New to Busy?

SBI in an HF21 World - Part 2: Convergent Linear & Downvote Pools


8 months agoBusy8 min read


Much has already been said about HF21 and the EIP by other Steemians. We may have been premature about the level of consensus; maybe it will not actually happen. Initially I was in favor of the EIP, but the deeper I dig the more I think that it's a distraction that won't have any of the intended effects.

Last week we focused on the impact from the change to 50/50 curation rewards. Each of our value streams already passes through curation rewards one way or another, and will continue to do so with minimal impact on total value members receive over time.

SBI Logo design by @roldamn

In today's post, we explore convergent linear rewards and downvoting pools. We talk about the role of personal responsibility in a decentralized world, and about why we don't think the changes will have any positive impact.

What is Quality?

The most unpredictable impact of HF21 is how large stakeholders will manage their new downvote pools, and surviving in the new ecosystem will require a higher level of individual awareness about how much ‘quality’ your own material has.

Quality is a highly subjective characteristic, not an inherent attribute. The reward protocol of Steem is designed around 'Proof of Brain' ... the idea that the curation incentives will encourage the rewarding of quality content, and people that 'prove their brain' receive the most rewards from the pool.

The inherent flaw to this logic is that everybody perceives quality differently. There are 'cult classic' TV series with strong followings that were cut in their first season due to low viewership. Did they lack quality or did it just take too long for people to realize the quality was there all along.

With respect to blockchain... quality is defined and codified as the content that attracts the most stake-weighted votes.

The Downvote Pools

People decry the breakdown in proof of brain because they disagree with the quality of the most-rewarded content, but they don't cast dissenting votes either.

Dissenting votes are a critical part to establishing consensus on what material is 'quality' and should be rewarded. The EIP tries to resolve this by removing the economic 'opportunity costs' of casting dissenting votes. Even with the economic cost removed, the social costs (risks of retaliation and reputational damage) will remain just as high.

Whether consensus is improved by a rise in dissenting votes depends as much on a removal of the social costs as it does on removal of the economic costs. Downvotes are extremely unpopular and frequently described as 'stealing rewards from content' - instead of as mere votes against the existing consensus rewards level. This makes it unlikely that the social costs of downvoting will be removed along with the economic costs.

Removing the economic costs will result in substantially more downvoting from those willing to bear the social costs (people with well-established reputations that will be assumed to be in the right) and people that don't have social costs (people with well-established 'bad' reputations that have nothing to lose).

Graphics by @katysavage

Convergent Linear

Along with downvote pools and 50/50 curation, the third piece of the puzzle is convergent linear. Based on the initial parameters that I have seen in the github (the release is not packaged yet, so this may be totally different if it actually comes out), post payouts will start at roughly 1/2 their current levels until they reach about $0.10. From there, they will converge upward toward linear, though linear is never quite reached.

This means that whatever your normal vote values are, you can always add 'extra value' to your existing payouts by tacking on additional votes... even if those votes otherwise come at break-even (no built-in profit).

Convergent Linear and SBI

What does this mean for Steem Basic Income? While your payouts per rshares value received may be consistently lower under the EIP, actual change in your rewards will be amplified by your relative positioning.

The actual impact on your upvote values will depend on the payout of your post relative to all other posts.

Since the median payout is below the $0.10 threshold mentioned, just getting onto the slope will be enough for each additional vote to amplify the value of your existing votes and reduce the impact of the EIP on your post payouts.

You may receive less total rewards on each post, but you need Steem Basic Income (and any other promotion strategies you follow) even more: to get your post onto the upward sloping curve and amplify the value of all the organic votes that you receive.

(SBI Token design by @thekittygirl)

Personal Responsibility

Here is where the downvote pools come back into the picture. The only check on buying every possible vote you can for every post is the free downvote pools. If downvoting becomes accepted and prevalent, then promoting your content too high up the curve will result in downvotes that negate the boost you received from climbing up the curve.

It is the personal responsibility of every Steemian to evaluate the quality of their content with one simple question: Will this post attract more upvotes than downvotes? If you believe that you will attract more stake-weighted downvotes than upvotes, then posting will damage your reputation, and you probably should do it.

If you over-promote the post, then attracting more eyeballs will tip the scales and also result in more downvotes than upvotes. If downvotes are only made by those that don't care about social costs, then you have to be even more careful - because people that think you're 'abusing the reward pool' (by their own standards) will not just downvote a single post, they may attempt to destroy your account.

Summary Conclusion

With 50/50, we were able to easily compare the current mechanics with the new and identify that the rshares value that Steem Basic Income delivers to each post will modify in a way that the member share of those rshares is constant or even improves.

With Convergent Linear and downvote pools, we are no longer looking at the rshares that you receive, but at the conversion of those rshares into rewards. The potential outcomes range from a worst-case scenario of "same as now only worse" to "better than ever before". The results depend on how optimistic you are about people's responses, along with which outcomes you would consider to be better or worse (also quality values judgments). What's clear is that the actual outcome is completely unpredictable.

While I initially supported the proposal, the deeper analysis I have made leading up to this post has convinced me that the likelihood of it improving anything that people claim it is intended to improve is extremely low. The EIP is a red herring, a distraction from more important development, an experiment in bad tokenomics.

If it does happen, the need for regular Steem Basic Income support on your content will be even greater than before, as a sufficient level of support will give you greater resilience to the prospect of attack.

It's not the conclusion that I expected to write, and I hope that if the EIP happens the best in people overcomes the broken incentives and Steem becomes a little more civil. I am not as optimistic as I was last week after writing part one.


If you want to join, or to increase the level of basic income to which you are subscribed, enrollment is pretty straightforward:

Just send 1 STEEM to @steembasicincome. Include the name of a Steemian to sponsor in the transaction memo (preceded by @). You and the person you sponsor will each receive 1 unit in the program. You can sponsor any active Steemian, it does not have to be a current member.

If you're unclear, please check out our full transaction memo guidelines and then let us know if you have any questions.

The official currency for enrollment is STEEM. If we choose to accept SBD without refunding it, you do not receive extra value for it.

Enrollments are processed automatically every 144 minutes.

Questions about Steem Basic Income?

Please read our recently published FAQ. Most questions are addressed in our FAQ or in the additional resources that it suggests. If you still have questions, ask in the comments section or join us in our discord channel.


Sort byBest