Should we discontinue EVMs?

Pranav · Post by **Pranav** » 22 Jul 2009 18:31

Rahul Mehta wrote:
One cant put video camera in booth as of now. Too expensive. May be 5 years down the line, it is possible, but not today. But one can have camera which takes pix and sends every 30-60 seconds or so.

Cams dont have to be very high resolution. Need to be just good enough to recognize faces. Can get webcams for about Rs 1500, I think. If you order in bulk, price would be pretty low. Would need a hard disk to store the video, and that would add some cost.

Dileep · Post by **Dileep** » 22 Jul 2009 18:32

"They are coming to get me" is the fundamental note to RM's psyche. Let is be CIA, NBJPRE, or BRF members. He lives on scaremongering.

He would do good as a Paki politicians, downhill skiiing included.

Tanaji · Post by **Tanaji** » 22 Jul 2009 18:38

The topic at hand is how vulnerable EVMs are large scale attack from top.

No the topic is about how feasible it is to attack a EVM. What you fail to understand or are deliberately pretending to is the difference between possibility and feasibility. It is theoretically possible to do what you are saying, but practically not possible.

All cryptographic algorithms are theoretically broken by brute force, but practically not possible with current or future hardware. You for some reason want to discount the feasiblity part of it.

And I proposed camera-in-booth (and so did Pranav, another papervaadi). This makes booth level rigging impossible in both cases.

Why would you believe the camera? If you dont believe the district collector, why would you believe the person in charge of the camera? He can show what he wants or run a tape of what he wants.

Pranav · Post by **Pranav** » 22 Jul 2009 18:46

Tanaji wrote: Why would you believe the camera? If you dont believe the district collector, why would you believe the person in charge of the camera? He can show what he wants or run a tape of what he wants.

Very difficult to fake a video of people voting one by one, especially if video shows recognizable faces. It was already tried on a limited scale in May, and it can be expanded.

Tanaji · Post by **Tanaji** » 22 Jul 2009 18:50

Pranav wrote:
Tanaji wrote: Why would you believe the camera? If you dont believe the district collector, why would you believe the person in charge of the camera? He can show what he wants or run a tape of what he wants.
Very difficult to fake a video of people voting one by one, especially if video shows recognizable faces. It was already tried on a limited scale in May, and it can be expanded.

Yes, I agree. What I am pointing out is the same feasibility versus possibility issue to Rahul Mehta. Just as he claims we can solder desolder replace BU/CU/Cables, bribe DCs etc etc. it is also possible that video cameras can be subverted. Doesn't mean its feasible.

Rahul Mehta · Post by **Rahul Mehta** » 22 Jul 2009 18:58

Tanaji wrote:All cryptographic algorithms are theoretically broken by brute force, but practically not possible with current or future hardware. You for some reason want to discount the feasiblity part of it.

If key size is 128 bytes, then encryption is not breakable even if you put all PCs in world for 1000 years. Here CIA can make some 100000 BUs with some 1000-2000 offshore engineers and have them replaced with some 10-20 people at top in CEC, PMO, HomeMin and some mere 100 agents. So this falls within "do-able" limits of CIA.

Why would you believe the camera? If you dont believe the district collector, why would you believe the person in charge of the camera? He can show what he wants or run a tape of what he wants.

The camera send pix to server that can be seen by anyone over internet for a small fee to cover costs. So CIA cannot rig all the online streams as there is no technology to fake 700,000 streams or even 1000 streams. Mind you, CIA will not bother to rig 100-200 booths. If they can rig 20000 booths, they are in, otherwise they wont bother. And no local guy has means to rig even 5 streams.

And to rig 10% booths, one would need to bribe 10% of presiding officers i.e. 70000 of them. Thats cant be done by anyone. So industrial scale rigging of camera is ruled out.

Rahul Mehta · Post by **Rahul Mehta** » 22 Jul 2009 19:03

Pranav,

In trojan theory, one issue was : how will trojan know the candidate number and when should rigging start.

IMO you should look at RF technology. That may give some answers. The RF chips are now 2mm * 2mm and need no power to exchange 100-200 bytes. Can such an RF chip be integrated with micro-controller made in Japan? Or placed at some point in PCB?

I cant tell now.

Rahul Mehta · Post by **Rahul Mehta** » 22 Jul 2009 19:11

Tanaji wrote:
Pranav wrote:Very difficult to fake a video of people voting one by one, especially if video shows recognizable faces. It was already tried on a limited scale in May, and it can be expanded.
Yes, I agree. What I am pointing out is the same feasibility versus possibility issue to Rahul Mehta. Just as he claims we can solder desolder replace BU/CU/Cables, bribe DCs etc etc. it is also possible that video cameras can be subverted. Doesn't mean its feasible.

