the technical details about the machines.
The most recent report of the Commission (July 2006) concluded that the machines needed some modification but were basically okay, but that the software used to manage an election was basically a joke and should be scrapped. The Government tried to use this as vindication of their actions in procuring the system, even though they had been perfectly willing to let a nationwide election go ahead with dodgy software.
Even that fig-leaf of respectibility has now been removed, and I expect that the Government will soon be moving the machines out of their costly storage facilities, and into the nearest recycling centre. As the Dutch hackers showed that they could be used to play chess, perhaps an amusement arcade will take them off their hands.
Lots of info at the Irish Citizens for Trustworthy E-Voting site linked above, including a discussion list archive which has covered every imaginable angle on E-Voting.
The Internet is obviously still something of a "linguistic hinder".
"It should read, "Oracle to layoff 2000 people" Not jobs, people. People are losing their jobs. Its a sad thing."
Now that you've broken the ice on the pedantry thing, can I just say that it should be "Oracle to lay off 2,000 people"?
"In a world that ... knights Tim Berners-Lee without an understanding of the pre-WWW background of stateless client/server document architectures ... on which he built..."
The world didn't knight Tim Berners-Lee, the British Government did, presumably because he's a British Citizen who has made a distinguished contribution to technology and society. We will probably never know whether a deeper understanding of the pre-WWW background of stateless client/server document architecture on the part of Queen Elizabeth and Tony Blair would have had any impact on this decision.
"Safari's tabbed browsing isn't as elegant."
Doing a feature comparison between the two browsers' implementation of tabbed browsing, all I can come up with is as follows:
Favicons
Safari: Not supported
Firefox: Site's favicon is displayed in tab
Verdict: Win for Firefox
Close button
Is there a single other significant difference?Safari: Each tab has its own close button
Firefox: Single close button to the extreme right of the tabs bar closes the active tab
Verdict: Win for Safari
"What hypocracy! When Apple wanted to "prove" that Intel was slower, they used GCC. Now that they need to use Intel, they're using the manufacturer's recommended compiler and getting better results."
That word is usually spelled "hypocrisy" or, in this case, "marketing"....in fact no mouse at all with the Mac Mini. Perhaps the following line from Apple's Mac Mini page was intended to answer the Slashdot crowd's inevitable response:
Darwin is licensed under the Apple Public Source License (APSL), which is a non-free license in Stallman's opinion.
Why not just offer Itunes in English, with whatever languages they already have translated, and have the currency in Euros? That'd cover a whole bunch of countries. People all listen to a lot of American music anyway with a handful of local artists. Just do the American music and work with the local artists/labels after the site's up.
If only it were that simple! It's not the case that Apple can negotiate a deal in the USA to sell tunes by US artists all over the world. Deals have to be done in each country with the people who hold the licenses for the music in that country. So we're not talking about negotiations about carrying tracks from local artists - the negotiations are about carrying the whole iTunes music catalogue.
I'm not being ignorant. ...
I based my statement on the national language of countries. I thought Afrikaans was the language of South Africa, and I just plain forgot about lil' old Liberia!
You also forgot about Ireland, where English is by some considerable distance the most commonly spoken language, as well as being one of two official national languages (the other being Irish).
So, yes, you were being a little bit ignorant - in the sense of lacking knowledge - but then nobody knows everything, so you can usually get away with any amount of ignorance with a little bit of humility.
The lack of iTMS for Ireland is particularly galling as Apple were ready to roll it out, but have been stymied by IMRO, the people who collect music royalties here. As Ireland is a market of only 4 million people, I can't see Apple putting too much effort into overcoming the objections of these local princelings.
As Groucho Marx almost said, I wouldn't want to be part of any club where I could afford to be a member.
I'm not too worried about the syncing aspect of this, as Apple or other developers will step in to fill this gap. What is sad, however, is the end of development of Claris Organizer/Palm Desktop.
I've used Palm Desktop continuously since 1998, when it was still Claris Organizer. The application has hardly changed at all in the intervening years, but in my view it remains the most elegant PIM available. It's also remarkably feature-complete for such an old product.
"Palm Desktop 4.0" brought OS X compatibility and some terminology changes ("Contacts" became "Addresses", "Tasks" became "To dos" etc.), but beyond that it was the same app. It even retained the scripts to open URLs in Cyberdog, or create form letters in MacWrite.
Now that the product has been orphaned, I'll probably switch to Entourage, which I find nowhere near as elegant. What are the chances Palm could be convinced to open-source Palm Desktop and allow it to live on?
