Should we discontinue EVMs?

Sanku · Post by **Sanku** » 18 Jul 2009 01:08

BEL etc should audit always. I would scream for it till I died if I was at EC/BEL, irrespective of whether BJP or anyone else won.

A clear test plan is the corner stone of all engineering.

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 01:21

Sanku wrote:BEL etc should audit always. I would scream for it till I died if I was at EC/BEL, irrespective of whether BJP or anyone else won.

A clear test plan is the corner stone of all engineering.

The simplest audit will be give the losers (all those who earn deposit money back/top 3-4 of a parliamentary constituency) to be able to choose 1 to 3 EVMs of his choice and ask to test however he wants getting any experts or equipment he needs all captured on camera. This will stop any accusations other than replacement of EVMs. Those who wants to avail of this facility will have to pay for replacement costs of EVMs.

But for this it will require that the complete design, source code, firmware code of EVMs to be open source. Why is the EC afraid to do this simple thing.

Basically instead of the EC saying we "know" that EVMs are fool-proof, shift the onus of proof burden to the otherside, saying prove that EVMs are not fool-proof. If any vulnerabilities are found in this testing, they can be rectified up in the next iteration.

The thing is Ceaser's wife has to be beyond doubt.

Pranav · Post by **Pranav** » 18 Jul 2009 07:00

Pranav wrote:
As regards how to call the trojan, you need logic on the processor through which the key-presses are filtered. When the unusual key presses which constitute the trigger mechanism are encountered, control is transferred to the known location of the trojan code. (RMji - the key in favour of which the rigging is to be done would be the first key to be pressed once control is transferred to the trojan.)

To clarify: the scenario is that you press, for example, keys 1 and 7 simultaneously to transfer control to the trojan. Then press key x to rig in favour of x.

Special-purpose logic for filtering key-strokes and an additional storage area for trojan code would be required only of the PROM were being audited, and the riggers wanted to defeat such PROM audits.

In the current scenario, where no audits are being allowed, you could put the trojan in the PROM itself and nobody would know.

One question is whether the PSUs are writing the code themselves or sourcing it from elsewhere. Another question is how many people at the PSUs would know if the master copy of the binary was tampered with by a manager.

And of course the totalisers. Would be interesting to find out if totalisers were used in Chidambaram's constituency.

Rahul Mehta · Post by **Rahul Mehta** » 18 Jul 2009 07:31

Pranav, Raju etc

Pls ignore the insults the EVM-lovers are throwing. If you reply back, they will immediately file complaint that "thread is spiraling down" and ask admins to kill the thread. So if you want this thread to go on, simply ignore their posts which have insults and merely focus on what they write, and also point out what they refuse to write.

So pls focus on the fact that pro-EVM are shying away from answering this question : what is cost of 71 cr paper ballots? Why is Tanaji refusing to answer this question? Why is Dileep not answering it? Why is NO pro-EVM guy answering it? I must have asked this question to them 10 times. Now I request you to ask them the SAME question.

---

Muppala,

Here is my wish : for N hours anti-RM elements write against what I write, I wish they would spend N seconds in reading what I write. Because in the EVM replacement conspiracy I proposed. I did NOT say that CIA agents replaced EVMs at booth. I said that CIA agents replaced EVMs in CEC warehouse. Replacing say 200,000 EVMs at booth level would need CIA to bribe 10000s of officials and ask them to look aside. And they would need 1000-2000 field agents. Where as replacing 200,000 EVMs in CEC warehouse needs only the support of PM, Chawala some top 10 officials in CEC and no more than 100 field agents. This is manageable for CIA. And I wont worry about manufacturing 1400,000 fake EMVs and 700,000 fake Control Units. That is child's play for CIA. The only problem is replacement.

So say 50% of EVMs are stored in district warehouses and 50% in CEC warehouses. The ones in CEC warehouse got replaced with ease. And the ones in district warehouses will get replaced ONLY iff Collectors are Missionary/CIA agents. Possible in many districts in AP, very difficult or almost impossible in Gujarat.

In any case, I dont think EVMs can be replaced at booth levels in large scale.

-----

Dileep,

Following what I wrote

Rahul Mehta: 1. If the chip-manufacturer knows the full code inside the ROM of the EVM, he can know which RAM locations are used to store the counts of each row. eg say counts of Row1 are stored in word number 80. Say count of Row2 are stored in word number 81 and so forth.

2. Now code in ROM will make predictable changes in CPU registers when a key is pressed. eg say row-1 is pressed, then ROM will write 01 in CPU register-D. Say row-2 is pressed, then ROM will write 02 in register-D and so forth. So if ROM code is known, one can state the predictable changes ROM will make in CPU register when a key is pressed. So the microcode can know which key was pressed.

3. So the microcode inside controller can know which keys were pressed from (2). So when a key combination comes, microcode will increase the contents of the RAM location corresponding to Row #k. (The k will be part of activation key sequence).

Following is what you said

0. I appreciate this RM. You have a technical proposal here, which could be rationally analyzed and answered. Thanks.

1. The RAM locations are decided by the compiler at the time the source is compiled. So, it would be different from version to version. There is no guarantee that the "Stolen" version of the software is going to be ultimately used in the EVM where the chips are getting soldered.

2. The registers are GENERAL PURPOSE, and will be used for all calculations. If you look at them, their values go on changing. No one can detect a key press by looking at the registers.

3. Since 1 and 2 can't be done, and those are necessary conditions for 3, it can't be done.

0. You are welcome. If you notice, thru out the thread, I have confined myself to scenario building only.

1. BEL will not compile the code every now and then. Once the testing is over and industrial production starts, the code will be recompiled only if there is design change or a critical bug is found. So yes, if there is re-compile the trojan in micro-code will fail. But the trojan in microcode will be effective as long as ROM does not change. And trojan can be written in way that RAM location would be input from the keypad. So activation code for 10 different versions of EVMs will be different. But that is still manageable.

2. Each time keyX is pressed, register value will show a predictable change that corresponds to X. So if microcode has a routine inside it that monitors changes in the register values, it can detect the pattern if and when the pattern comes.

