 Thank you, Manthi, for the introduction. Like you mentioned, this talk is going to discuss rather some of the design considerations in implementing digital voting systems, electronic voting systems. And I must thank the previous speaker, Shivam. He pretty much covered everything that I wanted to cover in my talk, all by it in a much simpler and nicer way. So my presentation would mostly be the formalization of the ideas that Shivam spoke of, primarily for operationalization purposes of electronic voting technology and such. Now, why these properties are important because election security concerns are ever present as very nicely put by Shivam. And this could be due to interference by foreign powers or due to unauthorized voting, voter disenfranchisement and technological failures. So basically, they all inevitably call into question the integrity of elections. Now, what is clear in the last three decades of very fine research, at least in the area of computer science on designing voting systems, is that it should not be necessary to trust any authorities individually or collectively for establishing the correctness of the election process. So any digital voting should essentially be constructed, keeping this in mind that it should be allowed to operate without some explicit trust requirements. So for a digital voting system to be able to operate in such a way, it should satisfy certain minimum requirements before these instruments could be considered for voting, right? So for enabling electoral democracy. So it's because they are sort of in a manner enabling electoral democracy. So what is important to note is that these set of technical requirements is in fact driven by certain key obligations. And what are those obligations? The first obligation is that the losing candidate has to be provided with a convincing proof of their loss. So that's from the candidate who's participating in the election. And from the voter side, the voter should she demand be supplied with the guarantee or a proof that her vote was indeed cast as intended, which essentially means indicates that the voting machine has registered the vote correctly and recorded as cast, indicating that the cast word is correctly included in the final tally. And finally, counted as recorded, which means that the final tally is correctly computed from whatever records were essentially available. And finally, some other key obligations is that no vote should be recorded other than by those who passed through this process of eligibility verification. So it should not be the case that I can come into the polling booth and somebody else votes on my behalf, as Shivam was pointing out, that there were cases which cropped up during Indian elections at some parts of India. By the way, I must say that to solve such problems, electronic voting systems will not cut it. Why? Because technological solutionism can only make things messy if not done right. And there are very high chances that technology, if being constructed without putting due thought and due diligence, will inevitably create more problems than solve issues. So the actual solution lies in designing the process rather than focusing too much on technology. And finally, all votes. So the last obligation for designing such voting systems is that all votes must be kept secret during and even after the polling, all the way up to tally. So now, the first three obligations where votes need to be casted as I mean, so voter has to be provided these guarantees that the votes were indeed cast as intended, recorded as cast and counted as recorded. They together form the correctness aspect of a voting system. And this correctness, when you want to take this property and sort of operationalize it as if I want to design a system, how should I interpret this correctness property? It invariably ends up splitting down into two properties. The first is universal verifiability, which essentially says that through publicly posted data, anybody should be able to verify the correctness of each vote from the tally and from the records that are available. Now, before I go any further, I must also spit out what is the threat model under which I am defining these terms. So I have to essentially talk about what are the powers that an adversary can have which can create a ruckus in an election process. So under an adversarial setting, your adversary can do quite a few things. What are those things? So it can corrupt. It can be a corrupt election authority itself. So that's a modeling of an insider attack. Or it can corrupt the polling officers through whatever means. Or it can corrupt the voting machines themselves that are being used to conduct the election. The adversary has the power to alter or delete cast votes during the polling or during the collection process or even during the or while publishing the recorded votes into some public bulletin board. It can introduce fake votes in the system, which are essentially those votes that are not certified by the polling officer. So under such a threat model, assuming these are the powers that an adversary can have, what voting systems must guarantee, what voting systems must guarantee, and those two properties are essentially universal verifiability and individual verifiability. So under universal verifiability, like I said, a voting system is universally verifiable. If anyone in the public can verify using publicly posted data that each vote is indeed recorded as cast and counted as recorded. So the last