a) a cubic curation reward curve, so discoverers are more effectively incentivized.
@twinner suggested a sigmoid function. It would make self-votes on 'empty' posts/comments less attractive, but at the same time would prevent extremely high rewards on posts where 'everybody' is placing his upvotes (different than the n^2 reward curve).
b) diminishing returns on voting for the same authors over and over again.
I am happy to read this elsewhere finally. I was suggesting it some time ago, but the article was commented mainly by minnows. I described the core idea like this:
"How about if after each vote on a specific account (including ones own account) each further vote on the same account would lead to significantly less curation reward for the voter and less profit for the upvoted account? Thus, when upvoting an account which I had already upvoted before, my voting power would be smaller than in case I upvote an account which I didn't upvote before."
I think diminishing returns should apply for downvotes as well to make 'personal battles' less attractive and use them instead for envisaged reasons.
Of course the details were still to be discussed, for example how strong the returns should diminish and how long the timescale would be for any specific user to recover again.
To be clear, I was talking about the curation reward curve, not the author reward curve. This used to calculate how much of an advantage early voters have.
As for author reward curve, I wouldn't support a sigmoid function. Among other issues, it would kill demand for holding a lot of SP and discourage top authors (and maybe curators). High activity and SP holding would incur a very steep tax, in effect. Linear is OK - not perfect, but the problems lie elsewhere. Once these other issues are fixed, linear offers the best combination of liberty and equality. I would certainly support stricter bandwidth limitations - that would effectively minimise the spamming problem. A brand new account starts with 37 MB - way excessive, in my opinion. Oh, and an overhauled Rep system.
I like some the ideas in your post though, hope people see it now.
Yes, you are right, we were talking about different things here ...
Is that so obvious? After a flat start (to make self-votes less attractive) it would rise rather steep actually ... only the end would be shallow again ...
I have to admit I didn't think about it in detail yet, but the idea looked interesting to me at a first glance (maybe one could use a sigmoid function which is not that steep in the middle ...).
The problem of the linear curve is that self-votes (even in case no other user votes) have a rather strong effect.
As a matter of fact, self votes by whales have a much smaller effect. It's true that the effect is stronger for minnows, but given the massive disparity in distribution, linear means less rewards are allocated to possible self votes.
The real culprit is the vote regen change from 40 > 10. That means a possible 100% self vote is 4x as powerful as before.
To believe that I needed an example. If the curve starts flat, of course also whales should profit less when upvoting an own article on which nobody else has voted so far.
Yes, I mentioned this in my linked article, too. It makes self-votes much more lucrative than before - together with the linear reward curve.
Addition: I also think the former idea to have only four fully rewarded articles per day was reasonable as it made it less attractive to make many short low quality posts per day just to upvote them oneself.
I was comparing linear to the quadratic curve we had before. I see now that you meant sigmoid. Rewards are calculated by total Rshares contributed. Depending on the exact implementation, a megawhale might just get to the peak all by themselves with a sigmoid curve. Don't underestimate the disparity between whales and minnows - it's enormous. Anyway, it's not possible to talk about details like that without knowing the exact implementation.
I have to say this sigmoid function is 'fascinating' me as it could avoid the disadvantages of the linear and the quadratic curve. Therefore I wanted to attract some attention for this idea and see if some other steemians might be interested, too. :)
Of course the exact implementation is what finally would matter, but before to rack my brain too much, I wanted to plumb if the idea is able to attract some interest.
Anyway: thanks for drawing the attention on diminishing returns again!
This drop from 40 to 10 votes, and the drop in VP changing from 0.5% to 2%, are due to one single parameter doing two jobs within the core code; this is why the product remains constant. It is not possible to change this without splitting that parameter into two separate ones.
Good thinking
Diminishing rewards for the same author over and over again makes a ton of sense.