So microcode level trojan is possible , if the chip manufacturer is willing

---------------

Muppala,

BEL does not have capability to make chips like Intel 8087, 80287 or 80387 or similar such chips made by Motorola. Nor does anyone in India. There are no such foundries in India. BEL imports these chips. Say BEL imports these chips from Intel or Motorola or some company in Japan. If CIA gets holds of ROM code inside EVM, and asks the chip manufacturer to make chips that would have a trojan activatable with key sequence 178535-K (K = row number of the candidate CIA wants to push up in poll) , then the company which is making the chip can put such a trojan inside microcode of the chip. Now I claim that the company which provided chips are under influence of CIA. IOW, BEL imported the chips from CIA . Now do you believe that EVMs dont have trojan in microcode?

The trojan would need some human to input key sequence 178535-K AFTER candidate number is assigned and before poll ends. That can be done at CEC warehouse, Collector warehouse, or during the poll.

And the scary part is - if the trojan is inside the microcode, NO audit in this world can detect it !! Why? Because there are no equipment to read microcode once the chip is made.

Dileep · Post by **Dileep** » 18 Jul 2009 07:35

Let me summarize the technical part:

There are three modes of attack proposed till now:

1. Corrupt the binary executable by influencing BEL.
2. Corrupt the source of the chips by influencing the chip vendor
3. Fabricate spurious machines and insert them in the process.

The first thing to consider is, how do we verify these were not done in the machines used in E-2009. The second thing is how to institute a system where these are taken care of in a reliable and transparent manner.

1, is the only issue that has any chance of being pulled off. It can easily be verified by pulling out random samples of the machines, hook up a JTAG reader, read the binary and compare to the master.

2, is very far fetched, but it can be verified by removing samples of the chip, de-package it, and visually compare the top mask of the chip with the master mask. Any change, be it a single transistor or interconnect, will show up on the top mask. This process is routinely done for failure mode analysis in chip packaging houses.

3, is easily verified by pulling random samples of the EVMs and verifying them. The machines are expected to retain the counts, and counts are on record. So, you can first verify that the machines are the same as the one used in the poll. Then open them an see what is inside. Even the mighty CIA can't match all components and PCB to be exactly the same as that manufactured by BEL.

Rahul Mehta · Post by **Rahul Mehta** » 18 Jul 2009 07:48

Dileep wrote:Let me summarize the technical part:

There are three modes of attack proposed till now:

1. Corrupt the binary executable by influencing BEL.
2. Corrupt the source of the chips by influencing the chip vendor
3. Fabricate spurious machines and insert them in the process.

The first thing to consider is, how do we verify these were not done in the machines used in E-2009. The second thing is how to institute a system where these are taken care of in a reliable and transparent manner.

1, is the only issue that has any chance of being pulled off. It can easily be verified by pulling out random samples of the machines, hook up a JTAG reader, read the binary and compare to the master.

2, is very far fetched, but it can be verified by removing samples of the chip, de-package it, and visually compare the top mask of the chip with the master mask. Any change, be it a single transistor or interconnect, will show up on the top mask. This process is routinely done for failure mode analysis in chip packaging houses.

3, is easily verified by pulling random samples of the EVMs and verifying them. The machines are expected to retain the counts, and counts are on record. So, you can first verify that the machines are the same as the one used in the poll. Then open them an see what is inside. Even the mighty CIA can't match all components and PCB to be exactly the same as that manufactured by BEL.

There is NO way to read microcode of chip. !!

So if the trojan is in the microcode of the chip, you can never even dig it out.

And as I said that CIA replaced real EVMs in CEC warehouse with fake before poll, and after counting they have put the real ones back with counters updated. Even though CEC does not make EVM counts public, Chawla will have with him. He gave it to CIA agent in CEC warehouse, who connected the Control Unit (or EVM) with PC, the PC simulated the protocol and so the EVM has now counters set to same value as on counting tables. So replacement theory says that post-mortem will not reveal any foul play.

[To give an analogy : God made the world 5000 years ago along with fossils which look 500,000 year old. So a 500,000 year old fossil does not prove that God did not make earth 5000 years ago.

]

=======================

Dear audit lovers,

1. If the trojan is in ROM, audit will help ONLY if one can get the ROM contents of the micro controller chip.It is possible, but not easy.

2. If the trojan was implanted in microcode by the chip manufacturer himself, NO equipment in world can dig that trojan out.

3. And if EVMs are stored in CEC warehouses, they can be replaced within few hours with look alikes.

Rahul Mehta · Post by **Rahul Mehta** » 18 Jul 2009 07:53

.

Paper ballots cost no more than Rs 70 cr, or perhaps less than Rs 35 cr.

To replace them, EVM lovers made Rs 5500 * 700,000 = about Rs 400 cr. Or may be Rs 700 cr if we assume EVM cost be Rs 10000.

And now EVM-lovers want to audit EVM, ROM, microcodes, chips, PCBs and all the software like randiomization etc. These audits would run into crores of rupees of cost !!

WhereTH is savings?

.

Dileep · Post by **Dileep** » 18 Jul 2009 07:56

Rahul Mehta wrote: 1. BEL will not compile the code every now and then. Once the testing is over and industrial production starts, the code will be recompiled only if there is design change or a critical bug is found. So yes, if there is re-compile the trojan in micro-code will fail. But the trojan in microcode will be effective as long as ROM does not change. And trojan can be written in way that RAM location would be input from the keypad. So activation code for 10 different versions of EVMs will be different. But that is still manageable.

You can't input the RAM location from keypad. It needs entry of at least eight hex digits. How will you manage it by the available keypad? You will have to be an android to accurately key it in using a non-numerical keypad.

2. Each time keyX is pressed, register value will show a predictable change that corresponds to X. So if microcode has a routine inside it that monitors changes in the register values, it can detect the pattern if and when the pattern comes.

No. There will not be any predictable change. The value is always read into the accumulator, which is used for all calculations, so the value of the register will have no pattern.

So microcode level trojan is possible , if the chip manufacturer is willing

Microcode within the chip is NOT executable program. It is a digital logic block used to do a certain arithmetic or logic operation. It can't do complex operations like the executable program does. Here is what wiki says about microcode http://en.wikipedia.org/wiki/Microcode. It can't do any calculation, comparison etc.

It is impossible to do what you say, ie monitor the key press and modify RAM locations from microcode. It is possible only from executable instruction sequences.