two properties specified in the first line of the slide. And then there are, of course, no spurious votes in the final tally. Individual verifiability, on the other hand, essentially means that if a voter can obtain a proof that their vote was recorded as intended in the final tally, which means cast as intended as well as recorded as cast. So these two properties together can compose or rather compose into recorded as intended property. So what I'm doing is I'm essentially capturing all those thoughts that were already covered by Shivam in the last talk, but formalizing them if you want to take them up and operationalize in a system. Now, so those were the key obligations for correctness. Our security secrecy, rather, vote secrecy is also a key obligation. And under vote secrecy, one must never reveal how a given voter essentially voted to prevent what? To essentially prevent vote selling or coercion. So voting systems where votes are directly recorded, which is such kind of systems are also known as direct recording electronic systems. It is assumed that for such kind of systems, it is assumed that the machines will not yell out and leak the vote. Or they are constructed in a manner that through side channels, they will not leak out information thereby compromising the vote secrecy. So if I have to consider secrecy and coercion resistance from these properties, keeping these properties as pivots, I would have to then start thinking about what can an adversary do. So I'll have to essentially provide a threat model under which I have to situate these properties. So as a threat model for an adversary for secrecy would be that the adversary would be able to observe all voting receipts, voter receipts. Or they would be able to observe all voter verified paper trails or paper records, VVPRs, if there is one instituted in the voting process. Or the public outputs that the Election Commission decides to put out after the voting process has been conducted. Of course, adversary can participate as a bare-handed voter, can have all the powers that the previous adversary had when we are talking about correctness. And here, the adversary can also corrupt the voting machines to reveal votes and other secrets. What the adversary of keeping the secrecy as private cannot do is they cannot corrupt the election authority to reveal the votes or other secrets. So under such an environment, the system ensures voter secrecy. Then given possibly even malicious voter receipt and publicly posted data, no information can be derived how the voter voted. So that is an individual voter cannot prove to anyone else how she voted. And that is essentially tagged here as individual vote secrecy. Community vote secrecy, on the other hand, is when a bunch of voter receipts, having a bunch of voter receipts and publicly posted data and adversary cannot derive how a community voted. So well, this is community vote secrecy. It is very, very important to preserve this community vote secrecy as well from unintended targeting or intended targeting for that matter. And finally, before I go forward to quotient resistance, it is important to state here that individual vote secrecy turns out to be a necessary condition for quotient resistance. And now I come to the point of quotient resistance. It's essentially a property which says an adversary instructing a voter to vote in a certain way cannot determine if the voter did follow the adversary's instructions or not. So quotient resistance is very, very important. Otherwise, all fairness and the democratic principles behind electoral, behind voting is then lost. Finally, so community vote secrecy is also important to prevent. Why? Because you want to avoid profiling and targeting of voters belonging to a particular locality. And revealing the vote tally of a polling booth mapped to, say, a small neighborhood compromises community vote secrecy. So essentially, I would come back to a point that Shivam was talking about that if I were to put out this data of what votes were casted on a particular polling booth, that polling booth essentially captures the voting patterns of a particular neighborhood. It is small. And therefore, there is a high chance that community vote secrecy would be compromised. So a suggestion made here is that only an aggregate tally is published to avoid community profiling. You don't publish polling booth level data, but you aggregate the votes across polling booths and then publish them. All right, so these were the important properties of correctness and secrecy. But in addition, if you want to make sure that not only a voting system has this important property of correctness and secrecy, it also gives you some amount of efficiency, then you will also have to start considering other properties, such as recoverability. Now, recoverability has to do with the graceful recovery of a voting system in an event of an attack. How can a voting system recover back? Now, note that an adversary can participate as a voter, as I had mentioned in the previous slide. And the voter, the adversarial voter, can come back and falsely claim that her vote was not recorded as intended. So what happens in such a setting? So basically, either there is dispute resolution protocol which figures out whether the voter is indeed saying true things or making false claims. In addition, what one would want to have in such digital voting systems is that this. So basically, the recoverability property is primarily tied to a software