Voting Abuse and Ineffective Curation: A proposal for blockchain-level change

Of late, there has been a lot of controversy on Steem regarding vote buying, purchase of Steem delegations, self-voting, etc. Actually, a lot of this isn’t really new to the Steem blockchain: similar arguments about fairness of reward distribution have been around since shortly after the first payout and people realized that the rewards were “real”. Another related complaint I’ve seen is that “good posts” aren’t rising to the top of the trending page and topic pages like they should (i.e. posts aren’t being curated effectively).

This last complaint is actually the most serious one, in my opinion, because one of Steem’s primary promises should be that it is good for content discovery (it’s not the only one: it also has great promise as a reward mechanism for software development, humanitarian efforts, etc, but it is certainly one of the original design goals as a social media platform).

Several changes have been made to the blockchain rules over time in order to attempt to improve the reward distribution, some that were helpful and some that were harmful, and currently there’s several proposals out there now for future changes. One of which I’ve seen recurring lately is some form of change to encourage downvoting (aka flagging).

Is a change in downvoting rules the answer to curation woes?

The idea behind most of these proposed changes to downvoting would be to encourage users to downvote posts that were self-voted (or voted up by a group of similar minded individuals) to reward levels that others consider unreasonable.

Under current rules, there are several economic and social problems with using downvotes in this way:

  1. downvotes cost the downvoter curation rewards that could otherwise be gained by using the voting power for an upvote (this was by design, to discourage casual or malicious downvoting),
  2. downvotes tend to make the person who got downvoted feel cheated about their lost rewards
  3. downvotes tend to create an adversarial relationship between the downvoter and the person being downvoted
  4. downvoting of a powerful poster or a group of such posters could result in retaliatory downvoting, which tends to severely discourage curators from downvoting posters with a lot of Steem Power.

I’ve read through various proposals that have been suggested for trying to alleviate these problems, none of which solve all the above problems, and some of which are difficult to implement on a blockchain (anonymous downvoting, for example, as a method of reducing retaliatory downvoting). So, in my opinion, most of the negative aspects of downvoting can’t be eliminated by changing downvoting rules, so I think changing downvoting rules won’t work well as a method of improving curation effectiveness and reward distribution on the blockchain.

Another option: Change the economic incentives to encourage “better” upvoting

Steem was designed to reward effective curators with higher curation rewards. Effective in this sense means curators who could quickly identify and upvote posts that others would likely upvote if they saw them. The idea is simple: early curators would upvote a good post, raising it up on the hot list to bring it to the attention of more readers/voters, creating a cascading effect of increased attention (and rewards) for good posts. Similar voting systems exist on other social platforms, of course, but one of the advantages of Steem was the additional of a financial incentive for good curation.

However, there’s a competing financial incentive that currently effectively counters the financial benefit to effective curation: the current blockchain rules favor self-voting over effective curation. To understand why this is the case, a little history is needed:

Bot curation and introduction of the “30 minute rule”

Originally, Steem was designed to create a 50/50 split of rewards between authors and curators. For example, on any given post, 50% of the rewards would go to the author of the post, and 50% of the rewards would go to the people who upvoted the post. The 50% awarded to the curators was and is determined by two factors: the order in which they upvoted (early upvoters get more of the curation rewards) and the amount of voting power they upvoted with.

But early on, bots began to vote on posts by popular authors immediately after the post was created in order to grab the lion share of the curation rewards. To counter this tactic, the “30 minute rule” was introduced into the blockchain code in an attempt to level the playing field against early bot voting. The 30 minute rule imposed a declining penalty on curators who vote in the first 30 minutes after the post is created. If you vote on a post immediately after it is posted, you lose all your curation rewards. If you vote at the 15 minute mark, you lose half your curation rewards. Any vote made after 30 minutes gets it’s full curation rewards.

30 minutes was chosen somewhat arbitrarily as the time it would take to read a post. 30 minutes might sound like a long time to read a post on Steemit, but if you’ve read any of the posts made by the guy who came up with this idea (and you probably have, if you’ve been on this platform for long), then you’ll understand why he set it to 30 minutes. In fairness, as one of my longer posts, this post may take 30 minutes to fully read as well.

