The Economic Theory of HF21’s EIP

The EIP, or the Economic Improvement Proposal, that is part of Steem’s 21st hard fork seems to have been grossly under-explained to date. Day after day, I continue seeing all kinds of complaining and fearmongering, mostly based on one single protocol: the proposed 50/50 content rewards split between authors and curators.

What I’m not really seeing is an honest post from any of these fearful and sobbing users about the full set of protocols and how they can potentially impact user behavior. Instead, they mostly take current behavior and rewards in today’s system and extrapolate their new “data” under 50/50 distribution protocols. That kind of extrapolation – isolating one proposed variable and plugging it into a soon-to-be defunct set of protocols – is precisely how to not make predictions about future behavior and reward allocation...and certainly not a wise way to “inform” others or to make economic decisions.

So instead of talking about whether or not the proposed hard fork will be good or bad for some users over others, or isolating the protocols and formulating opinions based on that isolation, I’d like to offer my thoughts on the actual economic(-ish) theory for this set of protocols – how they may affect user behavior and the value of the blockchain and its currencies based on incentives and disincentives.

What are the protocols?

1 – Convergent linear rewards curve

This curve is a change from pure linearity and is designed to mitigate exploits at the lower end of the curve – where users can upvote themselves small amounts of rewards, mostly undetected, without corroborating votes from other users. The purpose would be to help prevent “leaks” in the system where spammers and other “abusers” or exploiters can reward themselves for low-value content due to today’s linearity in the allocation of the daily rewards pool.

2 – 50/50 content rewards distribution

This distribution change for content creators and curators moves the current 75/25 split in favor of authors to an even 50/50 split. It is designed to entice more staking of STEEM tokens and more participation in the system for content discovery and ranking.

3 – Separate and free downvote pool

This creates a new pool of free daily voting power at the equivalent amount of 25% of current voting power to be used only for downvoting. It is designed to encourage users to help with the ranking and rewarding of content by giving them a limited number of downvotes that will not cause losses in potential curation rewards. Once this daily downvote pool is consumed, additional downvotes will consume the normal amount of a user’s voting power.

How can these individual protocols affect user behavior?

We cannot possibly predict precisely how user behavior will change. But we can make some reasonable assumptions about behavior based on both incentives and disincentives and past and present behavior.

In my opinion, this is what we could possibly or even probably see from each protocol change.

Convergent Linear Reward Curve

Changing the reward allocation algorithm from linearity to convergent linear will mostly impact post payouts at the low end, which means that the least valued content will likely see less rewards due to being extremely unpopular. This would include a large amount of spam and plagiarism but may also include content that has not been shared or promoted by its creator, not otherwise seen by any meaningful number of people, or that is simply of a low quality as determined by the community.

Not rewarding low-value or non-marketed/unseen content is nothing outside the norm for social media. A reward curve that progressively rewards content creators as their content receives an increasing number or an increasing value of upvotes should entice creators to post higher quality content and to increase their promotion/networking efforts. Both of these outcomes would be beneficial for the content creator and the platform alike.

Those users who do not wish to compete (for viewers, voters, and rewards) or simply do not have the skill set to do so will likely not benefit from a convergent linear rewards curve. The lack of skills and/or the lack of effort will be commensurately rewarded – i.e., they will likely not be rewarded at all or will receive very small amounts.

From both a blockchain economic and content ranking perspective, this would be a positive result. Less rewards being “leaked” to low-quality content means more rewards will be available for higher-quality content. The inability or the lesser ability to earn profits from low-quality content should result in less of that content, which should make the remaining content more visible, further pushing it towards profitability.

50/50 Content Rewards Distribution

Increasing curation rewards by 100% (from 25% to 50% of content rewards) will likely entice more users to power up or remain powered up in order to maximize their risk-taking through staking STEEM tokens. As more tokens move from the liquid supply to the non-liquid, or as less tokens move from non-liquid to liquid, the total liquid supply of STEEM tokens will decrease and there should be less downward price pressure on the currency.

As curation rewards increase and author rewards decrease, and since curation rewards are paid out in STEEM Power and not liquid STEEM tokens like author rewards are, additional tokens will be distributed into the non-liquid supply instead of being readily available for selling on exchanges. As a result, we should see less downward price pressure on the STEEM currency.

With a larger pool of curation rewards, competition for those rewards should increase, possibly resulting in better overall curation. Finding the most undervalued content can greatly maximize curation rewards, which should encourage a meaningful percentage of curators to adjust their current habits and seek higher-quality content to reward with an upvote – and possibly even share that content with other consumers, further increasing their chance of earning more curation rewards.

Of course, some users will be content with their current habits and will continue curating as they have been, but they may very well lose out to more competitive users who can take full advantage of the gamification of curation rewards. As a result, those who do not adjust their behavior will likely receive a smaller piece of the curation pie.

Separate and Free Downvote Pool

A daily free pool of downvotes should entice more users to cast a limited number of free downvotes on content that they believe is over-rewarded, thus further contributing to the ranking of content based on their subjective preference. With these votes becoming available for all users, some of the lowest-quality content and some of the most “popular” and over-rewarded content will likely have at least some of their potential payouts reduced. As a result, more rewards will be returned back into the daily pool and will likely be reallocated to higher-quality and undervalued content.