independence property, which demands that a small detected change or error in an election outcome can be corrected without rerunning the election. So software independence to take place, one needs to have mechanisms in place to identify, A, that error has happened in the voting process, and B, to be able to fix the accountability where the error has happened, which polling booth did the error originate from, et cetera, et cetera. So that there can be a graceful recovery. Instead of rerunning the entire election, you could perhaps do a reelection only at a particular polling booth later. So this is the traditional definition of a recoverability, which is strong software independence, which essentially demands that a detected change in error in an election outcome can be corrected without rerunning the election. So now, if this software independence recoverability property has to be guaranteed in a voting system, it would inevitably require you to go out of the digital system and maintain some additional record. So that's where this idea of VVPR comes in. So public VVR existence is very, very important. Not only that, one should be able to essentially, so pretty much the common understanding is that if you have VVPR facility in your voting system, that is going to be a trusted entity. So I will look at, because this is where the intention of the voter is recorded in a manner of speaking. So this is assumed to be correct. Now, the problem is that if I have to show the software independence, I would have to have a one-to-one correspondence of VVPR, voter verified paper record, with a digitally recorded vote or the electronic vote. And so what would this allow? So this would allow one to precisely identify those votes where there is a difference between the VVPR and the electronic votes. So the electronic votes don't match up. So of course, one can. So VVPR verifiability is a very important property for recoverability to exist for voting systems. Now, this one-to-one correspondence, the VVPR that I have been speaking about, if it can be publicly demonstrated for any VVPR without compromising individual and community vote secrecy, then we call the voting system to be VVPR verifiable. Now, designing such a voting system is a challenge. And I think there are a lot of voting protocols that are in existence. And whether or not all such voting protocols provide public VVPR verifiability is a matter of question. So and it is also worthwhile to note that there is an inherent tension between this strong software independence and community secrecy. So verification, say, of a VVPR against a digitally recorded vote may fail in many types of cases, where receipts of VVPR exist, but there is no corresponding electronic vote, or there exists electronic vote with no corresponding receipts of VVPRs. And if during audit, it turns out that the margin of votes between the winner and the next highest vote-getter is too narrow to ignore, then to ignore these verification failures may not be the right thing to do. And it may become imperative to identify the polling booths from where this fault arose, or a set of polling booths from which the fault arose, and conduct polling again in them locally. So however, the problem is, so this process can leak information. So an adversary monitoring the published votes before and after the repolling can really determine which votes have changed. And the adversary can then use this information to correlate the changes with the polling booths involved in the repolling. So such community secrecy compromises, ideally should be limited to recovery procedures and should be minimized to whatever extent possible. So the general wisdom would be to conduct polling at a larger scale, repolling at a larger scale. The larger the size of the repoll, the lesser is the leakage of information. So that's the wisdom. Finally, I want to talk a little bit about dispute resolution a little more. So a voting system should not just be fair, but also appear to be fair. And for that to exist, dispute resolution is one of the key components for, and it should be there in one of the voting systems where the intention of the voter is not recorded directly by the voter. If it is recorded by the machine, then dispute resolution becomes a necessity. All right, so having understood these properties, what I will do now is I will talk about some electronic voting protocols, which are primary. I mean, so basically two electronic voting protocols, one representative of each type in each category that these voting protocols fall in. So essentially, electronic voting protocols are often of two type, optical scanning electronic voting protocols, where there are hand marked paper ballots. So voter comes in to the polling booth, looks up at the paper ballot, marks the paper ballot herself, and then gets it scanned. And that's why they are called optical scanning paper ballots. Of course, the system will provide an encrypted part of the vote as a receipt of the ballot as a receipt. And these, the encrypted receipts are displayed on essentially public bulletin boards, or the encrypted vote rather is displayed on the public bulletin boards. And there are many such voting protocols over the years that have been designed, which guarantee end-to-end verifiability, right? So many of them provide universal verifiability as well as individual verifiability. The other type of voting protocols are these direct record electronic voting protocols, where the vote is not recorded