5 ms·

And how do we know the machines no-one was able to (publicly) hack are the same ones, running the same code, that are then used in the elections? That no on-chi

And how do we know the machines no-one was able to (publicly) hack are the same ones, running the same code, that are then used in the elections? That no on-chip memories were re-flashed, that no secret hardware bug/backdoor was used? No matter how open source/open hardware/open audit you go, in the voting booth they're still black boxes that could have been tampered with.

jezclaremurugan 10y ago
I'm not saying these machines are unhackable, but with the addition of additional constraints they are _almost_ perfectly secure. For ex., On election day, mock voting and counting is done in the presence of all party representatives - the results are then erased before the actual counting. And representatives of all parties are mostly around when these machines are moved around and counted. Also, the way party symbols are listed is not pre-determined etc. and hence cannot be predicted by someone wanting to add votes to a particular party. Source - my mom has been a presiding officer during elections a couple of times, and I've looked at the training manuals..
- nileshtrivedi 10y ago
  Mock voting can only detect malfunctions, not backdoors. Randomization of candidate ordering is a trivial challenge to overcome for a backdoored EVM. Don't fall for EC's PR. Here is an example: Is NOTA ("None of the above") option's location also randomized? If not, a backdoor would listen for this sequence: NOTA, NOTA, NOTA, X, NOTA, NOTA, NOTA, X. And then it would start re-assigning votes in favor of X with say 60% probability. It doesn't matter where X is located. The backdoor might trigger only after a few 100's or 1000's votes have been given. How will mock voting or randomization help here? The sad thing is that all these weaknesses have already been detailed in Hari Prasad's original paper. But EC continues to dish out the same flawed defense: https://indiaevm.org/evm_tr2010-jul29.pdf https://indiaevm.org/evm_tr2010-jul29.pdf

jezclaremurugan 10y ago

I'm not saying these machines are unhackable, but with the addition of additional constraints they are _almost_ perfectly secure. For ex., On election day, mock voting and counting is done in the presence of all party representatives - the results are then erased before the actual counting. And representatives of all parties are mostly around when these machines are moved around and counted. Also, the way party symbols are listed is not pre-determined etc. and hence cannot be predicted by someone wanting to add votes to a particular party. Source - my mom has been a presiding officer during elections a couple of times, and I've looked at the training manuals..

nileshtrivedi 10y ago
Mock voting can only detect malfunctions, not backdoors. Randomization of candidate ordering is a trivial challenge to overcome for a backdoored EVM. Don't fall for EC's PR. Here is an example: Is NOTA ("None of the above") option's location also randomized? If not, a backdoor would listen for this sequence: NOTA, NOTA, NOTA, X, NOTA, NOTA, NOTA, X. And then it would start re-assigning votes in favor of X with say 60% probability. It doesn't matter where X is located. The backdoor might trigger only after a few 100's or 1000's votes have been given. How will mock voting or randomization help here? The sad thing is that all these weaknesses have already been detailed in Hari Prasad's original paper. But EC continues to dish out the same flawed defense: https://indiaevm.org/evm_tr2010-jul29.pdf https://indiaevm.org/evm_tr2010-jul29.pdf