Up to now e-vote experiments have all been paired with a usual ballot-paper vote. It has been compulsory to vote also in the old fashioned way since ballot-papers were the official votes for any recount. Thus we can say e-vote experiments are an expensive way to demonstrate how hardware & software vendors are good in producing voting machines.

Pairing paper voting with electronic voting tends towards giving us confidence in the latter. If software & hardware vendors succeed in not having (or masking) problems, in a few years time public opinion will be convinced that e-vote gives the same results as paper elections and thus we'll accept to vote without any alleged useless paper verification.

At that point we'll be in a mess since technology never stops and thus there always will be new software releases, new hardware architectures, new network technologies, new hacking technologies... Furthermore we will possibly have to trust new shareholders, new management and new employees of companies manufacturing e-vote hardware & software.
We will have to passively accept any electoral results coming from such much changed situations because in the lack of old-fashioned ballot-papers we'll have no way to do any "visible" verification.

P.S. In real electronic elections as in India, Brasil and USA there has been no parallel paper vote and infact electoral results has been accepted without any verification.