BEL does not have capaibility to make chips like Intel 8087, 80287 or 80387 or similar such chips made by Motorola. BEL imports these chips. Say BEL imports these chips from Intel or Motorola or some company in Japan. If CIA gets holds of ROM code inside EVM, and asks the chip manufacturer to make chips that would have a trojan activatable with key sequence 178535-K (where K is the row number Congress candidate), then the company which is making the chip can put such a trojan inside microcode of the chip. Now I claim that the company which provided chips are under influence of CIA. IOW, BEL imported the chips from CIA . Now do you believe that EVMs dont have trojan in microcode?

The fact that microcode can't do what you say notwithstanding, microcode programming is not like your C programming. It involves re-design and re-verification of the chip, plus re-making of the masks and running the fabrication. Do you know how much time, effort and manpower that takes? Certainly you can't do it within the normal lead time of 4 weeks for the chips!

And the scary part is - if the trojan is inside the microcode, NO audit in this world can detect it !! Why? Because there are no equipment to read microcode once the chip is made.

The fact that microcode can't do what you say notwithstanding, microcode is stored in transistor arrays on the chip, and any additional code will show up as additional transistors and interconnects. You can de-package the chip and visually verify it.

Dileep · Post by **Dileep** » 18 Jul 2009 08:07

Rahul Mehta wrote: There is NO way to read microcode of chip. !!
So if the trojan is in the microcode of the chip, you can never even dig it out.

You can't read it, but you can verify its existence, because it shows up on the chip layout.

And as I said that CIA replaced real EVMs in CEC warehouse with fake before poll, and after counting they have put the real ones back with counters updated. Even though CEC does not make EVM counts public, Chawla will have with him. He gave it to CIA agent in CEC warehouse, who connected the Control Unit (or EVM) with PC, the PC simulated the protocol and so the EVM has now counters set to same value as on counting tables. So replacement theory says that post-mortem will not reveal any foul play.

There is no PC interface for EVM. If you make a simulator for the ballot unit, you still need to press the key on the CU for each vote. And I know that there is a time delay of at least 20 seconds from the moment the PO presses the key on the CU and the BU accepts the vote key. Because I saw the guy press the key, and the green light came on my BU only after that time.

[To give an analogy : God made the world 5000 years ago along with fossils which look 500,000 year old. So a 500,000 year old fossil does not prove that God did not make earth 5000 years ago.

]

You know what, your arguments sound exactly like those of the 5000 year proponents.

2. If the trojan was implanted in microcode by the chip manufacturer himself, NO equipment in world can dig that trojan out.

A dremel tool and a microscope will show you.

3. And if EVMs are stored in CEC warehouses, they can be replaced within few hours with look alikes.

Like a few days in fact, if you want to update the counts.

Muppalla · Post by **Muppalla** » 18 Jul 2009 08:39

Dileep wrote:Let me summarize the technical part:

There are three modes of attack proposed till now:

1. Corrupt the binary executable by influencing BEL.
2. Corrupt the source of the chips by influencing the chip vendor
3. Fabricate spurious machines and insert them in the process.

The first thing to consider is, how do we verify these were not done in the machines used in E-2009. The second thing is how to institute a system where these are taken care of in a reliable and transparent manner.

1, is the only issue that has any chance of being pulled off. It can easily be verified by pulling out random samples of the machines, hook up a JTAG reader, read the binary and compare to the master.

2, is very far fetched, but it can be verified by removing samples of the chip, de-package it, and visually compare the top mask of the chip with the master mask. Any change, be it a single transistor or interconnect, will show up on the top mask. This process is routinely done for failure mode analysis in chip packaging houses.

3, is easily verified by pulling random samples of the EVMs and verifying them. The machines are expected to retain the counts, and counts are on record. So, you can first verify that the machines are the same as the one used in the poll. Then open them an see what is inside. Even the mighty CIA can't match all components and PCB to be exactly the same as that manufactured by BEL.

Why (2) is far fetched? Between all possibilites I see (2) along with hacked randomizer is a probable one. I do not believe in external vendors. At least there should be open verification and all the names of vendors involved everything related EVMs should be published. Transparancy is key to get over these conspiracy theories.

To me (1) is not possible. There will be someone who can spill the beans.

Muppalla · Post by **Muppalla** » 18 Jul 2009 08:57

Rahul Mehta wrote: Muppala,

BEL does not have capability to make chips like Intel 8087, 80287 or 80387 or similar such chips made by Motorola. Nor does anyone in India. There are no such foundries in India. BEL imports these chips.

May be India manufactures these chips. What is they type of the chip in EVM? Is it of 8087 type architecture? They may not have at the ones that produces volumes. See the following:

http://www.business-standard.com/india/ ... wn/237631/

Chips developed by SCL, which was under the department of information technology till 2004-05, have been used in Indian satellites for imaging and other applications. The Defence Research Development Organisation has also used SCL chips and so has the Atomic Energy Commission.

SCL is also learnt to be participating in a PSU consortium--including Bharat Electronics Ltd, Electronics Corporation of India Ltd—in the Indian citizen’s multipurpose national identity card project.

SCL has recently added to its capability to fabricate micro-electro mechanical systems (MEMS), an initiative funded under the National Programme on Smart Materials. Under this, SCL is believed to have taken up fabrication of a variety of sensors including those for use in ‘radio-sonde’ application for the India Meteorological Department.

SCL’s wafer-fabrication plant started in 1984 and is currently at 0.8 micron level of technology. The plan to spin it off into a society is almost two years old and comes at a time when private sector chip manufacturing giants are being wooed by the Indian government to come and set up shop in India.

Muppalla · Post by **Muppalla** » 18 Jul 2009 09:00

http://www.youtube.com/watch?v=EdretTflWEw

Dileep and others - What is your take on the above video and claims?

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 09:16

http://www.bel-india.com/BELWebsite/ima ... atures.pdf

Easy transportation, set up and operation, operates on battery. Very low Mean Time Between Failure (MTBF), more than 10 years of guaranteed life cycle, simple maintenance

Since it says 10 years of guarenteed life, I give it 15 years life, when it will have to be replaced. Rs 10000, for 4 general elections and say 2 state (the states which have common elections need dual EVMs), 3 local elections. So Rs 10000 for around giving it generous voting cycles give 15 elections before it needs to be replace. In a booth normally with 1000 voters (again more than average) , avg cost of each vote in each EVM = 0.67 Re.