manually on a paper and then later scanned, but it's directly recorded by the machine. And this is where problems can arise because how can one then establish that the machine has truly recorded your intent in a correct manner. So there are quite a few DRE protocols as well. I have listed some of them here on the slide. So going forward, there are of course, some strengths of these voting protocols in the sense that they provide you these correctness properties and end-to-end verifiability, but there are also limitations. So for example, in DREs, there is a dispute resolution problem and there's no easy fix for this dispute resolution for DREs. Of course, the dispute resolution problems do not arise in optical scanning protocols because the intent of the voter is directly recorded by herself. But DRE or not DRE, there are these randomization attacks, which are a problem. And so what happens under these randomization attacks? So many voting protocols try to prevent electronic recording machine from learning the vote. And in process of doing so, what they do is they randomize the permutation of candidate order on the prepared ballots. So that the voter's selection appears to be random and can only be decrypted at the backend. So every voter will see a different permutation of candidate orderings. Now, these are essentially susceptible to coercion attacks where a coercer can ask the voter or force the voter to vote for whichever candidate that appears at some fixed point or position in the permutation. So the vote may then get randomized. In a manner of speaking, you can understand this as a denial of voting attack, right? So instead of denial of service, it's a denial of voting attack. Now, the other limitation of these electronic voting protocols is the fact that they rest on very sophisticated and complex cryptography. And so the German Supreme Court in 2019 made an observation where it said that the use of voting machines which electronically record the voters' votes and electronically ascertain the election result only meets the constitutional requirements if the essential steps of the voting and of the ascertainment of the result can be examined reliably and without any specialist knowledge of the subject. Now, these last few words are the words which are very, very important when I want to lay the emphasis without any specialist knowledge of the subject. And what's happening with most of these electronic voting protocols is that they take the agency away, not just from the voter, but also from the polling officers. So it's some magic that's happening in the voting machines. And I do not have a perfect, as a voter, I do not have a perfect understanding of what's going underneath. Neither does the polling officer, but we are expected to just trust the workings of this machine to have recorded everything correctly and secretly. So, many of these protocols do not provide... So, and of course, the last limitation is that I was talking about this idea of public VVPR verifiability where this one-to-one correspondence with VVPR slips is important. And it so appears that many of the modern-day, end-to-end verifiable voting protocols do not provide this correspondence. And the absence of which the trust on the electronic voting system is hard to establish. So I'm left with 10 minutes. I had prepared a couple of slides talking about the process of what happens, say for an optical scanning voting protocol and a DRE-based voting protocol. So in the interest of time, perhaps I will not go through all the details, but limit my discussion. I'll keep it at a very high-level description of what goes through, right? So as a first step, there is eligibility verification where a person who is eligible to vote is authenticated at the polling place in accordance to whatever procedures are put in place for voter registration and et cetera. Now, here I want to come in and comment on what Shivam was talking about. Electronic voting machines, of course, theoretically are hackable, but much of the fraud actually takes place in the electoral roll lists, right, at the eligibility verification phase. So it's very easy to delete a voter from an electoral record. Maintaining the integrity of electoral record is of paramount importance. And we'll see that perhaps one, this is perhaps one area where a very specific variant of blockchains could be applicable without the network, just a cryptographic, untemparable bulletin boards, append only bulletin boards, can implement electoral records. That is essentially a part of blockchain because blockchain is this plus our distributed consensus. So Shivam is absolutely right. I mean, voting on blockchain or blockchain for voting doesn't make a whole lot of sense now. Or in posteriority, but there's a lot of hype around it. OK, so the second procedure, yeah. So before the voting happens, let me very briefly talk about how the ballot is prepared. So at least in this optical scanning protocol that I'm discussing, which is Scantigrity 2, the ballot consists of a voting portion, right? So this is the voting portion, a receipt portion, which is shown below after this perforated doted line, with a unique and every ballot has been provided a unique ballot ID number, right? So the ballot ID number is there in the voting portion as well as in the receipt portion. And the voting portion of the ballot includes a list of candidate names with an optical mark, because this is a hand mark based protocol. So there are these optical marks. And