In voting scemes I have seen so far that are based on homomorphic encryption, a group of partially trusted "authorities" share the task of decrypting the result.
The voter has to trust at least one of the authorities to be sure that his personal vote will not be decrypted individually. At the same time, every voter can verify that the (joined) authorities are not tampering with the result.
Is there a fundamental reason that the voters can not simulataneously all be authorities?
If this would be possible, the only way to find out the vote of A is to compute and decrypt the votes of all voters, and the votes of all voters minus A. But as A has to help decrypting, he/she can inhibit the latter.
I can only think of practical reasons:
- Every voter could also inhibit obtaining the overall result. I guess there should be ways to handle that. These ways may be costly.
- It would be too costly.