Is this is cheaper than paper ballots, even giving very very generous estimates, I do not think so. Atleast according to the data provided, cost is NOT a benefit of EVMs. To get cost competetiveness for EVMs, they will require a minimum average life of 25 years and anyone in computer field knows, electronic items do not usually work that long without problems.

Muppala,
no where it says SCL is manufacturing the chips. That SCL manufactures chips doesnt mean it manufactures EVM chips too.

Muppalla · Post by **Muppalla** » 18 Jul 2009 09:19

ravi_ku wrote:Muppala,
no where it says SCL is manufacturing the chips. That SCL manufactures chips doesnt mean it manufactures EVM chips too.

I am not claiming that BEL used SCL manufactured chips in EVMs. I am refuting the point that India does not manufacture chips at any scale.

Rahul Mehta · Post by **Rahul Mehta** » 18 Jul 2009 09:28

Dileep wrote:Microcode within the chip is NOT executable program. It is a digital logic block used to do a certain arithmetic or logic operation. It can't do complex operations like the executable program does. Here is what wiki says about microcode http://en.wikipedia.org/wiki/Microcode. It can't do any calculation, comparison etc. It is impossible to do what you say, ie monitor the key press and modify RAM locations from microcode. It is possible only from executable instruction sequences.

Yes. In general, there is no point in putting complex code inside microcode. But we are talking of SPECIFIC application.

Inside microcode, you can do ANYTHING you want - even matrix inversion and FFT calculation. No one would in general do it, as there is no value for putting such code in market. But if someone wants to put complex code in microcode, there is no technological barrier. Also EVM rigging is no complex. All you need to do is get activation sequence, get candidate number and add 10% to 20% votes to that candidate number and deduct same from rest. This is TRIVIAL code.

The fact that microcode can't do what you say notwithstanding, microcode programming is not like your C programming. It involves re-design and re-verification of the chip, plus re-making of the masks and running the fabrication. Do you know how much time, effort and manpower that takes? Certainly you can't do it within the normal lead time of 4 weeks for the chips!

It id difficult, but it is do-able and there is no technological barrier. And where is 4 week limit? EVMs have been around for 10-15 years.

vera_k · Post by **vera_k** » 18 Jul 2009 09:37

EVMs can be tampered

Elaborating his point, Prasad told a gathering including reporters that in such a case, the first 10 voters who cast their votes can be the basis of manipulations. The particular party will get 60 per cent votes by manipulations as per the programme, they added.

Rahul Mehta · Post by **Rahul Mehta** » 18 Jul 2009 09:38

Rahul Mehta: There is NO way to read microcode of chip. !! So if the trojan is in the microcode of the chip, you can never even dig it out.

Dileep: You can't read it, but you can verify its existence, because it shows up on the chip layout.

If two 8057 have different microcodes, the chip layout inside chip will be almost same. Further, small variations in layout can come anyway - they dont prove functional difference. And btw, when you say things like layout comparisons are do-able, pls do give us cost estimates. How many crores will such chip-layout level audits will cost? The paper ballots cost no more than Rs 35 cr to Rs 70. Here, your audits alone are costing us crores.

--------------

Also, the latest batch of EVMs do have port for connectivity so that Totalizers can read it. Even if Totalizers may not have been used, CEC did use EVMs with ports. If there is a port, one can always make a connector so that driver program in PC drives that device.

----------

Audit lovers,

The kind of hi-tech audits you guys demand, your auditors and auditing will cost us more than paper ballots. And top of that, you guys will settle for nothing less than PwC level auditors or Arther Anderson level auditors ( are they still around)? So where is the savings?

Rahul Mehta · Post by **Rahul Mehta** » 18 Jul 2009 09:50

ravi_ku wrote:http://www.bel-india.com/BELWebsite/ima ... atures.pdf

Easy transportation, set up and operation, operates on battery. Very low Mean Time Between Failure (MTBF), more than 10 years of guaranteed life cycle, simple maintenance
Since it says 10 years of guarenteed life, I give it 15 years life, when it will have to be replaced. Rs 10000, for 4 general elections and say 2 state (the states which have common elections need dual EVMs), 3 local elections. So Rs 10000 for around giving it generous voting cycles give 15 elections before it needs to be replace. In a booth normally with 1000 voters (again more than average) , avg cost of each vote in each EVM = 0.67 Re.

Is this is cheaper than paper ballots, even giving very very generous estimates, I do not think so. Atleast according to the data provided, cost is NOT a benefit of EVMs. To get cost competetiveness for EVMs, they will require a minimum average life of 25 years and anyone in computer field knows, electronic items do not usually work that long without problems.

ravi_ku,

I have been beating EVM-lovers black and blue on cost comparison. Despite repeated chest beating, these guys are so dheedh that they dont respond.

Look at today's newspaper. Take one whole sheet out i.e. 2 pages or 8 A4 sheets. One paper rim has 500 pages. How much is the RETAIL price of such rim? Rs 300 only. How much do offset printers charge for printing one newspaper page size page? Rs 70 for 1000 i.e. 7 paise per sheet as big as newspaper page. Look at your later printer paper. How muich does it cost? Rs 120 for 500 sheets A4 size only. And these all retail prices for small quantities. For 15,00,000 ballots, costs would be half or even less.

Do the maths whichever way you like. Ballots dont cost more than 50 paise per ballot or Rs 1 at most.

---------------

[Chest beating ON]

Hai koi mai kaa laal EVM-lover willing to comment on relative costs of paper vs EVM assuming EVM's 10 years

[Chest beating still ON]

Rahul Mehta · Post by **Rahul Mehta** » 18 Jul 2009 10:08

http://www.mcclatchydc.com/226/story/64711.html

Most electronic voting isn't secure, CIA expert says

WASHINGTON — The CIA, which has been monitoring foreign countries' use of electronic voting systems, has reported apparent vote-rigging schemes in Venezuela, Macedonia and Ukraine and a raft of concerns about the machines' vulnerability to tampering.

Appearing last month before a U.S. Election Assistance Commission field hearing in Orlando, Fla., a CIA cybersecurity expert suggested that Venezuelan President Hugo Chavez and his allies fixed a 2004 election recount, an assertion that could further roil U.S. relations with the Latin leader.

In a presentation that could provide disturbing lessons for the United States, where electronic voting is becoming universal, Steve Stigall summarized what he described as attempts to use computers to undermine democratic elections in developing nations. His remarks have received no news media attention until now.