all that voter has to do is use their pen to mark on. So initially, all of these optical marks would not show these very unique IDs that are behind them, right? So the ballot includes basically these list of candidate names. But each bubble contains a sequence of randomly generated alphanumeric characters, so to speak, which are essentially also called in literature as confirmation codes. And these are printed by an invisible ink. So before the ballot is marked, none of the confirmation codes are essentially visible. And if a voter has to indicate her preference, all she has to do is just use her decoder pen to just mark over these bubbles. And that particular alphanumeric encoding would present itself, right? So once it is done, so the receipt portion is optional. I mean, if the voter chooses, she can record this alphanumeric text in the receipt portion. And because it's perforated, she can air it off and keep it herself. Actually, she has to show it to the polling officer and the polling officer will stamp it by saying that, OK, so the voting has been recorded, right? But at least cast. Now, so that's essentially the receipt stamp, stamping and the collection. And then what the voter does is she, in the presence of a polling officer, scans the top part or the voting part of the receipt of this ballot, and it shows up on the screen so she can cross check whether the ordering is correct. In fact, what is important to note here is the actual vote would not be available to the machine. So the machine will not get to know that the voter has actually voted for RT. All that they would be able to scan is that alphanumeric encoding. So now the voter can go ahead and cast her ballot and then basically take this receipt home. And then there is this entire post process of posting these votes and tallying these votes, which is a very interested, sophisticated, cryptographic, cryptographically based system where the Election Commission is maintaining a bunch of tables, some private, some public. And these tables are connected to each other via mappings, which are essentially cryptographically protected. And these mappings can essentially be approved that nobody has essentially gone and changed data across these tables. So the cryptographic primitive that is typically used to prove whether at the time of tallying that these mappings are indeed correct from the vote to whatever encrypted data that has been put out in the public domain, there is this cryptographic primitive called MixNet. And so this is the case in optical scanning scantigrity protocol. Of course, one could change this cryptographic primitive. The other kind of protocols use something called as homomorphic encryption, where tallying is done using this homomorphic property, where votes being added, the encryption of votes being added is essentially equal to the encrypted vote being multiplied together. So this is the additive homomorphism. And so some of the voting protocols, their back ends essentially use this homomorphic property. I think there are a lot of technical details. I would not want to get into it. Otherwise, this will just confuse a lot of audience. And so in the interest of time also, let me move ahead. So optical scanning voting protocols, the VVPR verifiability is essentially non-public. It can be done. So the public VVPR verifiability does indeed exist, but it is non-public in the sense that they would have to open up the entire MixNet in order to make these mappings between the electronic vote and the VVPR. And in the process of doing so, the vote secrecy would be breached. So it cannot be a public process. The other thing for DRE voting protocols is the fact that VVPR verifiability is only partially available because, again, that would mean. So the technical detail behind it is that the homomorphism has to be, again, they will have to open up everything, which is counterproductive because it defeats the purpose of having this homomorphic encryption in place in the first place. More importantly, DRE voting protocols such as StarVote provide no secrecy from DRE scanners. So somebody use scanning the vote or printers. So that's that. And like I said, what happens in the back end is, so once the tallying has happened, the election commission has to provide a proof that the tallying has been done correctly. And again, a very sophisticated computer science technique is used here to show that correct tallying has happened. What the authorities do is they decrypt the encrypted tally of votes and publish a proof, which is a very interesting proof. It is called a zero-knowledge proof of the correct decryption. So essentially, the proof is saying that I have not been any hanky-panky during the decryption process. So the tally is indeed correct according to whatever recorded data I have. Let me skip this introduction to these crypto primitives of what is zero-knowledge proofs and other. So zero-knowledge proof, let me make one statement on it. It's a proof strategy where if there is a declarative statement that requires a proof, one is able to obtain a proof for that without knowing the knowledge required to design that proof. So no knowledge behind the proof is essentially leaked. But the verifier is convinced in a certain way that the declarative statement is indeed correct. The assertion made in a particular declarative statement is true. So what I'll do is I will take last two, three minutes to touch upon internet voting and block-based voting and primarily upon their