While we have been unsuccessful at reaching a consensus in HF17, I have looked at some data I want to share with you and propose that we remove the comment reward-pool idea and continue to live in the same ocean as 
pointed out earlier.
There are more comments than posts per day, but there are less comment-authors. The amount of accounts that comments may shock you, so let me just deliver it to you raw:
Today:
203 new accounts (2 mined)
6410 active accounts;
2056 posts (1068 authors);
7300 comments (426 authors);
234833 votes (5173 voters)
Yesterday:
263 new accounts (16 mined);
6613 active accounts;
2299 posts (1168 authors);
8157 comments (399 authors);
338873 votes (5342 voters)
So what I see here is that a very large number of comments are made by an incredible small amount of authors (about 20 comments per author) compared to the amount of posts created by a recognisable amount of authors (about 2 posts per author).
Questions to consider:
- Can you make a bot that can comment random compliments?
- Is it hard to make a bot that comment random compliments?
- Do you think someone right now have a random comment-bot?
- Did you know that upvote-bots for posts and comments are open source and very available to all steemians?
- Should we be willing to risk 38% of the daily blog-rewards on these numbers?
My personal opinion is that a few people make a lot of comments and a lot of people make a few posts. My witness is currently running Fork 17 with Fork 16 consensus. I will upgrade to Fork 17.1 when consensus is reached and we can move forward.
