Thanks for this interesting article.
I would prefer a curve which started as n^2 / exponential (thus flat), and then later changed into linear which would work against self-voting as well as excessive rewards.
@clayop had a similar idea.
@yabapmatt: Overall I don't think tweaking the reward curve or curation % is worth focusing our time and efforts on. The vast majority of users don't understand any of this stuff and I don't feel like changing it will change their behavior or encourage more people to invest.
If a reward curve worked against self-voting whales and excessive rewards of single posts that would be already a huge progress, then it wouldn't matter if the majority of users changed its voting behavior or not.
5 minutes curation window.
...
3 minutes curation window.
I never really understood the sense of that curation window: bots can be programmed to upvote after 30, 15, 5, 3 or 0 minutes after a post was published anyway, so why should a curation window decrease the advantage of automated vote bots?
I can see what you are getting at with N^2 to begin with and reverting to linear at value 'z' for example. This way 'good' curating could earn more than selfvoting and post rewards won't blow out. I have a few questions regarding this.
What should 'z' be? Getting this right would be difficult and I can imagine would require trial and error.
Would the optimal 'z' value change as the price of Steem increased? In theory it shouldn't, as value in Steem remains the same (post value in dollars increases) but we are relying on human perception, which is often askew.
What happens with early whale upvotes that shift the curve straight to linear?
How would you be able to explain how this curve functions to the average user?
Auto-upvoting accounts favoured by particular whales will still be a problem which is the case with any curve that is not linear.
In my discussion with @eonwarp (under the here discussed article) I already explained why I am not really interested to elaborate the details of my idea. It was just to bring the idea itself into the game. If witnesses find it to be interesting they should care about implementing it (they are getting paid for that). :)
Many equations are possible ... And as I wrote, by using spline interpolation the curve could look more beautiful than my example graph.
I think the average user should not care and just vote as he always does. The idea should work against upvoting own posts or comments with much STEEM power (as long as not many other people are upvoting as well).