I'm not sure about this.
1. There isn't a real correlation between length and strength in terms of comments.
For instance, this comment: "So, I wasted so much time on this. Not that it was a waste of time, but I finished it at once, chose love over will, and accidentally clicked forward...good thing I saved at the maze. I had to start over from there, but better there than the beginning. But, anyway, it was a fun read...sure I wanted to get my grammar Nazi gun ready at the excessive (and inappropriate)apostrophes, but no big deal. Also, despite the fact that the sex scenes were a bit silly (as if a 14 year old boy could last 20 minutes...the average is about 5) it made me sad to see the romance not work. As for the maze?...I cheated. :) But hey, I don't think it was a bad idea..."
is not as good as this comment: "An absolutely amazing piece of storytelling. The only problem (and it isn't one at all really, in my eyes) is that it is very linear, and cliché, but clichés become such because they are such good ideas. There is easily, however, enough unique ideas in the story that make any negatives infinitely small compared to the elegance of the story. I could see this actually being publishable, and would pay for this, knowing how good it is."
2. The author shouldn't necessarily have a say in which comments are the best. A community vote is appropriate, but having author's choosing could simply lead to the most positive comments being consistently labeled as the top comments. From there, anyone seeking top comments (for the EXP points), will just write positive comments. If it's community choice (as in the comments which most represent the voice of the community), then the incentive will be to constructively speak one's mind, rather than appeal to the author.