Stigall told the Election Assistance Commission, a tiny agency that Congress created in 2002 to modernize U.S. voting, that computerized electoral systems can be manipulated at five stages, from altering voter registration lists to posting results.

"You heard the old adage 'follow the money,' " Stigall said, according to a transcript of his hour-long presentation that McClatchy obtained. "I follow the vote. And wherever the vote becomes an electron and touches a computer, that's an opportunity for a malicious actor potentially to . . . make bad things happen."

Stigall said that voting equipment connected to the Internet could be hacked, and machines that weren't connected could be compromised wirelessly. Eleven U.S. states have banned or limited wireless capability in voting equipment, but Stigall said that election officials didn't always know it when wireless cards were embedded in their machines.

While Stigall said that he wasn't speaking for the CIA and wouldn't address U.S. voting systems, his presentation appeared to undercut calls by some U.S. politicians to shift to Internet balloting, at least for military personnel and other American citizens living overseas. Stigall said that most Web-based ballot systems had proved to be insecure.

The commission has been criticized for giving states more than $1 billion to buy electronic equipment without first setting performance standards. Numerous computer-security experts have concluded that U.S. systems can be hacked, and allegations of tampering in Ohio, Florida and other swing states have triggered a campaign to require all voting machines to produce paper audit trails.

The CIA got interested in electronic systems a few years ago, Stigall said, after concluding that foreigners might try to hack U.S. election systems. He said he couldn't elaborate "in an open, unclassified forum," but that any concerns would be relayed to U.S. election officials.

Stigall, who's studied electronic systems in about three dozen countries, said that most countries' machines produced paper receipts that voters then dropped into boxes. However, even that doesn't prevent corruption, he said.

Turning to Venezuela, he said that Chavez controlled all of the country's voting equipment before he won a 2004 nationwide recall vote that had threatened to end his rule.

When Chavez won, Venezuelan mathematicians challenged results that showed him to be consistently strong in parts of the country where he had weak support. The mathematicians found "a very subtle algorithm" that appeared to adjust the vote in Chavez's favor, Stigall said.

Calls for a recount left Chavez facing a dilemma, because the voting machines produced paper ballots, Stigall said.

"How do you defeat the paper ballots the machines spit out?" Stigall asked. "Those numbers must agree, must they not, with the electronic voting-machine count? . . . In this case, he simply took a gamble."

Stigall said that Chavez agreed to allow 100 of 19,000 voting machines to be audited.

"It is my understanding that the computer software program that generated the random number list of voting machines that were being randomly audited, that program was provided by Chavez," Stigall said. "That's my understanding. It generated a list of computers that could be audited, and they audited those computers.

"You know. No pattern of fraud there."

A Venezuelan Embassy representative in Washington declined immediate comment.

The disclosure of Stigall's remarks comes amid recent hostile rhetoric between President Barack Obama and Chavez. On Sunday, Chavez was quoted as reacting hotly to Obama's assertion that he's been "exporting terrorism," referring to the new U.S. president as a "poor ignorant person."

Questions about Venezuela's voting equipment caused a stir in the United States long before Obama became president, because Smartmatic, a voting machine company that partnered with a firm hired by Chavez's government, owned U.S.-based Sequoia Voting Systems until 2007. Sequoia machines were in use in 16 states and the District of Columbia at the time.

Reacting to complaints that the arrangement was a national security concern, the Treasury Department's Committee on Foreign Investment in the United States launched an investigation. Smartmatic then announced in November 2007 that it had sold Sequoia to a group of investors led by Sequoia's U.S.-based management team, thus ending the inquiry.

In the former Soviet republic of Georgia, Stigall said, hackers took resurrecting the dead to "a new art form" by adding the names of people who'd died in the 18th century to computerized voter-registration lists. Macedonia was accused of "voter genocide" because the names of so many Albanians living in the country were eradicated from the computerized lists, Stigall said.

He said that elections also could be manipulated when votes were cast, when ballots were moved or transmitted to central collection points, when official results were tabulated and when the totals were posted on the Internet.

In Ukraine, Stigall said, opposition leader Viktor Yushchenko lost a 2004 presidential election runoff because supporters of Russian-backed Prime Minister Viktor Yanukovych "introduced an unauthorized computer into the Ukraine election committee national headquarters. They snuck it in.

"The implication is that these people were . . . making subtle adjustments to the vote. In other words, intercepting the votes before it goes to the official computer for tabulation."

Taped cell-phone calls of the ensuing cover-up led to nationwide protests and a second runoff, which Yushchenko won.

Election Assistance Commission officials didn't trumpet Stigall's appearance Feb. 27, and he began by saying that he didn't wish to be identified. However, the election agency had posted his name and biography on its Web site before his appearance.

Electronic voting systems have been controversial in advanced countries, too. Germany's constitutional court banned computerized machines this month on the grounds that they don't allow voters to check their choices.

Stigall said that some countries had taken novel steps that improved security.

For example, he said, Internet systems that encrypt vote results so they're unrecognizable during transmission "greatly complicates malicious corruption." Switzerland, he noted, has had success in securing Internet voting by mailing every registered citizen scratch cards that contain unique identification numbers for signing on to the Internet. Then the voters must answer personal security questions, such as naming their mothers' birthplaces.

Stigall commended Russia for transmitting vote totals over classified communication lines and inviting hackers to test its electronic voting system for vulnerabilities. He said that Russia now hoped to enable its citizens to vote via cell phones by next year.

"As Russia moves to a one-party state," he said, "they're trying to make their elections available . . . so everyone can vote for the one party. That's the irony."

After reviewing Stigall's remarks, Susannah Goodman, the director of election reform for the citizens' lobby Common Cause, said they showed that "we can no longer ignore the fact that all of these risks are present right here at home . . . and must secure our election system by requiring every voter to have his or her vote recorded on a paper ballot."

===================================

And here are comments from readers from an IE article

http://www.expressindia.com/latest-news ... ut/475997/

RECOUNTING CANNOT ALTER THE VOTING IF EVMS ARE USED by n.krishna on 14 Jun 2009

The control units have protocol to electronically transmit results back EC. EVM data of previous elections, gives the number of votes, each party got from each booth. Through the controlling units, remotely using sophisticated codes to its transreceiver embedded in the controlling units CIA changed the results in favour of congress. For 2009 Election at the Sivaganga loksabha constituency in Tamil Nadu, P. Chidambaram was the congress candidate, M. G. Devar of BSP, Rajakannappan of AIADMK, Shakthivel of DMDK, Ramaswamy of PT and some 14 other individual candidates. The day of counting, 16th May, there was a heavy drama. The Congress Candidate P. Chidambaram was declared defeated by his rival Raja Kannappan of AIADMK by 3000 votes. P. Chidambaram who is in the knowhow of the CIA’s role in EVM, asked for a recounting.