No. Now you are just blabbering. The levels are NOT same. To send fake data from 100000 booths , you need to bribe 1-2 persons per booth or 150,000 in total. Whereas BU replacements in CEC warehouse is possible by bribing 10-20 people at top and 2-3 guards at warehouse. And there is no soldering etc in BU replacement theory. And after counting, 400 DCs need to be bribed to look aside, not take risks.

The number of people involved in "replace 100,000 BUs" theory are too small compared to your "fake 100,000 camera" theory. So your use of word "same" is misleading.

Tanaji · Post by **Tanaji** » 22 Jul 2009 19:12

If key size is 128 bytes, then encryption is not breakable even if you put all PCs in world for 1000 years

I keep telling you, if you dont understand a technical issue dont use it in an attempt to make yourself sound credible. You end up making a fool of yourself.

Key strength is dependant on the algorithm used and the entropy of the key. A 128 bit key for a asymmetric algorithm is very less, the recommended is 1024 at least. For a symmetric algorithm, 3DES which is on its way out uses 160 bit keys. So, your statement is provably wrong on many counts.

The camera send pix to server that can be seen by anyone over internet for a small fee to cover costs. So CIA cannot rig all the online streams as there is no technology to fake 700,000 streams or even 1000 streams.

Again a fundamental misunderstanding. All you need to change is the output of the camera. You dont need to change 1000 streams. Changing the output of the CCD will result in all streams being changed.

Please,. I beg of you, stop using wrong technical concepts.

Tanaji · Post by **Tanaji** » 22 Jul 2009 19:14

Rahul Mehta wrote:Pranav,

In trojan theory, one issue was : how will trojan know the candidate number and when should rigging start.

IMO you should look at RF technology. That may give some answers. The RF chips are now 2mm * 2mm and need no power to exchange 100-200 bytes. Can such an RF chip be integrated with micro-controller made in Japan? Or placed at some point in PCB?

I cant tell now.

So "RF technology" is the jargon for today?

At least use correct jargon. What you want is RFID not RF technology.

Pathetic.

Pranav · Post by **Pranav** » 22 Jul 2009 19:23

Rahul Mehta wrote:Pranav,

In trojan theory, one issue was : how will trojan know the candidate number and when should rigging start.

IMO you should look at RF technology. That may give some answers. The RF chips are now 2mm * 2mm and need no power to exchange 100-200 bytes. Can such an RF chip be integrated with micro-controller made in Japan? Or placed at some point in PCB?

I cant tell now.

See http://forums.bharat-rakshak.com/viewto ... 22#p702422 .

Yes, RF with MEMS antennas is within the realm of possibility. It is good to operate on the assumption that integrity of the system will be under continuous attack by the government and others with the means to do so. The goal is to make it as difficult as possible.

Rahul Mehta · Post by **Rahul Mehta** » 22 Jul 2009 19:31

Pranav wrote:Cams dont have to be very high resolution. Need to be just good enough to recognize faces. Can get webcams for about Rs 1500, I think. If you order in bulk, price would be pretty low. Would need a hard disk to store the video, and that would add some cost.

We need cam with mobile connectivity so that pix can be sent to servers directly. And my cost "limit" is Rs 3000 per booth including batteries. Nothing hard and fast. If video can be accomplished, I have no issue.

Basically, the device would need cam, mobile connectivity, memory and should run on a small battery for 10 hours. The battery is what becomes issue in such cases. To keep battery small, you cant have too much functionality like constant video recording. So I settled for 1 pix every 60 seconds. The mobile connectivity is useful and needed. Using that, the cam can upload the pix every minute and citizen can see the booths he wants online. This gives him feeling that nothing wrong is going on, and the urban myth of booth capture can be killed ASAP. And that will also kill the argument that "EVMs are the only way to stop booth capturing" and make way for paper ballots.

Rahul Mehta · Post by **Rahul Mehta** » 22 Jul 2009 19:39

Tanaji wrote:A 128 bit key for a asymmetric algorithm is very less, the recommended is 1024 at least. For a symmetric algorithm, 3DES which is on its way out uses 160 bit keys. So, your statement is provably wrong on many counts.