unsuitability. So internet voting, in my opinion and in many others' opinion, of course, in fact, a lot of very big computer scientists in a sense that who have been working in this space for decades altogether have clearly talked about the unsuitability of internet voting. Why so? Because A, how do we guarantee coercion freedom? If you have an app or a mobile phone and from your bedroom you are allowed to vote, then coercion freedom goes for a toss. There is no way where you can guarantee coercion freedom. So it's just not possible. So there's an impossibility reserved right there. The other question is, how can we trust the device and the app themselves? And then in a day and age where Malwarez can see through a different app that you downloaded, not the voting app, but through a different app. And that Malware can break havoc. So not only can it break your voting secrecy, but voter secrecy, but it can do all sorts of unimaginable stuff, breaking all the correctness properties. And how do we really establish that your device and app is indeed correct? So that itself is a computationally infeasible exercise, as has been established in literature. Again, how do we guarantee software independence in the presence of such device and app? So software independence goes for a toss. And therefore, if you don't have strong software independence, you don't have recoverability, and so on and so forth. And again, then auditability and other properties also come under question. So a very serious computer scientist who has done a lot of work in the space at Honored Revest. In fact, he is also a winner of, during award, Noble Prize Equivalent. He was the guy who gave an RSA, an encryption strategy. He says that internet and blockchain-based voting would greatly increase the risk of undetectable and national scale election failures. And he substantiates it with a lot of information. So to say that, to have internet voting just for ease and convenience, and to increase the voter turnout, apparently that also turns out to be incorrect. There have been various studies done in Switzerland and Belgium where they have shown either no impact on voter turnout or rather negative impact on voter turnout. And there have been other studies done in Canada which show that, in fact, so this study was not in Canada. I think there's a typo here. It was done in Estonia where they said that, well with internet voting, what they saw was that it would favor groups with higher income or higher education. So demographics who have higher income or higher education. And that is understandable. So how would a person who is not lettered would navigate such technology? So usability also is under. So yes, all of these properties that people typically point to while promoting internet voting or blockchain based voting, people have done studies and they don't inspire confidence. Again, unsuitability of blockchain based voting. So well, this is again a non-starter because blockchains are essentially built on top of these nuts and bolts called smart contracts. Who's going to establish the correctness of these smart contracts? Your smart contracts are wrong. Your entire blocking blockchain cannot be trusted. So it's USP falls flat on its face. Well, it's a network and then therefore it will open itself up to a much larger set of attacks like denial of service attacks, right? Or collusions and cartelization. Group of users operating through their pseudonymous identities get together to collude, create cartels and create a fork in the network. And then what they do is they show one tally to one set of people and other, you know, so one kind of information to one group of people and one other kind of information to other group of people, right? So this is opening a whole can of worms at a point in time when we are still, you know, working on unconnected sequential voting machines, digital voting machines, right? Again, so if you were to consider permissionless blockchain voting systems, then clearly the visibility of votes is under no question. The votes have to be visible to all the players, all the participants in the blockchain. And then how do you talk about preventing coercion and vote selling when the vote information is freely available for everyone? Again, where so a limited setting where a component of blockchain technology that could be used, which I had referred to earlier in my talk, is in implementing electoral roles. And what is that component of blockchain? Essentially this idea of cryptographically chaining the records of a blockchain. Through this cryptographic chaining, one could obtain properties of untemperability, append only untemperability to these records. And this is what we want in electoral roles. So once you have been added as a registered voter for a particular constituency, nobody should be able to manipulate those details, right? So these are the properties that would be guaranteed through certified, through cryptographic hashing or chaining, hash chaining of these records. It gives you these properties of certified publishing in an honorable history. So that's a very limited component where a component of blockchain perhaps could be used. I think with that, I will end my talk. Thanks for your attention. The early version of our article from which many of the ideas that I discussed in the slides is available here. Thank you. And I'll open to taking questions now. Thank you very much, Pabza. Very interesting presentation from you. I'm very informative. And