Earlier quoted context omitted.
There is plenty of research and sample implementations of electronic voting schemes in which the final outcome is verifiable without anyone being able to see anyone else's vote. As an example, I worked with Michael Clarkson on an implementation of Civitas: http://www.cs.cornell.edu/andru/papers/civitas-tr.pdf
Agreed. There are a couple of other papers as well. The problem is they are far too complicated to explain to voters which makes them unlikely to be adopted and unlikely to be trusted. Having a secure vote is obviously the primary goal, but having a vote that people trust is pretty important too.
This system was designed by a couple cryptographers, one of them Ron Rivest of RSA fame. For details on this and a couple other simple voting systems, see here: http://rangevoting.org/RivSmiPRshort.html