Pls read my post. I said 128 BYTES, not bits. And strength depends on both, algorithm used and key size. With small key size, it becomes easier to break. And time to break increases almost exponentially with increase in keysize. So large size give a security that even if ALL PCs of world are put to work, they cannot break the code in 1000 years.

---

Again a fundamental misunderstanding. All you need to change is the output of the camera. You dont need to change 1000 streams. Changing the output of the CCD will result in all streams being changed.

To change output of 100,000 CCDs, you need to bribe 100,000 presiding officers. Not even 1% as easy as bribing 20 top guys in India and getting 100,000 BUs replaced in CEC warehouse. So replacing 100,000 BU is easier than changing 100,000 CCD outputs --- by several orders of magnitude.

Raja Bose · Post by **Raja Bose** » 22 Jul 2009 19:43

Rahul Mehta wrote:If a small RF unit has to send/receive a few bytes to/from a big RF unit a few meters away, the energy small one would need is negligiblel. (eg RFID circuits do radio communication without any battery).

Again RMji please educate yourself before spouting techno-babble. Communication with passive RFIDs from a few meters?!

BTW even with an active tag, do you know how big the RF transceiver unit is - certainly not something your CIA agent can carry around innocuously.

suryag · Post by **suryag** » 22 Jul 2009 19:53

RMji here i would like to bring up some points
1. CIA can generate graphics based voting queues and replace the videos. May i suggest that a voter come in and after voting look at the camera and make a contortion.
2. CIA can rig RFID circuits too, you already proved they can change Microcode so playing around with a passive rfid circuits is child's play for them
3. Servers in which videos are going to be stored can be hacked in, again we cant have 700000 people to watch over the live transmission

Rahul Mehta · Post by **Rahul Mehta** » 22 Jul 2009 19:55

Pranav,

Pranav wrote: No exact pseudocode, but a possible approach that was mentioned was:
1. special key combination to transfer control to trojan, for example by pressing 1 and 7 simultaneously.
2. press key x
3. a simple version would be to allot third vote thereafter to key x.

Instead of keys, can one activate trojan by RF?

We know that RFID chips can send/receive a small number of bytes upto few meters with no power source.

Can CU chip (that came from Japan) have such send/receive functionality?

If ones knows the OTP ROM code, he can know which RAM locations stores count of Row1 to Row64.

So if CU chip has RF capability, and a microcode instruction to subtract N from RAM location X and add N to RAM location Y, then one can change counts after poll ends and before counting starts !! The memory locations X and Y can be sent over RF.

See if you can find more information on how far RF tech has moved, and see if such chips can be made.

Pranav · Post by **Pranav** » 22 Jul 2009 20:03

Rahul Mehta wrote: Instead of keys, can one activate trojan by RF?

I'm sure that would not be impossible.

Rahul Mehta · Post by **Rahul Mehta** » 22 Jul 2009 20:04

Raja Bose wrote:Again RMji please educate yourself before spouting techno-babble. Communication with passive RFIDs from a few meters?! BTW even with an active tag, do you know how big the RF transceiver unit is - certainly not something your CIA agent can carry around innocuously.

The CU has battery in it. So RF inside EVM (if there is RF) is not passive.

I am giving RFID as example of RF. As such "ID" has nothing to do with EVM rigging we are discussing.

(Aside : https://acc.dau.mil/CommunityBrowser.aspx?id=143395

It says that passive RFIDs can be read from 30 fts. But that is not relevant for us.)

Tanaji · Post by **Tanaji** » 22 Jul 2009 20:11

Can you be more specific on where the RF interface is , BU or CU , how one activates etc?

Rahul Mehta · Post by **Rahul Mehta** » 22 Jul 2009 20:44

Pranav wrote:I'm sure that would not be impossible.

I think so too. But we must give a compete operation manual type write up.

--------

Tanaji,

In the BU replacement theory, RF is in BU. Thru RF, a field agent from a distance of few meters communicates candidate number etc. The RF unit here can be 1 cm * 1cm or even bit bigger. As as long as BU does not noticeably heavier it would be fine. And if one opens BU, it should not be drastically different from real one.

In "Trojan on chip" theory, the RF is in CU and in Japanese chip. And it has to be real miniature. So what is the smallest RF that one can put on micro-controller chip that can take data from say 10 meters away?

--------

So if the Japanese chip has say 2mm * 2mm RF unit that can send-receive say 10 bytes from a device 20 meters away, then trojan activation is possible. What makes me think that such RF unit can be placed is from the news I read on RFID. Handheld RFID readers can communicate with tiny RF chips (I dont know distance yet).