"Bob Geldof's knighthood is not honorary. He is addressed as Sir Bob Geldof."
Sorry, but Geldof's knighthood is honorary, as will be confirmed by any of the biographical material found here. As an Irish citizen Geldof cannot hold a full knighthood unlessa) He takes British citizenship (or subjectitude, or whatever) AND
b) He gets approval from the Irish government (if he also wants to retain his Irish citizenship).
A couple of Irish citizens who also hold British nationality have accepted full knighthoods with the permission of the government, notably former Heinz supremo and current newspaper magnate Tony O'Reilly (or Dr Sir Anthony J. F. O'Reilly as he probably prefers to be known).
It's true that Bob Geldof is often styled 'Sir Bob' in the media, but this is not an official title, and the status of his knighthood is no different from that of Spielberg or Greenspan.
"Regardless of how you dislike the embattled outcome of the last election it was *orderely*. There was no military coup, there was no mass unrest. It was orderly. A process was followed."
Ah yes. In fact, I think it was Kodos who once said, "I am looking forward to an orderly election tomorrow, which will eliminate the need for a violent blood bath."
"Since when is "architecting" a word?"
Ever since we were tasked with leveraging mission-critical neologisms and jargoning, in order to grow units and upsize language weirding across all territories.
From the Independent story:
I don't know what weighting they gave to the internet vote, but in any case Ive was the choice of the jury also. Normally in these processes, if the organisers have any sense at all, they don't give the internet portion more weight than a single juror.
Well, my point was that if you have the required access to implement such a fix, there are plenty of other fixes you could also implement which don't rely on there being a random element to the counting. So the system would not be significantly more exploitable if that random element existed, which in any case it doesn't.
But the general point is that any of these exploits are made significantly more feasible by keeping the source closed, as you say. The actual voting system should not be a significant factor - PR-STV may be complicated for humans to count, but I would expect that any software written to perform a count under this system would be relatively simple and the sources thereof easily readable.
"Both of these ways of fixing such an ellection would be impractical in a paper system, but quite easy in an electronic black box."
The anomaly whereby the order of the ballots can (theoretically) affect the outcome of the count does not apply in the electronic system, as far as I'm aware. This is because the computer can perform operations which are considered too time-consuming for manual counters, e.g. allocating third and subsequent preferences in proportion to all a candidate's ballots, rather than simply re-allocating those ballots which make up the surplus. In a paper system, these anomalies are tolerated because they generally even out over the course of the count, and it would, as you say, be very difficult to fix.
What's more, the type of fix you describe is quite complicated - you'd have to be able to make the system perform sorts and multiple counts, and somehow cover this up. If somebody had enough access to implement this, they would be better off implementing a much simpler fix, such as generating extra votes for a particular candidate etc.
The government's own guide to the electronic voting system has this to say on the source code:
If this is not a publicly accessible server, why bother telling us where to find the source?
The other thing revealed by the document linked above is that the whole process makes heavy and totally unnecessary use of Microsoft software. We'll never have the source code to Microsoft Access, but that's the DB application which stores the ballots. Also, counting requires a bank of computers running Microsoft Window - surely this can not represent value for money for the State?
"Electronic Voting (while at the same time, eliminating "exit polls" which might have shown a different picture) allowed the Irish Government to obtain large "YES" votes in heavily populated areas that typically vote the party line, though not usually in numbers large enough to outnumber the rural population"
Given that five times more money was spent by the Yes campaign than the No campaign, why bother with a conspiracy theory?
In any case, yours doesn't hold up to even the most rudimentary checking of facts. You're suggesting that the Yes vote was artificially inflated in urban constituencies, in order to make it outweigh the No vote in rural constituencies. If you took some time to examine the actual results, you'd see that this makes absolutely no sense:
I'm not saying that fixing of the e-voting system couldn't or wouldn't happen, but your suggestion that this is what did happen in the 2002 referendum stretches credulity.
You only pay $9.99 for albums on the iTunes Music Store - I understand this is a fairly reasonable discount from retail prices in the US.
Mary is no longer Minister for Public Enterprise - she lost her seat in the last election and has been kicked upstairs to the Senate. In fact the Department of Public Enterprise no longer exists: it has been replaced with the Department of Transport under Seamus Brennan.
I think responsibility for the Telecommunications sector now lies with the clumsily-titled Department of Communications, Marine and Natural Resources. There are also a number of quangoes operating under the Taoiseach's Department, such as the Information Society Commision. Government chief whip Mary Hannifin is involved in all this somewhere, I'm not sure how. They keep touting her as the "Information Society Minister".