I very much like the idea of reading journal articles using e-readers. They are easy on the eyes, and allow you to naturally keep and organize articles digitally instead of keeping giant stacks of paper or file folders around. The downsides are that most of them have screens that are smaller than normal sheets of paper, and also that they can only display static grayscale. So we give up what could be one of the greatest advantages to modern publishing - the easy inclusion of videos and animations. Perhaps soon technology will provide a unity of both benefits.
As to allowing comments on articles - not sure if you had in mind just during the peer review process, or always. I do think that one of the greatest benefits to internet publishing is that if you read something you disagree with or believe is wrong, you can tell the author and any other readers so immediately. (Indeed, how many blog posts have we seen appended with "commenter ___ kindly points out that...") This seems less applicable to most science articles, since we hope that we are producing verifiable truth, and because the audience is much smaller. (If you have a question or problem with an article, you're much more likely to just email the author.)