Sanku · Post by **Sanku** » 22 Jul 2009 22:32

Pranav wrote:
Tanaji wrote: Why would you believe the camera? If you dont believe the district collector, why would you believe the person in charge of the camera? He can show what he wants or run a tape of what he wants.
Very difficult to fake a video of people voting one by one, especially if video shows recognizable faces. It was already tried on a limited scale in May, and it can be expanded.

Technically very very easy. Ask any one who has written one of the modern video games.

Rahul Mehta · Post by **Rahul Mehta** » 22 Jul 2009 22:41

Sanku wrote:
Pranav wrote: Very difficult to fake a video of people voting one by one, especially if video shows recognizable faces. It was already tried on a limited scale in May, and it can be expanded.
Technically very very easy. Ask any one who has written one of the modern video games.

Sanku,

There are 700,000 booths. If each booth has 1 camera which uploads one pix every minutes and puts on District Collector's servers. And the pix can be seen by anyone (candidates, media, public) for a fee to cover cost of servers and bandwidth.

Now are you claiming that someone in world can fake even 2% of this data on real time basis?

Pls explain the whole setup.

Sanku · Post by **Sanku** » 22 Jul 2009 23:18

Rahul Mehta wrote:
Sanku wrote:
Technically very very easy. Ask any one who has written one of the modern video games.
Sanku,

There are 700,000 booths. If each booth has 1 camera which uploads one pix every minutes and puts on District Collector's servers. And the pix can be seen by anyone (candidates, media, public) for a fee to cover cost of servers and bandwidth.

Now are you claiming that someone in world can fake even 2% of this data on real time basis?

Pls explain the whole setup.

Cant, cant explain Physics to some one who doesnt even knows the laws of motion.

Meanwhile please use google to find how crowds are rendered in CG pictures etc.

--------

Where did real time get into the picture this is not real time. Real time you dont need cameras. There are eyeballs.

Virupaksha · Post by **Virupaksha** » 23 Jul 2009 00:05

Sanku wrote:
ravi_ku wrote:That is why I am trying to learn what exactly is stopping these people from trying to rig for themselves in the case of EVMs as some people are mentioning that EVMs CANNOT be rigged by thugs and so on.
Well the EVMs just cant work on a "dump in a lot of votes in a short span of time" mode. A thug will spend the better part of the day doing that. Even if you get PO in picture.

And you will need lots of thugs, one per machine, in the manual system one thug can do the trick, a lot of preparation can be done before the polling day.

A thug will not be able to press the button very fast. So the gap of 8 sec per vote is the only real deterrence. Is this what you are saying?

So all a system needs to do is make sure that one can vote only with a minimum gap of 8 sec, then the protection level of thugs will be the same for both the systems.

Sanku · Post by **Sanku** » 23 Jul 2009 00:12

ravi_ku wrote:So all a system needs to do is make sure that one can vote only with a minimum gap of 8 sec, then the protection level of thugs will be the same for both the systems.

Well not completely, thugs can do homework on ballot papers before (related to time but not only time also ease), also the comfort factor of a thug w.r.t. to electronic system as opposed to a manual paper (a piskological issue) etc.

A Mupalla said he has posted a "why no paper ballot before" and it had more points, I will let him take the stage.

Virupaksha · Post by **Virupaksha** » 23 Jul 2009 01:09

Sanku wrote:
ravi_ku wrote:So all a system needs to do is make sure that one can vote only with a minimum gap of 8 sec, then the protection level of thugs will be the same for both the systems.
Well not completely, thugs can do homework on ballot papers before (related to time but not only time also ease), also the comfort factor of a thug w.r.t. to electronic system as opposed to a manual paper (a piskological issue) etc.

A Mupalla said he has posted a "why no paper ballot before" and it had more points, I will let him take the stage.

Sanku,

let me see what you are saying

i) its a piskological thing, means it was new in 2004. Today it is not new and I dont give credence to this piskological factor anymore.
ii) I think you have read what Rahul Mehta has said about ballots. All ballots have to be signed by returning officer. So what can be pre prepared?

Muppalla · Post by **Muppalla** » 23 Jul 2009 06:11

Dileep wrote:
Muppalla wrote: Dileep,
A request - Could you write a summary of what is possible to subvert EVMs technically (even if it is a long shot) that does not need too many people from EC, districts etc. Anything that involves more people from various locations is not even a long shot and in India it will get on to the newspapers. Folks like me lost the track becasue of you know what is going on here.
Thanks
If I write, that will be about what is NOT possible, because I am yet to see a vulnerability that could be exploited without compromising a large number of people.