you provided an excellent framework for judging suitability of different types of porting systems. And you discussed some of them also. I wish I could say for myself that I understood everything. And now it's clear what is possible and what's not possible. But clearly, it's difficult to compress so much into one session from both sides for the presenter because you know so much and also for the audience to grasp all of it. I have a question and there are a couple of questions from the audience. So we quickly take them up. The one question that came to my mind was that I need to be able to verify that my vote was correctly recorded and counted. But if my vote, if I can verify that my vote was directly recorded and counted, it opens me up to the risk of somebody asking me to do that verification in the presence of a witness. So these two things will always be in conflict. Would it, would there not be? Right. So essentially your vote is never revealed. No, vote will get revealed. If I can verify that my vote was correctly counted. So somebody can stand on my head and say, okay, reveal to me that it has been correctly counted and which means reveal to me that it has gone to the candidate that I wanted to vote for. Correct, but this would be something that is a mapping which you only know of and which is not given as an information in the receipt that you got. The included receipt. That's okay. That's okay. That's okay, but if I can verify this. Yeah, sorry. Go ahead. If I can verify it, right? After the votes have been recorded, if I can, if I can access that and see that yes, it is correctly recorded, then somebody can watch over me to see that, yes, what am I seeing? It's not only an assurance to me. It is, it's a verifiability that I can execute later after casting the vote. And that opens me up to some kind of fact. But it's not a piece of information that you can provide to anybody. You always have this freedom of lying. If I may come in, you can give this proof in zero knowledge. So there are graphic techniques where you can give the proof in zero knowledge without revealing. And all, all such proofs will have to be zero knowledge proofs. And there are easy techniques over the last 30 years. That's true. That's true. That's true that zero knowledge proofs can be constructed, but this would make it, make the German courts absolutely, absolutely. You know, it would make it very difficult for people to deal with zero knowledge proofs unless it was their apps which were doing it, in which case the trust shifts to the app or to some other system, you know, and it creates its own problems. So thanks a lot. And, and there are a couple of questions also from Mr. Balum Netra. He says, won't targeting of voters in the community level voting rate as widely and publicly available be more easy to detect and therefore it may prevent or help quien prosecute such assaults as well as build social pressure against targeting. won't targeting of the voters when the community level voting data is widely and publicly available will be more easy to detect. And therefore it may prevent or help prosecute such assaults as well as build social pressures So of course, I mean, how do you establish a correspondence or rather some sort of a correlation between targeting of voters and this community profiling. So that's where the problem lies right. I mean, of course, the targeting is indeed happening but in ways where you cannot say that this targeting is happening because that targeting the people behind that targeting have this this sort of an information. So, if a party A, if a particular party A knows that in a particular community, they don't get a vote, they will, they are anyways doing it they will preserve their resources and not do a whole lot of campaigning there, where they know that there are fensiters perhaps they will put in a lot more resources and try to swing the vote there. So it's already happening. How do you establish that. How do you establish that. And you can't also be sure that publicly if something is available something is known that pressure can be built enough to prevent it from happening. That's also not necessarily, you know, that's debatable as well. He has another question. You have used some kind of optical scanning for ballots, thus keeping two parallel tracks of votes, the digital and the paper. Did this make the system very fine. So, basically, so yeah so basically what happens is the paper ballot paper. The paper copy of the ballot is typically shredded and put into dustbin otherwise some correlations can be made and motor secrecy can be compromised. So there are some districts and consequences where optical scanning based ballots are indeed used. So whether they really do maintain the paper copy of the ballot. So once it's been scanned, they maintain the digital copy. Perhaps they can keep the paper copy for VVPR public VVPR verifiability. That's not correct in United States you are not allowed to do electronic elections. Yes, as a nation doesn't do it but there are consequences for example. No, no constituents is allowed to do it. So you always do only paper elections. The problem is that verifiability does not come because the VVPR and the electronic elections has run as two independent parallel elections. They're not coupled together. So that's why the verifiability does not come. Okay, so one to one correspondence between them is the missing piece. Yeah, so thank you. Thank you very much for this very interesting talk.