Curation: Now and to come


The upcoming Hardfork 25 will change the curation scale to a far simpler model than it was, which is summed up in @blocktrades' recent update post regarding development work.

The new curation reward algorithm works as follows:

  • first day (24 hours) linear rewards (equal weight to all voters in that window)
  • second window (24 hours to 72 hours/3 days) with reward weight/2
  • remaining votes in 3rd window with reward weight/8

Under the new algorithm, anyone voting with the first 24 hours of the post receives the same proportional rewards. In other words, for any given voting strength, the voter will get the same percentage return-on-investment as any other voter during that period.

Voters voting during the second and third window receive a smaller proportional curation reward (and voters who voted during the first 24 hour period receive a little more reward when voters vote during the 2nd or 3rd window). Note that if no one votes during the first window, then 2nd window voters will receive the same amount of curation as if they had voted during the first window.

This creates a far greater parity in opportunity between manual curators and automated curation, as "frontrunning" is far less likely. This change might not seem like it will do a lot, but I think there is going to be quite a difference in how it plays out, as there are a couple of factors that will heavily increase the value of curation for manual curators.

From the current curation model, there are two factors that will change heavily - firstly is the most obvious to have an effect, which is the removal of the curved stacking of votes to enhance curation return into the three time buckets. This means that all of those early bot voters that don't read a post, will have no curation advantage over those who come in within the first 24 hours, and those in the next two days will have no advantage over anyone else in the same bucket etc.

The second reason is harder to evaluate.

Currently, there is a "cost" to voting early, which will be removed in the upcoming hardfork. This is a straight line system that returns value to the pool when voting earlier than the 5 minute period. It looks like this.

If voting at:

@ First second - 100% curation returned to pool
@ 1 minute - 80%
@ 2 minutes - 60%
@ 3 minutes - 40%
@ 4 minutes - 20%
@ 5 minutes and after - 0%

While the author will get 50% always What this means for the curator is, if they were to vote at 2.5 minutes and would have got a 10 HIVE curation return, they would instead get a 5 HIVE curation return and the other 5 HIVE will return into the rewards pool for distribution. This might not sound like a lot, but it can add up fast and can actually be around 20% of the total value of the curation portion of the post.

So in a hypothetical post with a 20% curation return (10% of total post value) that was paying out 100 HIVE in total, The author will get 50 HIVE and the curators will have 50 HIVE distributed among them, minus the effects of the lost HIVE - meaning that 10 HIVE would return to the rewards pool for redistribution.

If you then imagine that 10 HIVE will get redistributed to one more post under the same conditions, the next author would get 5 HIVE, the curators 4 and 1 HIVE would go back into the pool again. This I think means that authors will actually get 55% of the rewards, while curators 45% - all things remaining equal for the scenario across these two posts, which they generally aren't.

We can see this illustrated here on a post of mine from a few days ago:


Author payout is in blue - 50%
Curator payout in green - 32.13%
Total payout returned to pool - 17.87% (that is 35% of the curation total)

However, because of the current stacking mechanism where those who vote early benefit from those who vote late, it is possible to make up for the loss of the "curation return to the pool" function by frontrunning other voters, especially on posts that are expected to get voted highly late. This means that autovoters are able to position their vote in an optimal point to benefit from those voting after and undercutting them to reduce other voter curation rewards, but get a multiplier that makes up for the loss of the return to pool costs.

While there is game in this, it heavily favors the people who are autovoting early and those who are able to evaluate conditions and place votes with precision - which isn't the average person by far. The new curation mechanism should affect this problem by firstly taking a lot of the game out of the curation in stacking and then not returning curation HIVE back into the pool. What this does is gives the curators the full 50% of the curation return, only affected by which time bucket they voted in and overall - this will be far better for curators as their full vote will have an effect where they want to curate, rather than putting it back into the pool for redistribution.

This means that no matter the size of the account voting, curation will be more valuable all round, because no one will be undercut from maximizing frontrunners and the entire curation amount will be paid out. The difference in these two factors might be far more significant than first thought and should also give a level playing field, meaning that while the autovoting frontrunner maximizers will see a decrease in their curation returns, the manual voters here daily will see an increase in theirs.

I think this will also have an effect on what kinds of posts get highly voted, since there are one or two very large voter groups who only vote on posts (often low quality) with no votes on them to maximize their curation return, since they do not have to worry about the HIVE convergent curve costs. This means that quality posts might end up attracting more value overall, since there is no reason not to vote on them, even if they have already attracted voters.

And then, the convergent curve removal in the next hardfork comes into curation play in this also, as there will be equal incentive to vote on comments, as the vote value will be the same regardless of how much HIVE is going to be paid out, meaning that there will be equal curation and equal author rewards as voting on a post that has already been curated or, hasn't been discovered yet.

For someone like me who has always voted comments, this works out as a significant difference, since about 20-40% of my vote values go out to comments, which means I have been "losing" curation since they do not pass the convergent curve. However, it isn't only that - because they don't pass the curve, the comment authors are also not getting the full weighting of my vote.

After Hardfork 25, I suspect that comments I vote on will receive more value than today, plus I will also earn more curation from them. This is good for everyone who is active commenting and voting manually and could mean for many people around a 20% increase in the value of their interactions, while a clear drop off on autovoting maximization return will happen, with the difference plus the return to pool going directly into the wallets of curators.

While I completely understand why these mechanisms were originally introduced in the EIP, I think that this simplification will bring with it a lot more incentive to stake HIVE also, as the curation return becomes far more predictable, without having to think much about when or what to vote. Instead, people can just vote on what they enjoy and still get the benefits of the curation without having to miss out. There will be a few people out there who are maximizing their curation that will get a reduction, but all in all, most of the truly active stake on the platform will get a better result overall, as they will no longer have to compete with the maximizers with bots that they are very unlikely to beat, let alone beat consistently.

However, there will also be more incentive for abuse, so I think that people will have to also come to terms with understanding that as owners of the network it is our responsibility to keep our streets as clean as we can too. Every change has pros and cons, though many people want the upside without ever having to personally deal with the down.

Anyway, perhaps this is useful for people who might not understand curation to get a basis of where we are now and what is to come - but I think that in general, there is going to be a lot of benefit in HF25 for the platform of active users and hopefully, we will be better incentivizing the curation of good posts and encouraging far more interaction than we have seen over the last few years.

We will know how it plays out in the weeks after HF25, which is a couple months away. Should be interesting to see how disruptive it is and, how good it is for those who are manually active on Hive :)

[ Gen1: Hive ]