Sort:  

He gives a formula of the form: n^2 / (n + 1)
But in the example he uses: n^2 / (n/5 + 1)

My reading was that the post was a broad discussion of the general premise of CLC curves. If the final formula for the HF is confirmed then that would be useful to know.

Oops, you're right! I missed that bait and switch, where one formula was used for introduction, a different formula to demonstrate examples, and then switched back to the first for the conclusion.

The actual formula they implemented is ( rshares + s ) * ( rshares + s ) - s * s / ( rshares + 4 * s )

It's a bit more complex than explained in the deep-dive, and I don't feel qualified to explain it properly.

Thanks Joseph, that's helpful!

I had a check on github and it looks like "s" may be 2,000,000,000,000 rshares. Working through the formula on that basis gives a curve that:

  • Starts at 50% of linear (rather than 0% - this is a welcome change!)
  • But takes quite a long time to move towards approximating linear.

So a HF20 $0.10 upvote would:

  • initially add $0.05 to a post / comment under HF21.
  • be worth $0.08 ("80% consensus") around $6 payout.
  • be worth $0.09 ("90% consensus") around $16 payout.

Although there is also some scaling to take into account (since the lower rewards on smaller value posts would be spread across all posts over time).

Coin Marketplace

STEEM 0.30
TRX 0.12
JST 0.032
BTC 58773.92
ETH 2988.08
USDT 1.00
SBD 3.74