Number 2 is a question I’ve had myself. I did some sample calculations a few days ago, using visual inspection to approximate values for the new reward curve and the median payout from these two graphs:

^{The new reward curve, depicted in red, from @steemitblog's post}

^{Median payout per post, from @arcange's 7/19 stats post}

This includes a reduction of the Content portion of the inflation pool from 75% to 65%, which lowers it to 87% of what it is now, the change in the reward curve, and reward distribution between authors and curators.

Median Payout

According to the stats from 7/19, the median payout has recently been between 0.15 and 0.2 SBD, so I called it 0.175 SBD. At an internal median price of ~0.24 that gives 0.729 in STEEM.

The total payout on the median post goes down to 56% of what it is now.

Authors of posts by the median will earn 38% of what they would today while the people who vote for them will earn 113% of what they once did.

What Happens to a Haejin Payout

Posts paying out over 16 STEEM (~4 STU today? I'm not sure how to calculate STU exactly and translate this into the payouts listed on most frontends) will pay out more. I took a specific example from a predictable earner to see what that looks like. This post had a total payout of ~87.970 STEEM, well over the 16 STEEM line.

The total payout here drops down to 98% of what it was.

He (?) takes a hit and only brings in 65% of the author rewards while those who curate his posts double their rewards. Since this is a large self-voting example, a portion of that curation goes back to haejin's accounts.

For the same post haejin, ranchorelaxo, and starjuno together earned 18.701 SP in curation rewards, 85% of the curation for that post. Assuming the same distribution of rewards to curators we get this:

So haejin takes a 6% hit on his draw from the rewards pool and it's now doubly advantageous to front-run his self-votes.

For a self-voter of a much lower amount, like the median payout, their self-curation would bring in a small boost. Assuming they also brought in 85% of their curation rewards, instead of earning only 38% of today’s author rewards they would earn 54% of today's rewards in total.

Is this correct?

I dunno. This is my understanding of the various changes to reward calculations. I'm not sure why the curve is shown with the payout in STEEM, since the payout is calculated in Rshares, or how any changes to the price have shifted it since it was posted.