Why not in the fabs outside India? You may have answered this question but I might not have understood it. What if there are certain lots among numerous chips that are manufactured in Japan come with preloaded binaries or with trojans. Even when there are significant differences in power consumption at BEL(is that what you said?), as these are only few lots, isn't it possible to ignore these in an ocean of chips? Don't you think this does not need the entire BEL/ECIL to be corrupt?

Stamping, sealing and quality checks are manageable when doing the bulk production.

What's your take on such a scenario. If the external forces wanted to subvert our electoral process ( CIA

), all they need to do is 5% of booths in about 75-100 seats out of 543 seats.

Muppalla · Post by **Muppalla** » 23 Jul 2009 06:38

Sanku wrote:A Mupalla said he has posted a "why no paper ballot before" and it had more points, I will let him take the stage.

I am actually scared to get into that longer and lengthier replies

. Hence I kept away from that topic. Here are few things I will write:

(1)The party with muscle power can easily manipulate the ballot based system as compared to EVMs. EVMs will not work if some one tried to sabotage the booth. Laloo's victories can be atributed to ballot boxes only.
(2) To carry the ballot boxes you need trucks not suitcase carrying jeeps. Securing them is challenge. Now imagine the manufacturing of the boxes and carrying them to booths and then securing them. You need good trains.
(3) Even if there are 64 candidates in a constituency, you can imagine the size of the paper and the trouble to fold them and stuff them. There is no randomization of the symbol's slot on the ballot paper. Rajmohan Gandhi knows the trouble. In his famous election against Rajiv Gandhi in 1989, his symbol was chakra and the EC/Congress made it sure that there are 15 more circle based symbols for independents and this chakra was in between the fifteen circles. The sysmbol hand was printed on top-right corner. Due to this there were many false votes and invalid votes.

India has moved forward and should find a way to make EVMs more reliable and EC has a responsibility to assure the public even from perception based doubts. They need to bring a lot more transparency.

I cannot take the line that all officials are corrupt and hence EVM are more prone to fraud where as for ballot box usage the same officials are not corrupt.

Raja Bose · Post by **Raja Bose** » 23 Jul 2009 07:14

Rahul Mehta wrote: (Aside : https://acc.dau.mil/CommunityBrowser.aspx?id=143395

It says that passive RFIDs can be read from 30 fts. But that is not relevant for us.)

Then you need to find out how large is the RFID reader/writer unit going to be

(Hint: big enough to make your RF scheme impractical). Dont expect a tiny hand held unit talking with some active or passive tag from 30ft. Hence, your RFID idea is a non-starter.

Rahul Mehta wrote: In the BU replacement theory, RF is in BU. Thru RF, a field agent from a distance of few meters communicates candidate number etc. The RF unit here can be 1 cm * 1cm or even bit bigger. As as long as BU does not noticeably heavier it would be fine. And if one opens BU, it should not be drastically different from real one.

In "Trojan on chip" theory, the RF is in CU and in Japanese chip. And it has to be real miniature. So what is the smallest RF that one can put on micro-controller chip that can take data from say 10 meters away?

So if the Japanese chip has say 2mm * 2mm RF unit that can send-receive say 10 bytes from a device 20 meters away, then trojan activation is possible. What makes me think that such RF unit can be placed is from the news I read on RFID. Handheld RFID readers can communicate with tiny RF chips (I dont know distance yet).

Again you are constructing wild theories without looking at the practical aspects. For all these chatting with tiny RF devices you need antennas....the RF chip can be tiny (in terms of packaging form factor) but it will require a large antenna to be able to TX/RX over the distances you propose. Re. handheld RFID reader/writers, have you ever used them? - then you will know their range (it ain't certainly in meters!). If you can try to look at some of the SmartDust work done by UC-Berkeley and Dust Networks - you will see why what you claim as feasible are in fact not. In fact whenever they used to proudly show off their "Smart Dust", one would always find one evil fella nastily asking them to show the huge antenna the spec of dust needed to get any communication going

. BTW Dust Networks is funded by In-Q-Tel which is a CIA-funded venture captial firm so maybe now we can shift from RF to Smart Dust based EVM exploits

BTW how do you propose to use RF inside metal casings of the EVM? And in general do any of your ideas take into account Murphy's Law?