Then P. Chidambaram was declared elected by a marginal vote of 3,354. The total number of votes P. Chidambaram got was 3,34,348 and RajaKannappan got was 3.30,994.

-----

EVM chip comes from Japan by n.krishna on 14 Jun 2009

In 2004, the stupid BJP had adopted Electronic Voting Machines (EVM) for its elections to the Parliament with 380 million voters had cast their ballots using more than a million voting machines. The Indian EVMs are designed and developed by Bharat Electronics Limited (BEL) and Electronics Corporation of India Limited (ECIL). The System is a set of two devices running on 6V batteries. One device, the Voting Unit is used by the Voter, and another device called the Control Unit is operated by the Electoral Officer. Both units are connected by a 5 meter cable. The Control Units has Three buttons on the surface, one button to release a single vote, one button to see the total number of vote cast till then and one button to close the election process. The result button is hidden and sealed, It cannot be pressed unless the Close button is already pressed. Unfortunately the microchip used in EVMs is manufactured in Japan and CIA programme in it cannot be detected.

----------------------------------

IOW, there is one more wise men on earth who thinks that microchip had a trojan implanted by CIA

As I say, AWMTA

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 10:12

RM,
Do you remember who (basically at what level) was the tender for printing ballot papers used to be given?

Rahul Mehta · Post by **Rahul Mehta** » 18 Jul 2009 10:37

ravi_ku wrote:RM,
Do you remember who (basically at what level) was the tender for printing ballot papers used to be given?

When paper ballots were used,

1. The tender to issue paper ballots is issued by District Collector BEFORE election starts.

2. The the final price is set as (X * size of ballot paper). This agreement is done days before election starts.

So size of paper becomes known about 15 days before poll, when candidate list becomes final. And the printing company has some 7 days to finish printing. This was 15 years ago. Now offset printing technology has become so powerful and miniature that offset press can deliver lakhs of copies within 1-2 days.

And given that technology is so common place, competition is high and profits are thin. Even with corruption, a ballot with 20 candidates will not cost more than Rs 1 per ballot or Rs 70 cr.

Now transportation costs also apply on EVMs. And in case of EVM, it is more expensive as it is delicate. The paper punch can be thrown across the room and there will be no damage. So transporting paper would cost less.

--------

EVM-lovers,,

And here is one more proof to show that EVMs were replaced by CIA. If elections actually were on made in India EVMs, move than 20% would have shown faults. Where on polling days, less than 0.01% EVMs had failed. Such low failure rate shows that EVMs were made outside India

. Hence someone had replaced them

(I am just kidding, no offense against BEL people).

And there is one more reason to use paper. Paper is STRONGER !! eg I can throw ballot papers bunch from 3rd floor and yet they will be intact and usable !! While try throwing EVMs from even 1st floor and it would become useless. Indian conditions require STRONG media which is shock and so we must use paper

Or other, how about we solve the debate by throwing 100 bunches of ballot paper and 100 EVMs from 3rd floor, and see which is still usable?

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 10:41

Rahul Mehta wrote:
ravi_ku wrote:RM,
Do you remember who (basically at what level) was the tender for printing ballot papers used to be given?
When paper ballots were used,

1. The tender to issue paper ballots is issued by District Collector BEFORE election starts.

2. The the final price is set as (X * size of ballot paper). This agreement is done days before election starts.

So size of paper becomes known about 15 days before poll, when candidate list becomes final. And the printing company has some 7 days to finish printing. This was 15 years ago. Now offset printing technology has become so powerful and miniature that offset press can deliver lakhs of copies within 1-2 days.

And given that technology is so common place, competition is high and profits are thin. Even with corruption, a ballot with 20 candidates will not cost more than Rs 1 per ballot or Rs 70 cr.

Now transportation costs also apply on EVMs. And in case of EVM, it is more expensive as it is delicate. The paper punch can be thrown across the room and there will be no damage. So transporting paper would cost less.

Thanks, basically all I wanted to understand whether EVMs can also be interpreted as centralisation of corruption. Based on what you said, it definitely can be understood in that fashion.

krishnan · Post by **krishnan** » 18 Jul 2009 10:52

Well, you can tear a ballot paper, you have to safe guard it from rain , you need to print new set for re polling or for next election(you could go for recycling though). You need 1000's of ballot paper , when a single EVM is enough.

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 10:57

krishnan wrote:Well, you can tear a ballot paper, you have to safe guard it from rain , you need to print new set for re polling or for next election(you could go for recycling though). You need 1000's of ballot paper , when a single EVM is enough.

possibly you missed it when I tried to calculate, cost per vote. No, for reelection, even today EC uses spare EVMs. The original EVMs are not reused, just in case for the court cases.

That single EVM cost Rs10000, whereas printing ballots for 1000 voters costs at max in super duper vinyl paper printing 1 Re max.

Try to calculate cost per vote. I have tried to calculate them above. If you have problems with any the calculations, recalculate and please show where I may be mistaken. Otherwise your blanket statement, without any iota of proof is just another form of hot air.

Hint: The standard operating procedure for voter destruction today is simple

, pour water into control unit/ EVM. There goes your rain theory because EVMs are carried in special briefcases. For ballet paper , all you need is a simple plastic cover.

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 11:22

krishnan wrote:Ok , regarding cost , lets take a polling booth with around 25,000 voters, so we need to print 25,000 ballot papers, so taking your 1 rupee per ballot paper, it comes to around 25,000 rupees. Lets even take it as 20K. Thats still more than 10,000 per EVM.

Krishnan,
I cant talk to you without you knowing even basics of Indian elections. The avg number of people in a booth 600-700, though there exists booths with around 1500 votes, but they are very less.

and who told you that EVMs can have 25000 votes, no an EVM can have a max of 3840 votes. though if I remember correctly it was more like 1500 votes.

http://en.wikipedia.org/wiki/Indian_voting_machines

Dileep · Post by **Dileep** » 18 Jul 2009 11:33

Muppalla wrote:http://www.youtube.com/watch?v=EdretTflWEw

Dileep and others - What is your take on the above video and claims?

