That might be outdated shortly, the main problem seems to be acceptance of online identities.
No, it is not that. (In fact, the companies that sell digital signatures would surely love internet voting.)
The problem is ensuring that voters can vote according to their free choices without fear of punishment - by government at all levels, employers, mafias, family members, etc...
Icelanders may not feel that that is a problem now, but it is a serious and real problem in many parts of the world, including many parts of Brazil. Indeed, putting an end to such "leash votes" was the main excuse the governmetn used when it pushed electronic voting in 1996. Even where it is not a problem now in your country, it may become one when some fascist group becomes strong enough to scare voters, or when a president turns out to be corrupt -- and by then it will be too late to switch to a safer system.
To prevent voter coercion, the system must ensure that no one can know how another person voted. More than that: the system must make sure that no voter can prove to others that he voted in a particular way. At the same time, it must ensure that even skeptics -- especially the losing parties -- can be convinced that the votes have been correctly added.
It is not trivial, but also not too difficult, to ensure these requirements with traditional paper voting. It is more difficult, but still possible, to ensure them with properly designed hybrid paper/digital systems. It is mathematically impossible to get all three with purely digital systems, including voting by internet. The latter obviously fails the first requirement: one's vote can be snooped either at the terminal end (by the husband/boss/mafioso in person), in transit, or at the receiving end (by corrupt government officials).
I recall a complicated cryptographic internet voting proposal that claimed to protect voting secrecy from corrupt government officials (thus only part of the problem). It almost worked -- except that one could still identify how each person voted by capturing the received data and processing it offline, simulating an election closure after each vote cast and comparing the totals.
I have learned many interesting stories about electronic voting but they are too long and too off-topic to tell here...