Rahul Mehta · Post by **Rahul Mehta** » 23 Jul 2009 07:16

Sanku,

I did not contest that 100 camera's data can be faked. I want you to show how you will fake data from 50000 cams. And the data is real time because in the setup I proposed, the cam in booth send one pix to District server every minute. And from server, anyone will be free to download them and watch them on real time basis for some fee. This will increase confidence in voters that nothing wrong is going on.

Now explain how you plan to rig data from 10000 cams?

(Aside : One reason you cant explain is because IYO I dont know basics. Another reason can be that you yourself do not know how data at large scale can be faked within a short span or real time basis, and hence such comments. eg in IRV vs Two-Polls debate, you could not show how twice expensive Two-poll had any plus point, and so there also you had finally you resorted to insult throwing, like you are doing now)

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

Pranav,

The trojan theory of yours now looks better BU replacement theory of mine

. Did a lot of reading on RF chips, RFID and RFID readers yesterday, and apparently miniature RF sender-receivers seem possible.

Also, we hear that some X thousand EVMs went to Congress MP's company for "repairs". If he replaced the whole PCB with a PCB which is identical to actual PCB, then placing RF based trojan is TRIVIAL.

If there is a trojan, it can be activated as follows

1. Say polling is over, and and counting is yet to start.

2. The CU can support 64 candidates at most and stored the vote fetched by each candidate in RAM location #N to N+63. If one knows the code of BU, he can know location for each. Say N is 101

3. Say Congress is candidate no. 2. And BJP is no. 3.

4. Then trojan thru RF is told to subtract 200 from RAM#103 and add the same to RAM#101. Thus after poll ended, Congress gained 200 votes

So no agent has to touch the machine, and from distance of 10-20 meters, you can activate all CUs in that distance.

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

Dear Audit lovers,

So far, audit-lovers have claimed that "if rigged machines are still in District warehouse, audit can discover some of them by random check". That is only if SCjs appoint honest auditors and that can happen only if SCjs are honest to begin with. But I disagree -- SCjs are more corrupt than that BWM-Nanda HCj. Now honesty of SCjs is a non-technical issue and I dont want to discuss here. But if you are proposing audit, and if you are honest, pls write the assumption you are making (that you are assuming that SCjs are honest) in bold. Pls dont push that assumption under the carpet. Because that assumption of yours will be useful for me to convince people that "audit will never ever report rigged machines, even if they find some".

My arguments against audit is as follows

1. Audit of existing EVMs cant be done without SCjs approval or Cabinet approval or CEC's approval
2. SCjs , Cabinet and CEC are corrupt.
3. So they will appoint auditors who can be blackmailed as well as who are known to be corrupt and at the same time have excellent image in media (eg most ex-SCjs)
4. So when they "randomly" select 1000 EVMs, they will only select the good EVMs. And so bad EVMs will never be found.

Now you guys are assuming that SCjs are honest,. I wont contest your devout shraddhaa in SCjs. I only want you to WRITE that assumption in you "audit will discover bad EVMs" proposal.

Virupaksha · Post by **Virupaksha** » 23 Jul 2009 07:21

Muppalla wrote:
Sanku wrote:A Mupalla said he has posted a "why no paper ballot before" and it had more points, I will let him take the stage.
I am actually scared to get into that longer and lengthier replies . Hence I kept away from that topic. Here are few things I will write:

(1)The party with muscle power can easily manipulate the ballot based system as compared to EVMs. EVMs will not work if some one tried to sabotage the booth. Laloo's victories can be atributed to ballot boxes only.

Muppala,
can you explain more? What is the difference in muscle power for EVMs and ballot boxes?

Except the actual process of voting I feel everything else is same. In EVMs, you need to press buttons with 8 sec difference while for paper ballots you need to stamp, fold and then put into the box.

switching the cu by EU official for every vote is the same as signing every ballot paper.

(2) To carry the ballot boxes you need trucks not suitcase carrying jeeps. Securing them is challenge. Now imagine the manufacturing of the boxes and carrying them to booths and then securing them. You need good trains.

Are you serious? securing a bigger box which is double the size of EVM + CU suitcase, needs trucks?

(3) Even if there are 64 candidates in a constituency, you can imagine the size of the paper and the trouble to fold them and stuff them. There is no randomization of the symbol's slot on the ballot paper. Rajmohan Gandhi knows the trouble. In his famous election against Rajiv Gandhi in 1989, his symbol was chakra and the EC/Congress made it sure that there are 15 more circle based symbols for independents and this chakra was in between the fifteen circles. The sysmbol hand was printed on top-right corner. Due to this there were many false votes and invalid votes.