There’s another aspect to the 30 minute rule which I believe has seriously skewed the economics of steem in favor of self-voting and pay-for-vote bots: the curation rewards lost by early voters was given to the authors. Why? Well, I can’t say for sure, but the same guy who came up with this rule was a prolific poster, and I guess he saw it as a way to allow curators to reward their posters with extra rewards. But it’s had a big unintended consequence: an author can upvote his post early both directly and with voting bots, and shift a large portion of the rewards from the curators to himself. And because many Steemians don’t understand the 30 minute rule in it’s entirety, it’s probably not a conscious decision on their part to give up their curation rewards in this way. And sitting around waiting 30 minutes just to upvote a post is probably asking too much of them anyways. But worst of all, the 30 minute rules makes pay-for-vote bots a logical economic alternative for posters that aren’t invested into the long term value of the blockchain regardless of the quality of the post they create.

I think pay-for-vote bots can have their place in Steem’s economy: it makes sense to me that a relatively unknown author may want to get his post upvoted to gain attention of curators if he’s confident that his post will attract votes when it gets curator attention. This could increase the effectiveness of Steem at creating a quality trending page as it would allow new authors to offset some of the advantage of established authors who are already being followed and routinely upvoted. But I don’t think it should be profitable to pay for an upvote if other curators don’t see the quality of your post and follow through with upvotes of their own. But the 30 minute rule makes this economically possible today and a result we have near contentless posts being upvoted by pay-for-vote bots.

My thoughts: Change the 30 minute Rule to 5 minutes and restore 50/50 rewards

On an intuitive economic level, if we agree that curation on Steem isn’t currently working well, it seems that one obvious potential solution is to reward curators better for curating better while at the same time reducing the incentive for authors to self-vote bad posts. There are many ways this could be done, of course, and I’m only outlining my initial thoughts on one way to do this without introducing unnecessary complexity into the existing blockchain code:

Replace the 30 minute period with a 5 minute period

5 minutes is enough for a competent curator to normally determine if he likes a post, in my opinion. Let’s change the period to something that fits the time to evaluate most posts, not just the longest and most complex ones.

Let curators keep their full normal reward and eliminate the huge reward for early self-voting

This could be achieved simply by discarding the 30 minute rule entirely instead of changing it to a 5 minute period, but then we would be left with the same problem it was meant to prevent: immediate voting on new posts by bots. So instead of eliminating the rule entirely, I propose that we put curators who upvote in the first five minutes on the same reward footing, eliminating the advantage of an immediate voter.

Now this change still leaves bots able to selectively upvote at the end of the 5 minute period based on the number of votes already received on the post, but that’s far less abusive than handing out larger rewards to immediate upvoter bots and I think it’s hard to even say if the first 5 minutes of voting is going to tally with the 7 day performance of the post, which could invalidate this method of bot gaming entirely. I suspect it’s likely to be no better a strategy than the current one employed by “blind curation” bots that simply upvote popular authors near the 30 minute mark.

Does this solve all Steem’s curation problems?

Well, I don’t know and since it’s a complicated issue, almost certainly not, but of proposal’s I’ve seen so far, it’s one that I think is relatively easy to implement, and I think its benefits versus the current rules are pretty clear. At a minimum, I think it will seriously deter the ability of pay-for-vote bots to reward bad posts.

The biggest argument I can see levied against this proposal will be that content creators will get less rewards than they do now. This is indisputable. But, I think the better question is: on average, will “good” content creators get less than they do now? I think it could actually go in the opposite direction, with more good content creators getting more, while bad content creators get less: after all that’s part of what I mean by increasing curation effectiveness. If the distribution of author rewards shifts enough between good and bad authors, it’s an overall win for good content curators.

This post is mainly intended as a spring board for discussion of curation improvement ideas that focus on changing curation rewards instead of changing the downvote system, since it seems to me this has been neglected as a solution to the curation problem. Your feedback and proposals on this topic are more than welcome!

H2
H3
H4
3 columns
2 columns
1 column
822 Comments