For lo, it did come upon this day that the ICZN did look upon electronic-only publication, and say that it was good. Provided that it met the following requirements:
(1) the published work must be registered with ZooBank prior to publication, with provision in the ZooBank record of a permanent archive in which the the published work will be held,
(2) the published work must include evidence of registration, such as the ZooBank registration number, and
(3) the published work must be associated with either an ISSN or ISBN number, so it must be a formal 'book' or 'journal'.
You can read further details here. As is standard for new ICZN requirements, they are not intended to be retroactive, and online-only names from before the new rules were introduced are still invalid.
I don't think that many people will be surprised by this (while you complain about the cost of prescriptions... Mike Taylor is looking very smug). As I've discussed before, I reluctantly came to accept that electronic publication would have to be allowed after I realised that most non-taxonomists (and many actual taxonomists) were finding it increasingly difficult to distinguish what was 'published' from what was not. So what does this mean in practice?
It became clear at an early date that successful management of electronic publication was going to require registration. In the past, most people were talking about registration of names, but the ICZN has decided to go with registration of the actual publication. This avoids any hiccups such as human error leading to some of the names in a publication being 'published' while others are 'unpublished'. I can see potential complications arising with the requirement that registration happen before publication, but ZooKeys has been publishing articles with ZooBank registration numbers for taxa included for a while now, so it would seem that the requirement is not insurmountable. Of course, there would have been potential problems had the requirement been registration after publication. It should be noted that, even though the published work has to include evidence of registration, there is a certain allowance for error, so long as it is clear that the work has indeed been registered (so, for instance, if a transcript error meant that the wrong registration number was included in the publication, that does not automatically invalidate the publication).
The requirement that the publication has either an ISSN or ISBN number means that names cannot be just published carelessly. If I refer to the features of an unpublished species in a blog post, I will not be accidentally 'publishing' that species and potentially confusing the paper trail (electronic trail, in this case). Note also that, while the ZooBank record of a published work must include the ISSN or ISBN, the work itself doesn't directly have to. So, for instance, the ISSN of a journal does not have to be included in every individual article.
One of the biggest concerns raised about allowing electronic publication is that it will make it even easier for would-be taxonomic 'vandals' to ply their irritating trade. The registration and ISSN/ISBN requirements, as well as making it clear what is intended for publication and what is meant to be just an online communication, are intended to impede such behaviour. They won't stop it entirely (as I've noted a few times before, that would be effectively impossible) but they do provide a couple of hoops that must be jumped through.
So all that remains to be said is: Let the games commence!
In the matter of Walter Lewin, MIT goes medieval
2 minutes ago in The Curious Wavefunction