People give Steem value by working together. The more everyone contributes the larger the pie we share grows. The core idea is that everyone should get a slice of the pie proportional to their contribution.
It would be nice if everyone would be content receiving a fair share, but the reality is that many people wish to maximize their take. These are the people that show up to a pot-luck with enough food for one person, but eat enough food for two or more.
A whale is someone who makes waves in the market every time they move. An Evil Whale is someone whose negative actions dramatically reduce the value of the whole pie.
Demonizing Doesn’t Help
The first instinct is to blame those who work the rules to their advantage. After all, they should be more respectful. Getting upset at this behavior doesn’t hurt the abusers, it just causes you to suffer.
We must design systems that minimize the incentive to be selfish and give the honest individuals the power to police. Anyone playing by the rules has rightfully earned what they receive. The problem is the rules, not the people.
Vesting Stake
One of the most significant factors Steem employs to keep people from abusing the system is Steem Power. Any widespread abuse will cause the abuser to lose more value due to capital losses than they earn from their abuse.
The impact of capital losses does more to keep large whales in line than small minnows. The actions of a individual minnow are unlikely to significantly devalue Steem as a whole, so the profits on their selfish actions can be high.
Prisoners Dilemma
The minnows are in a prisoners dilemma. If everyone defects then everyone loses, but if only one person defects then they might earn a small undeserved profit.
The value of Consensus
Steem minimizes the ability of minnows to cheat by distributing rewards proportional to n2. This means minnows have far less influence relative to their weight and whales have far more influence. This means that minnows only have influence when they work together.
By requiring people to work together to gain influence we bias rewards toward common causes that grow the pie. It doesn’t make sense to collude with others if you must share the reward. It will lower the profitability of defecting and make it more likely to get noticed and countered by others.
A whale is effectively the same as a large group of minnows colluding together. In this case the whale can act more efficiently and doesn’t have to worry about defectors.
The larger the group of whales and minnows agreeing to a particular payout the less likely that the payout is harmful to the pie and the more likely it is beneficial.
Good Whales
Good Whales act in ways that maximize the value of their capital. The side effect of their actions is to increase the value of everyone else’s capital at the same time. These individuals are motivated to counteract any abuse they see. The larger the whale, the more incentive they have to be a good whale.
So if a group of minnows get together to milk the system, a single whale can come along and eat them up. The more effective they are at policing abuse the less incentive there is to attempt abuse in the first place.
Good Whales are the primary defense against Evil Whales.
Cannot run and nowhere to Hide
Due to Vesting Stake evil whales cannot run and their actions cannot be hidden. This means that any evil reputation they earn will stick. Once the good whales learn who the evil whales are, it is trivial to write a bot that will counteract every attempt to profit at the expense of the whole.
Giving Good Whales the Advantage
A down vote is fundamentally selfless. There is no profit for the voter unless the down vote increases the value of their capital. The system as a whole needs to be designed to ensure that those wishing to down vote get the last say.
In other words, good whales running bots to police the activities of evil whales need to have the last say. We have systematically designed Steem to maintain this behavior in every aspect except curation rewards.
Curation Rewards
The current curation reward system is designed to encourage consensus. It makes it profitable to vote with others and less profitable to vote elsewhere. The result of this system is to increase the contrast between posts with high payouts and everything else. It serves to accelerate the process of reaching consensus without respect to the consequences of that consensus.
Not voting on something everyone else is voting on is a selfless act. The whale who abstains from piling on with other whales is losing a guaranteed profit opportunity.
An evil whale will join any pile. They will take any actions that maximize their reward even if it harms the whole.
In some cases an evil whale is motivated to be the first to pile on. This motivates them to predict what will be popular to maximize their return. On the other hand, their motivation isn’t to find good content, but to predict where everyone will pile on next. This can devolve into a self-fulfilling prophesy.
To maximize rewards these evil whales aim to devise algorithms that collude with other evil whales.
Impossible to Police Evil Curation
The only way to police evil curation is to down vote a post. This isn’t targeted policing. There is no way to negate the evil whale’s curation behavior completely without harming others as a byproduct. This means that there is a hole in the rules of our game that enables some people to walk away with more than they contributed.
Small Losses are Acceptable
Every business owner knows they will lose some merchandise to shoplifters. Every customer knows the prices they pay are high enough to cover the cost of the losses. Neither customers nor business owners would profit by adding TSA level checkpoints in an attempt to cut all losses.
There is a law of diminishing returns. It costs exponentially more for each additional bit of loss prevention. At some point the cost of preventing losses is higher than the benefits. This is the point at which shop owners buy insurance or simply write it off.
Steem is no different. The cost of preventing abuse is complete loss of decentralization and complex rules that reduce engagement and kill the project. At some point we must accept that we cannot prevent all evil. We just need to make sure that the amount of evil we tolerate doesn’t prevent Steem from growing.
Magnitude Matters
If 100% of all rewards were distributed via a flawed system, then it could devalue the entire platform; however, if just 1% of rewards are distributed by the same algorithm then any misallocations can be tolerated.
The trick is to identify the proper balance between incentives and the risk of abuse.
Conclusion
As the largest whale, Steemit has the greatest incentive to be a good whale. We have more to lose by tolerating abuse than anyone else. Any changes we propose are selfishly designed to maximize the value of the platform which benefits everyone. Some people may suggest that we make changes to increase our slice of the pie, but any rational actor would be foolish to want a larger piece of a smaller pie. We would be quite happy to have 1% of a trillion dollar pie.
There are some people that are upset about potential reductions in their pending curation rewards. We know that the only people that stand to lose from the proposed changes are the reasonably large whales, especially those who have written bots designed to maximize their profits. I encourage every whale to look at the bigger picture and realize that even though you may end up with a smaller slice of the pie, any change we propose is designed to grow the value of the pie.
So lets keep the debate about what rules will maximize the value of the pie. Then we can let the “chips fall where they may”.