There is no real "tip" of the curve, it is gradual. Anyone placing a vote with the intent of rewarding the content benefits from additional votes added on top of it because of the continual convergence to linear.
The idea of this curve is to heavily penalize microvotes which attempt to slip under the radar with a million comments each earning 0.03. It probably won't have much effect (though still a little) on larger payouts including most paid votes.
It probably won't have much effect (though still a little) on larger payouts including most paid votes.
This is the primary point I was trying to make. As for the 'tip' of the curve... using this curve from @vandeberg's post, the point I marked in red was the point I was thinking of...
While the curve itself is gradual, this is the vote level where rewards with n^2 and linear cross. I don't know what his scale is, but if the 0.5 shown represents a STEEM value of vote, then I think it's way too high.
Yeah I agree we will need to assess some actual curve if and when it is proposed. We may disagree on the need for some sort of inhibiting of small-ish payouts on the low end. I do think they are necessary, even if not ideal from a fairness or interaction point of views. Like most things it is a matter of tradeoffs.
Well, if the trade off you substantial stakeholders consider a-ok eliminate any interest new users have to come here, I'll be looking for the next platform where I can speak freely you didn't kill yet.
More pressure on retention is suicidal.
If I'm not mistaken, the intersection of where n and n^2 cross has no bearing on the significance of that 0.5 point on the X axis on the n^2/(n+1) curve.
Recall that these are graphs of the derivatives, not the curves themselves. The point you've marked there is roughly where the derivative of n^2/(n+1) would have a linear tangent. In other words, it's where the second derivative of n^2/(n+1) roughly equals to 1, which has no real relevance that I can see.
I agree with Smooth that the proposed n^2/(n+1) curve just as a superlinear head and approximates to linear. The effects are to ward off profitable micro voting which happens already, but would likely increase in an otherwise new working economic scheme (higher curation, some free downvotes) because other avenues of abuse would be more noticeable and thus less profitable.
Ideally, I'd probably prefer just a 'spam tax' of say 30-50% up until Steem value is around 1, and then completely normal linear after that. But if that can't be implemented, then I'm ok with whatever convergent linear approximation curve.
Ideally, I'd probably prefer just a 'spam tax' of say 30-50% up until Steem value is around 1, and then completely normal linear after that.
If the goal is to kill off the remaining steem user-base, then I would say that's a great idea. The average post payout is only a little more than 1 SBD (about 3 STEEM), but strongly skewed by the tiny percent of posts with high upvotes.
The median post payout is closer to $0.08, well below your 'spam' range. Calling the posts by more than 50% of all Steemians 'spam' would be a disaster.
Source: @arcange's Steem Statistics
Almost all of them are spam, though. There's currently no incentive to put in any more than minimal effort.
Yes these figures are true. In practice, my suspicion is the convergent linear curve would actually hit the lower payout posts harder than a 30-50% spam tax, but of course that depends on the numbers.
If we can stop most profitable spam behavior by making low payouts accept 70c to the dollar, it's not a terrible trade off. Keep in mind that when other measures are implemented, self voters and vote selling posts would likely be hit by downvotes and would consider elsewhere. A deterrent to an influx of micro spam votes in order to continuing reaping ones own voting rewards should be considered as this could be really bad for RC and blockchain bloat.
Actual figures can be adjusted. It's really about forcing all profitable behavior into the light and there's no real way to do it without some trade offs.
The median post payout is closer to $0.08, well below your 'spam' range. Calling the posts by more than 50% of all Steemians 'spam' would be a disaster.
Likewise, the median number could be strongly tilted to the low end due to the high levels of pre existing spam.
"...making low payouts accept 70c to the dollar, it's not a terrible trade off."
You clearly speak for yourself, selfvoter. Regressive taxes that benefit the wealthy at the expense of the poor are always terrible trade offs. The more I read your comments, the more I observe the dichotomy between your rhetoric and your actions.
We have an awful problem with retention now, and this would horribly worsen it.
But even if its intent is to cull comment spam farming, it utterly destroys rewards for organic interaction between small accounts. When people make good, legitimate comments on my posts I like to hit them with a 25% vote and reply. That gives them about .03 which is just above the dust threshold which already penalizes small accounts. Likewise most accounts posts see very small rewards of a few cents on their posts, but they are legitimate, organic, and quality content. To cripple all these actions and shunt that value more toward whale votes & bid bot voted content is a negative. We’d effectively be creating a system where the majority of accounts are more likely to see decreased or even zero rewards, as they never see a large payout on an individual post.
I think we need to see actual numbers and a specific curve before we can judge this. There will be some cases where small interactions are inhibited but there are already some cases where small accounts can't by themselves generate a reward large enough to actually pay out. These are all tradeoffs.
One thing to consider is that if massive reward-mining operations are curtailed even somewhat, there will be a lot more Steem left in the reward pool and the sorts of small interactions you describe may therefore generate larger rewards, despite being somewhat inhibited by a curve. There are two parts of any reward calculation, one being the curve and the other being the total size of the reward pool and what else is pulling from it.