Either we are talking about different things or your math is wrong.
If we are just talking about the amount of rewards that a single user with his level of SP can reward to himself or his friends with his vote alone (no other significant votes) then he has more power under n^2.
You are right, 40 to 10 did change things. Even if you rolled back both changes though, he would still be able to extract more with 40 n^2 votes spread across 40 posts/comments than he can today with the 10 using linear.
I understand the argument of n^2. Under any system with moderately even distribution of stake, you are right about n^2, and it is a perfect system to combat this type of abuse. The problem is that with the distribution of STEEM, there is such a distance between the larger stakeholders and everyone else, that individual stakeholders can with their single vote essentially reach the tipping point or critical mass at which point their vote squared significantly surpasses the votes of a majority of the other voters and actually pulls more rewards, even though it is technically just ‘one vote’.
We are talking about the same thing and my math is alright.
There are simply too many parameters to give you any example numbers.
Did you not notice, how selfvoting only became a massive problem after hf19 ?
Even with the disproportional distribution STEEM has, this is not necessarily true.
First of all: If any other post had twice the votes, it would get 4 times the shares in rewards.
The effect of flags would change significantly, too.
A shareholder with half the SP could 'steal' more than half the payout of those abusive posts.
It would also have some indirect effects; As it would influence voting behavior.
There has been a dramatic increase in self voting since Hf 19. For the average user (even someone such as myself with a decently large amount of SP) our votes are worth more under linear rewards. For those of us on our end of the spectrum (most users) who want to use their voting power to reward themselves, self voting is more profitable under linear rewards when compared to n^2. That is why we have seen such an increase since the switch to linear rewards. I agree with you on that.
In the interest of not just repeating the same arguments over and over, since you don’t seem to agree with my point, maybe let’s just end with the above statement - since I know on that at least, we agree.
I will have the last word then:
My initial statement still stands strong.
I invite you to read my post on n^2.
I did not tag it #witness-category, although I could and probably should have.
I have read it.