It’s probable that some users will use these free downvotes to advance their crusades against other users that they simply don’t like. There’s nothing that can be done to stop these so-called “bad actors.” But every user will have the ability to cast these downvotes, so being a “bad actor” can result in such an actor attracting more downvotes for their own content. Downvotes and retaliatory downvotes occur today and, in general, they don’t have much of an impact on most of the user base.

As a whole, the free downvote pool should help to “normalize” downvoting and further improve content ranking and the allocation of rewards to generally higher-quality content.

How can the EIP as a whole affect the Steem ecosystem?

Now that we have a better idea of how each individual protocol can affect behavior, let’s see how they could affect behavior and rewards as a package. Again, this is only my opinion, but this is what we could possibly or probably see as a result of the EIP protocol changes.

The Combined Effects of the EIP

The convergent linear rewards curve should direct more rewards to higher-quality content and reduce rewards to lower-quality content. This would reduce the incentive to produce low-quality content, thereby reducing the number of overall content to be ranked and rewarded by curators. Those curators could then more easily discover higher-quality content and/or rank more content with both their upvotes and downvotes than they could have ranked previously.

The competition for the increased curation rewards should further help with the discovery of both high-quality content and new content creators. The rewards that are kept in the pool from not being allocated to low-quality posts due to the convergent linear reward curve and the rewards that are redistributed due to the free downvote pool should find their way to these higher-quality content creators and their content. As a result, those content creators with proficient social media skills should benefit from increased attention and a higher share of daily rewards.

As higher-quality content and users with better networking skills are pushed up the ranks, the most popular and “featured” content on the various Steem interfaces should attract more viewers and more interest from additional potential quality contributors to the ecosystem. In turn, this should attract at least some additional interest in the STEEM currency and entice more people to purchase STEEM and participate in the content rewards system, thereby increasing upward price pressure on STEEM.

Other possible effects of the protocol changes could include the following:

Reduced use of bid bots due to both the non-linear to linear threshold and the free downvote pool. Bid bot use would likely/mostly be used only by those with higher-quality content to promote that is less likely to be downvoted for being over-rewarded.
Delegations to bid bots and other automated services may be returned to the original stakeholders and used for manual curating.
Additional purchasing of STEEM may occur in order to take advantage of staking in SP and earning better returns from curation rewards, which would further increase upward price pressure on STEEM.

There are caveats about HF21 and the effects that the EIP could have.

If the hard fork only included the EIP, then most of the above would likely hold true in theory. Unfortunately, the Steem Proposal System (SPS) is also part of this hard fork.

With the SPS included, content rewards as a whole will be reduced by 10%, taking them from 75% of total STEEM inflation to 65%. This will reduce the total reward pool for both content creators and curators.

This will most likely reduce the impact of the economic changes in the EIP.

Because of that and because of the potential impact that it could have on STEEM distribution and price pressure, I am not in favor of the SPS at this time and I think it needs to be set aside until the economic protocols for Steem have been improved. There is an immediate need to correct broken protocols that have been detrimental to the system for over two years. There is not an immediate need to create a new distribution scheme for the Steem blockchain’s inflated tokens and the funding for the SPS can be reallocated from other pools.

We have also recently learned that a protocol change for the reverse auction of curation rewards will be included in HF21. This change will reduce the auction time from 15 minutes to one minute. This represents a 93% reduction in the reverse auction time, which is a significant change in curation game theory and will likely have an enormous impact on voting behavior.

I do not currently have a position on this change because I have not seen any curation data or statistics from previous times or from the previous change that reduced the reverse auction time by 50%. Although I’m inclined to support a reduced time, I cannot definitively say that one minute is the best option versus three minutes, five minutes, or any other number of minutes.

Because of the lack of data, any statistically-supported rationale for adjusting that protocol, and the protocol changes already included in HF21 which could be unintendedly impacted by it, I am not in favor of this significant change to the reverse auction and think it needs to be reserved for a future hard fork.

Conclusion

The EIP is not going to "save" Steem, but it could help stop some of the hemorrhaging. The underlying economic theory of this package of protocols is pretty sound, but there are still several Steem protocols that need to be addressed in order to further push behavior in the desired direction.

The fact that we even have the EIP as part of a proposed fork is a pretty big first step in the right direction. For the past two years, the broken economics of the Steem blockchain has been mostly ignored by the people who should have cared the most. I would like to see the EIP isolated as its own hard fork - prior to the SPS being proposed.

Increasing incentives for curators and seeking improvement in the content reward economics, which can positively impact both content creators and curators, and then reducing the overall rewards going to those creators and curators is completely contradictory from an incentive perspective - not to mention that the upward price pressure via the incentive to buy STEEM and power it up is canceled out by the downward price pressure via a new pool of rewards that are most likely to be cashed out to pay for "projects."

If we can implement the EIP and help reverse the trend of dissatisfied users and an ever-decreasing market cap ranking for STEEM, then the SPS may not even be necessary. There's no harm in waiting a few months to observe the impact of the economic changes in the EIP. If the results are insignificant or abysmal, then the SPS can be proposed.

As I have stated before - my priority as a Steem user and witness is to fix what's broken first, then try new additions and schemes. I think that's the most reasonable approach today, given the current circumstances.