These kind of EC antics can be stopped with EVMs? put those 15 members along with Rajmohan on the same EVM with Rajiv being on the EVM at the top.

invalid votes might get reduced. I cant see how EVM can stop the above mentioned antics.

India has moved forward and should find a way to make EVMs more reliable and EC has a responsibility to assure the public even from perception based doubts. They need to bring a lot more transparency.

Definitely agree.

I cannot take the line that all officials are corrupt and hence EVM are more prone to fraud where as for ballot box usage the same officials are not corrupt.

It applies vice versa too. and also the points of corruption will be slightly different.

Muppalla · Post by **Muppalla** » 23 Jul 2009 07:36

ravi_ku,

If you remember the pictures of the boxes that are required for a booth that has 40+ candidates, you will understand. The man power to move them to booths and back to counting center. Imagine the number of counting officials at the counting centers to the current number. Just think through. Don't you thing the probabiliy of error/fraud increases with more people involved in the election conduction process?

Think through about muscle power to steal the boxes while they are being transported. All the EVMs going in a sealed police jeep Vs in huge trucks? Yes literally trucks. Just extrapolate 64 sysmbol paper and due to size of the paper, the size of the box increases. The number of boxes at a given polling booth increases. The carrying vehicle size increases. Now all a Laloo has to ensure is that each constituency has atleast 40 candidates to do the harakiri.

With EVMs 40 candidates is nothing and hence a large number of senseless/sabotaging independents also got reduced with the introduction of EVMs.

In addition imagine about a common voter and his inability to fold the paper and push through the slot as opposed to just pressing a button. Imagine the space that a booth needs for these boxes.

India has gone through a lot of horror and EVMs are a gift to the process.

I am not saying that EVMs are all perfect but instead of perfecting these, we are thinking to go back to those horrific days.

Rahul Mehta · Post by **Rahul Mehta** » 23 Jul 2009 07:41

Muppalla wrote:What if there are certain lots among numerous chips that are manufactured in Japan come with preloaded binaries or with trojans.

Dileep's MAIN point against Trojan in chip (or Trojan anywhere in EVM is ) : how would one tell the candidate row number to the trojan.

There are 700,000 EVMs. Say 100,000 of them have trojan or even 20,000 of them have trojan. If one person has to go close to EVM and actually punch in buttons to communicate with trojan, then personnel requirement shoots up . And the ink on the finger makes it worse.

If RF based circuit can be placed on chip or elsewhere, then issue is settled. Otherwise, trojan theory is incomplete.

---

Essentially, I dont want to stop at making claim that it can be rigged. I want to give complete operation manual type write up with personnel estimate and also mention which all people's co-operation will be needed to rig 20,000 to 100,000 EVMs.

Rahul Mehta · Post by **Rahul Mehta** » 23 Jul 2009 07:51

Even if there are 64 candidates in a constituency, you can imagine the size of the paper and the trouble to fold them and stuff them. There is no randomization of the symbol's slot on the ballot paper. Rajmohan Gandhi knows the trouble. In his famous election against Rajiv Gandhi in 1989, his symbol was chakra and the EC/Congress made it sure that there are 15 more circle based symbols for independents and this chakra was in between the fifteen circles. The sysmbol hand was printed on top-right corner. Due to this there were many false votes and invalid votes.

These kind of EC antics can be stopped with EVMs? put those 15 members along with Rajmohan on the same EVM with Rajiv being on the EVM at the top.

This is one thing I noticed: each time one supports paper-ballot, people start bringing issues which ALSO exist in EVMs. I request people not to bring equal-equal issues.

Also skip issues that have known alternate solutions.

--

If there more than 64 candidates, EVMs cant be used - DC will have to use paper ballots. So EVM does not solve issue of "large number of candidates". And if there are more than 16 candidates, then there are two BUs and with 33-48 candidates there will be 3 BUs and with 49-64 candidates there will be 4 BUs. So EVM does get bulkier as number of candidates increase. And since EVMs are fragile, one person cannot carry more than two EVMs - one in each hand. Where as in ballot, they can be handled without such extreme care.

---

The problem of "two many candidates" can be solved by raising deposit to 1951 level. We dont need ANY brains to solve this problem.

