RE: Taking a closer look at the EIP 2e12 Reward Curve
Nice analysis! I have been working on something similar, although using historical data for the rewards distribution to work out how the curve is scaled up.
I think you have a discrepancy between figure 1 and figure 5 though? You have scaled up the curve in figure 1 so that it crosses the axis at 16 Steem, suggesting that the impact on payout vs linear is zero at 16 Steem. However I don't think that you've scaled up the rewards in figure 5, as the difference in payout at 16 Steem is 2 Steem, rather than zero, in this latter chart.
One of my initial concerns, when I heard about the 2312 curve, was that users would need to buy votes to get their posts on the apparently linear portion of the rewards curve. This is unlikely to occur as the difference between linear rewards and 2e12 curve rewards is at its greatest at that point on the curve.
I think the vote-selling position is complex. The CLRC will encourage aggregation. It will make economic sense for an individual to sell their votes (or to delegate) to someone whose posts have higher payouts, rather than to self-vote their own lower paying posts.
However, since the reward curve is continuously increasing, it would make even more sense to sell / delegate to someone with even higher payouts.
This leads to two initial conclusions (I think!):
(1) The relative value of your SP (in terms of the price you can achieve for delegation) will increase the more you hold of it, in line with the CLRC.
(2) Assuming the vote-purchase price is fixed by the market, it will be (more) profitable to buy votes on larger earning posts, but potentially loss-making for smaller earning posts.
Potentially the vote-selling price point could be around the linear break-even - which I think is broadly the conclusion you came to. Although I think there could be a competitive aggregation element.
To prevent this type of behaviour, users will need to use their downvote mana pool.
Agreed. Downvotes will play a large part in influencing this behaviour.
Thanks for your response @miniature-tiger. Figures 1 and 5 are looking at slightly different things. Figure 1 shows the point that upvotes on a post are worth the same as the linear rewards curve. Whereas, Figure 5 shows the point where total payout of a post catches up with the total payout of a post upvoted using a linear rewards. As the early upvotes are worth less with the 2e12 curve, it takes sometime for the upvotes to lift the value of the post to what it would be if we used linear rewards curve. Based on the assumptions I have used, this might not be until the post reaches 54 Steem. It is possible this could be sooner or later depending on how people vote. The shapes of the curves change based on voting behaviour that's what makes them difficult to map using Steem. I guess that is probably why Steemit Inc. discuss the formula in terms of reward shares. Unfortunately, the majority of the users do not understand reward shares and an explanation of the effects in Steem is considerably more useful even if the exact numbers are impossible to predict.
I agree with your two conclusions. Your second conclusion is probably more concerning. Some users may choose to bid bot there posts to extremely high values to make as much as possible. It will be interesting to see how the bots respond to the new curve as well as all the other changes. I think it will be very difficult to calculate an accurate ROI as it will vary from post to post.