The title is WRONG. Nothing is PROVEN. Only scenarios are made, and all of them refuted by the report posted by KV Rao.

Dileep · Post by **Dileep** » 18 Jul 2009 11:55

Muppalla wrote: Why (2) is far fetched? Between all possibilites I see (2) along with hacked randomizer is a probable one. I do not believe in external vendors. At least there should be open verification and all the names of vendors involved everything related EVMs should be published. Transparancy is key to get over these conspiracy theories.

Let us assume the following:

1. BEL did the chip design, and sent the CAM files to FUJIMOTO fab in Osaka to fab the chips. The design itself is kosher.
2. The moles in BEL chip design team takes the complete source of the chip design and sells to CIA.
3. The mole in BEL circuit design steals the complete CAD files and sells to CIA
4. The mole in BEL software team steals the complete source code of the software and sells to CIA.
5. The mole in BEL logistics team tells CIA that FUJIMOTO is doing the fab.

Now, CIA need to give all these to their crack team of IC designers to design a "microcode", which essentially is going to take half as much silicon area as the original chip. They go through the cycle of design, verification, layout etc for the combined circuit. It takes six months at least. The design and verification of the "microcode" is in fact going to take much more time, because you are dealing directly with the registers interconnect here.

Meanwhile FUJIMOTO have to provide the chips to BEL, so they run the fab with the old design and deliver the prototype chips to BEL.

CIA tells FUJIMOTO to use their CAM files instead of the BEL files, which result in a 150% bigger chip, and 150% more power consumption. The processing power is slower, owing to the cycles used by the "microcode".

Any self respecting chip designer will run test on each lot of the chips delivered. So, BEL is going to find out the increased power consulption, and reduced performance. They do FMA, and find that the wrong chip is used.

Rest will be history.

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 12:02

Dileep wrote:
Muppalla wrote: Why (2) is far fetched? Between all possibilites I see (2) along with hacked randomizer is a probable one. I do not believe in external vendors. At least there should be open verification and all the names of vendors involved everything related EVMs should be published. Transparancy is key to get over these conspiracy theories.
Let us assume the following:
........................

Rest will be history.

or they take their own sweet time of 5 years and make similar EVMs and replace them during non election cycle storage

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 12:07

Dileep wrote:
Muppalla wrote:http://www.youtube.com/watch?v=EdretTflWEw

Dileep and others - What is your take on the above video and claims?
The title is WRONG. Nothing is PROVEN. Only scenarios are made, and all of them refuted by the report posted by KV Rao.

So you want me to believe some non-descript bureocrat KV Rao as incorruptible and all-knowing

, in a country where the National Security Advisor himself migrates immediately to a different country immediately after retirement

Dileep · Post by **Dileep** » 18 Jul 2009 12:09

Rahul Mehta wrote: Yes. In general, there is no point in putting complex code inside microcode. But we are talking of SPECIFIC application.

Inside microcode, you can do ANYTHING you want - even matrix inversion and FFT calculation. No one would in general do it, as there is no value for putting such code in market. But if someone wants to put complex code in microcode, there is no technological barrier. Also EVM rigging is no complex. All you need to do is get activation sequence, get candidate number and add 10% to 20% votes to that candidate number and deduct same from rest. This is TRIVIAL code.

The FFT and matrix inversion are "self contained" modules. The executable program feeds the data into the FFT module, and activates the module. It does the calculation and places the result in its own registers. Accessing external memory and IO is not possible just by microcode. You need to add busses to carry the data to and from the microcode logic. Essentially, your microcode will become a little micro-microprocessor.

It is not TRIVIAL code. It is extremely complex code if you want to implement in microcode. IT is going to take a lot of chip area also.

And don't forget that the microcode can't "monitor" the register for key pattern, because the registers are used for all instructions.

What about the additional chip area and power consumption? And what about the performance degradation?

Dileep · Post by **Dileep** » 18 Jul 2009 12:11

ravi_ku wrote: or they take their own sweet time of 5 years and make similar EVMs and replace them during non election cycle storage

They sure can. RM still have that scenario active, even though it is torn into pieces already. Read up the previous threads please.

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 12:12

Dileep wrote:
ravi_ku wrote: or they take their own sweet time of 5 years and make similar EVMs and replace them during non election cycle storage
They sure can. RM still have that scenario active, even though it is torn into pieces already. Read up the previous threads please.

can you please point where post?? or a gist of how it is shred.

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 12:17

Dileep wrote:
Rahul Mehta wrote: Yes. In general, there is no point in putting complex code inside microcode. But we are talking of SPECIFIC application.

Inside microcode, you can do ANYTHING you want - even matrix inversion and FFT calculation. No one would in general do it, as there is no value for putting such code in market. But if someone wants to put complex code in microcode, there is no technological barrier. Also EVM rigging is no complex. All you need to do is get activation sequence, get candidate number and add 10% to 20% votes to that candidate number and deduct same from rest. This is TRIVIAL code.
The FFT and matrix inversion are "self contained" modules. The executable program feeds the data into the FFT module, and activates the module. It does the calculation and places the result in its own registers. Accessing external memory and IO is not possible just by microcode. You need to add busses to carry the data to and from the microcode logic. Essentially, your microcode will become a little micro-microprocessor.

It is not TRIVIAL code. It is extremely complex code if you want to implement in microcode. IT is going to take a lot of chip area also.

And don't forget that the microcode can't "monitor" the register for key pattern, because the registers are used for all instructions.

What about the additional chip area and power consumption? And what about the performance degradation?

All I see is it is not trivial. Ok, it is not trivial. So what? you dont seem to say it is impossible, I wonder why? For what how much power and memory exactly, say a press of 7 9 2 X, bytes exactly takes?

AjayKK · Post by **AjayKK** » 18 Jul 2009 12:17

For the sake of discussion, excerpts from an article in Tribune during 2001 polls by Capt. Amarinder Singh (ex Congress CM Punjab).

Chandigarh, March 11, 2001

Can electronic voting machines (EVMs) be tampered with?

“Yes”, says Mr Amarinder Singh, president, Punjab Pradesh Congress Committee, supporting his assertion by giving a demonstration of how an EVM with a cleverly programmed chip installed in it can transfer votes polled by one candidate to another leaving no remnants of the original voting pattern.

