EIP: On fixing the broken down-votes/flags/reputation system for STEEM

We used to have flags on steemit, now we have down-votes. But nothing has changed, really, they are the same thing. The change though is meant, apparently, to entice users to become more involved in the curation process and to get rid of the idea that down-voting post is an act of aggression, a bit of psychology to make clear down-votes are meant as a curation tool that balances up votes made for profit rather than curation.

eip

As it turns out, changing the name from flag to down-vote is only the first step Steemit Inc envisions for making users use the feature to improve the quality of the curated content. As outlined in this post, Steemit Inc gives its public support to a proposal by @trafalgar that includes the idea of creating a separate down-vote pool.

The proposal, an "Economic Improvement Proposal" has sparked quite some debate. Many (including me) believe the proposal, if implemented will backfire and lead to damage to the platform and its economy due to the fact that the proposal reduces incentive to spent time writing good content, hurts platform growth by unfairly hitting new accounts, but worst of all, incentifies passive stake holders to act in a way least beneficial to the active stakeholder, content oriented economy and platform.

In this post I won't be looking at the curator reward or the convergent-linear reward curve parts of the proposal (for those interested, I've outlined here hat I feel is wrong with a convergent-linear reward curve, and what alternative would likely achieve what the convergent-linear reward curve aims to achieve without the damage), just the part about down the down vote pool and more general, the idea that broader use of down votes will help the platform.

Passive stake holders

To understand how changes to the blockchain's economic model could influence behavior by large passive stakeholders, we first need to look at the money making options that passive stake holders have, and how changes would affect the incentive to opt for one or another.

In this poll, I asked what behavior of passive stake holders would be most beneficial to the platform.

While opinions differ greatly on what behavior by passive stake holders would be most desirable, there was one striking outcome that stood out: There was one option that no one voted for:

• Running (or delegating to) bid bots

I think we can conclude that any EIP that actually works will have to create incentive to passive stake holders to do something else, anything else with their stake than running or delegating to a bid bot.

more down-votes (as is)

If we look at the incentives created by increasing down votes, we see it creates a strong disincentive to opt for any of these two options:

• Actual self-votes on automated posts
• Actual self-votes on comments on their own posts

In doing so, it creates an incentive to opt for one of the other options:

• Algorithmic curation (good)
• Running (or delegating to) bid bots (bad)

While giving little to no incentive to opt for:

• Powering down and selling their stake (good)

That means, with the ease and returns involved in delegating to a bid bot, we are creating a situation where the worst behavior option is incentified by a measure meant to improve the platform. Needless to say this isn't the most desired outcome.

75/25 up vote, 25/75 down vote

I think we can do better by inverting the incentive model for down votes. Doing so, would, I believe, make a separate down-vote pool a good idea. Consider the following idea:

• When doing an up vote, 75% (or slightly lower) of positive RSHAREs go into an author bucket, 25% (or shightly higher) of RSHARES go into a curator bucket.
• When doing a down vote, 75% (or slightly lower) of negative RSHARES go into the currator bucket, 25% of negative RSHAREs go into the author bucket.
• If at pay-out time, the curation bucket is negative, it is filled up to zero with whatever needs to be taken from the author bucket to do so.

So what would this do, and how would this be better? Well, consider an account with a 100MVEST stake selling votes. When this account upvotes crap at 100%, 25% of the vote RSHAREs will go into the curation bucket, ready to flow back into his account. Now consider a 20MVEST account wanting to aid in improving platform curation and countering the upvote with her 100% downvote. With the current numbers, doing so will hit the bid-bot user hardest, while hardly hitting the bid bot owner. Changing the curation share would help a bit, but that would also hit every good content provider unfairly. But now look at what happens with our new rules.

Let us look, assuming a current ROI of about 100% on bid-bot usage, how the different scenario's work out.

Without down votes:

• payment == 75% of vote value.
• curation == 25% of vote value
• ROI for buyer: 100%
• ROI for seller (compared to self-vote): 100%

With down-vote now:

• ROI for buyer: 80%
• ROI for seller: 95%

With down-vote and 75/25 25/75 buckets:

• ROI for buyer: 93.33%
• ROI for seller: 85%

As we see, where a separate down-vote pool creates incentive to move from self up voting to investing in bid bots, combined with the bucket approach that lets down votes hit curation first, the direction will likely get inverted, and bid bot investments, above all other behavior gets deincentified.