There are demonstrated vulnerabilities in SSL/TLS (Secure Sockets Layer/Transport Layer Security -- the method used to securely exchange data on the web). Most of them are very hard to implement in the real world, but they do exist. Further, any encryption standard that relies on the exchange of keys is subject to man-in-the-middle attacks, and with the recent development of
DNS cache poisoning, we have a real world example of an exploit that gives all the appearances of being a valid secure session, but where the user is never interacting with the intended website.
The point here is that all technological voting solutions will have vulnerabilities, and in a game where the stakes are so high it is only prudent to assume that those vulnerabilities will be exploited. While exposure to possible exploits can be managed, it comes down (as all security issues do) to how much you trust the humans involved.
So... I think it has to be asked, what advantages does electronic voting offer and do those advantages outweigh the risks? Additionally, could we gain similar advantages without incurring the risks by using low-tech solutions (pen & paper) by simply altering the methodology of voting.
Personally, I think the answer is yes. Pen & paper balloting coupled with easy registration (present the proper ID at the polling place), extended voting times (make it a week, or a national holiday), independent exit polling (as a check against the system and to provide short term feedback) and human ballot counting would be preferable to any technological solution. IMHO, of course.