You are viewing a single comment's thread from:

RE: All the different reward curves explained in an easy to understand way. And how they effect you.

in #exyle5 years ago (edited)

It will all depend on the parameter C and K: n^2/(n/C + K)

When they find reasonable parameter values, so that a dophin goes into the linear range, than the changes are not that bad.

Sort:  

Thanks for this info, man! That would make a huge difference for sure.

@exyle @holger80 the curve is even more specific:

https://github.com/steemit/steem/commit/e99f97de5a0e9c72b15172b623391bee2049cee4

(n^2 + 2cn)/(n + 4c)

There's a reason for this specific form that has to do with the relation the curation curve has to specify and specific implementation choices, and from the thread they will put out that derivation at some point (there's some in my post as well.)

Basically the slope doesn't start at zero, so actually this is a lot milder than we think it will be. Which I think is good for those thinking it's a bit too much of a penalty to stomach, but not for those that want to see a dramatic effect on the lower end.

I think that it's enough to incentivize stacking votes vs low end reward farming. (Comparing two sets of post farms, if one farms at x per post and the other goes slightly higher, the one farming higher will be getting more) and that can cause cascading effects for how they adjust, even with that.

Thanks so much much for the extra info man!

I plotted the graph as best I could and depending on the value of c it changes a bit.

The initial start is now between linear and n^2.

I recorded the change in a video.



for c=1 (blue line).


Screenshot 2019-06-01 at 18.14.12.png


Have I understood you correctly?

Yup that looks to be right. There's also this interactive tool someone pointed out to me: https://www.desmos.com/calculator/weerzfsi7b which allows anyone to play around with it too (can put more curves simultaneously as well, I believe)

nice! And guess what tool I used :)

LOL right, I feel dumb.

Think it makes more sense to divide by n in order to easily grasp the impact. And to use a logarithmic scale for the x axis.

  • green: lineair
  • red: n^2
  • blue: convergent linear as proposed (with some c)
  • orange: alternative n x S^log(n) curve proposed here (by me)

I hope, whatever curve and whatever parameters are chosen, they will also shift dust level accordingly. Massive amounts of comments seem to be about to shift beneath dust level as things look now, and I'm afraid this will really hurt social interaction on the platform.

This is what things would look like with dust level cut-off added:

Coin Marketplace

STEEM 0.27
TRX 0.11
JST 0.031
BTC 71054.38
ETH 3862.54
USDT 1.00
SBD 3.52