Mr Amarinder Singh carries a set of EVMs, including the control unit, which during elections remains with the presiding officer of a polling station, and gives a “demonstration of how the programmed chip transfers the votes of one candidate to another”.

“We got suspicious about what we call ‘sophisticated booth capturing’ when we found that there was 129 per cent increase in the votebank of Akalis at Nawanshahr, 100 per cent at Sunam and now 65 per cent at Majitha. The ruling party did well wherever EVMs were used while at other places, we did well.

“But once they put their electronics experts on the job, they could immediately find a solution. Whatever the Election Commission says about EVMs is not true. The mother boards, after being removed from the EVMs, do not crash but work perfectly after being soldered back in the machine. Similarly, wave welding, which the Election Commission maintains is not available in India, is very much available at various places in the country,” asserts the Punjab Congress chief.

“We put our hardware and software experts on the job.

A programmed chip will not cost much. It is both timed and programmed to convert the votes polled by one candidate to those of another. It is only the final position that will remain on the hardchip or all three memories, thus leaving no scope for anyone to find out the original pattern of voting,” he says during the demonstration. “Seventeen votes are cast of which three go to candidate number 1, one each to candidates number two and three, 11 to candidate number 5 and one to candidate number 7. And after a while, when the votes are counted, the machine gives 13 votes to candidate number 1 and four to candidate number 2 and nothing to the rest.

“So each machine can be programmed to transfer, say, every third vote polled by the Congress to the Shiromani Akali Dal.

“The EVMs remain in the custody of the government, thus leaving scope for their manipulation. We had requested the Election Commission that if it wants to use EVMs in Majitha, let it bring EVMs from any other state and use them. But our suggestion was turned down and the EVMs already with the election tehsildars in Punjab were used,” he added.

http://www.tribuneindia.com/2001/20010312/main4.htm

Dileep · Post by **Dileep** » 18 Jul 2009 12:18

ravi_ku wrote: So you want me to believe some non-descript bureocrat KV Rao as incorruptible and all-knowing , in a country where the National Security Advisor himself migrates immediately to a different country immediately after retirement

Read before you shoot your mouth off. I am talking about the report made by Prof Indiresan comittee, and I refer to it for the refutals they made, so that I don't have to repeat them here.

Some of the claims in the TV report is laughable. They say wireless access can be used. Where is the wireless receiver i the EVM? They say serial port can be used. There is no serial port on the EVM. They say a piece of code can be loaded from outside. There is no outside access, so, how will you load the code?

Everything raised in the video is already torn to pieces. The only standing arguments right now is the "chip modification at the fab" and "replace the EVM at the warehouse". The only reason the latter is standing is because RM is propping it up ignoring all the evidence.

Dileep · Post by **Dileep** » 18 Jul 2009 12:19

ravi_ku wrote: can you please point where post?? or a gist of how it is shred.

It is on page 1. That is required reading for discussion here, because it answers a lot of questions.

Dileep · Post by **Dileep** » 18 Jul 2009 12:24

ravi_ku wrote: All I see is it is not trivial. Ok, it is not trivial. So what? you dont seem to say it is impossible, I wonder why? For what how much power and memory exactly, say a press of 7 9 2 X, bytes exactly takes?

Is is "possible" to land a man on mars? It sure is.

I am talking about the additional power consumption that the microcode unit takes. Whether active or not, all the circuitry need to be powered all the time, and the microcode executes all the time, monitoring key-presses.

Added: Read the scenario again. The power consumption will be caught as a defect in the chip. It is standard procedure to verify the power consumption of samples from every lot of custom ICs. It is done because it gives a direct verification of the correct silicon process, EMC damages etc. We have in fact rejected chip lots based on power consumption issues, which were subjected to severe FEMA and traced to handling problem of the chips.

It is also standard to run benchmarks for processors through JTAG.

Also, the presence of the extra microcode, and its interconnect will be readily visible, and the chip size itself will be bigger. Readily verifiable by anyone who cares.

Dileep · Post by **Dileep** » 18 Jul 2009 12:30

Rahul Mehta wrote:So say 50% of EVMs are stored in district warehouses and 50% in CEC warehouses.

You say don't make it so.

I asked you to PROVE THAT. Ask the Gandhinagar District collector and find out how many units were under his custody, and how many under the CEC custody. Ask, or file a PIL.

Till you prove that your argument of central storage of EVMS don't stand.

Then there is the issue of setting the old machine with the correct counts. Until you provide a solution for that your "CIA Replaced the machines" argument doesn't stand. The one you gave, ie connect a PC won't work. Connecting a robot to punch buttons won't work either, because the robot will be big, and it still takes at least 20 seconds to cast one vote.

Virupaksha · Post by **Virupaksha** » 18 Jul 2009 12:32

Dileep wrote:
ravi_ku wrote: So you want me to believe some non-descript bureocrat KV Rao as incorruptible and all-knowing , in a country where the National Security Advisor himself migrates immediately to a different country immediately after retirement
Read before you shoot your mouth off. I am talking about the report made by Prof Indiresan comittee, and I refer to it for the refutals they made, so that I don't have to repeat them here.

ok, not kv rao, but some xyz who goes by the name Indersen or Ravan or Inderjit. Who cares for the name of that bureacrat. So now my previous statement looks

you want me to believe some non-descript bureocrat Indersen as incorruptible and all-knowing

What difference did it make to the import of my statement? Do you want me believe him as all-knowing and incorruptible? If you want to believe that Santa Claus exists and pigs fly, who am I to say no?

I have read completely the report. You should read carefully. It bases the absence of trojan horses on the basis that every key pressed is recorded, as though what can be written by such a dumb machine cannot be erased.

shiv · Post by **shiv** » 18 Jul 2009 12:35

pgbhat wrote:In all this debate about fraudulent EVMs .... can anybody give the expected number of people that have to be involved in perpetrating this fraud?

Oh one or two..and the CEC is one. Now to nail the other guy

Pranav wrote: Manufacturers are PSUs under government control. Not hard to appoint the right person in the critical position. You don't need to have a large number of people involved - just one or two people would be enough. And the CEC himself is known to have a dubious background. Maintenance contracts given to Congress companies. So it would be rather naive to trust the system to police itself.