Muppalla · Post by **Muppalla** » 23 Jul 2009 07:59

25 candidates is enough for the ballot box to become bulkier. Upto 64 candidates there is no big deal in EVMs. If there is a chance to sabotage, the main scoundrels will spend money and make sure there are more candidates. Now the costs are increased becasue of EVMs as they have to bring extremely large number of candidates to beat the system. The parties will have money for few constituencies but not too many. Bihar under Laloo and many INC ruled states used to resort to such tactics. I have see ballot boxes of the size of general overhead size Syntex tank. They used many kinds of transportation to transport these boxes. 1980s and 1990s are all splashed with such elections.

Dileep · Post by **Dileep** » 23 Jul 2009 08:02

Since RF came up:

You can't integrate an RF receiver into a silicon chip. It can be done only on a hybrid or MMIC chip. Even those chips need external antenna elements, like PCB traces.

It is all a matter of size. The antenna need to be of size comparable to the wavelength of the signal in use. For remote activation, you can't use signals more than 2GHz. You need an antenna of several centimetres to receive those signals.

So, it is a physical impossibility (ie laws of physics, not technical limit) to integrate the RF receiver onto the microcontroller IC.

That notion can be dropped.

Muppalla · Post by **Muppalla** » 23 Jul 2009 08:05

Rahul Mehta wrote: Dileep's MAIN point against Trojan in chip (or Trojan anywhere in EVM is ) : how would one tell the candidate row number to the trojan.

If it is feasible (I would wait for Dileep's reply), then the binary itself is written so that the certain votes from certain buttons are transffered to some other buttons. The randomizing program that assigns the buttons to candidates is also compromised to match the pattern in the compromized ROM program. That was my take if such a thing is possilbe in a long shot.

My take is the party that sabotaged the EVMs ( if they did ) did not do using any EC/district officials. They did at source with a lot of planning. Something like the computers' hacking shown in DieHard-4. If feasible then it is probable. ofcourse verification can disprove that and anything like that may not have happened.

Dileep · Post by **Dileep** » 23 Jul 2009 08:19

@Mupalla's query about embedding a "trojan" on the chip at the fab:

There are both technical and logistical issues with that.

Technical:

1. You need to add an extra execution unit (ie another core) to the microcontroller. You can't "sneak" something into the "microcode". To understand that, you need to understand what is microcode, and how it works on a microcontroller.

2. Adding the extra core, which is comparable in size to the original core, (because it is a small microcontroller, not a pentium class processor), the power consumption of the chip will be remarkably high. Since the added core "steals" cycles (unless you add a multicore manager) the original processor will become slower. Both will get caught in QA testing at BEL.

Logistical:

1. Fabs can't do small lots. They will have to run the rigged devices as a separate run, and the regular devices in a separate run. The chips come out in packaged trays from the production line. If you want to mix the rigged chips, you will have to open each and every tray, and randomly replace the packaged ICs on them. Then you will have to re-package it. You can't feed them back into the packaging machine.

2. The legend that is stamped on the ICs have a date code system. The fab will have to force its production system (which is very stringent) to make two different runs have the same part number and date code. This is IMPOSSIBLE in a semiconductor plant. You can't go an explain to all the quality staff why two entirely different chips have the same part number (which you can blame on the customer maybe), same mask id (no way acceptable) and date code (no way acceptable).

You can't bribe or balckmail a commercial fab into that kind of an operation.

Dileep · Post by **Dileep** » 23 Jul 2009 08:22

Muppalla wrote:
Rahul Mehta wrote: Dileep's MAIN point against Trojan in chip (or Trojan anywhere in EVM is ) : how would one tell the candidate row number to the trojan.
If it is feasible (I would wait for Dileep's reply), then the binary itself is written so that the certain votes from certain buttons are transffered to some other buttons. The randomizing program that assigns the buttons to candidates is also compromised to match the pattern in the compromized ROM program. That was my take if such a thing is possilbe in a long shot.

My take is the party that sabotaged the EVMs ( if they did ) did not do using any EC/district officials. They did at source with a lot of planning. Something like the computers' hacking shown in DieHard-4. If feasible then it is probable. ofcourse verification can disprove that and anything like that may not have happened.

The order of candidates on the ballot is as follows:

Recognized Parties
Other registered parties
Independents.

In each of those blocks, the individual candidates are arranged in ascending order of the alphabet in the local language. So, here in DMA, the order was "Thomas", "Radhakrishnan", "Sindhu". If "Hybi" was the cong candidate, as the favourite, he would have been last.

Raja Bose · Post by **Raja Bose** » 23 Jul 2009 08:29

Dileep wrote: You can't bribe or balckmail a commercial fab into that kind of an operation.

ah! But Dileep-ji we are talking of CIA/NSA's own fabs here

(they do have them BTW).