Math of Steem: Where does the profit from voting bots come from?

So a couple of weeks ago @yabapmatt made a post about voting bots and @abh12345 asked a question that I thought was interesting: what if everyone started buying votes? He characterized his issue thusly:

Profits are there to be had by everyone in the chain: delegator - bot owner - user (sometimes). The reward pool, and thus the blogs of the non-users suffer though.

I had some short thoughts at the time, but it stuck with me. There is a whole lot of profit in the vote-buying system, and figuring out where it all comes from took a little while. I knew that "it comes from the reward pool" was wrong: moving pieces of the reward pool (RShares) between users cannot create profit. Vote-buying or vote-trading in and of itself is fundamentally mathematically neutral. So where does the profit come from? Four primary places, listed in order of how easy they are to understand, not necessarily how important they are.

1. The cost of powering up

Users pay a cost by committing to keep their money in the system for a certain period of time by staking Steem as SP, and in return when the currency inflates, the new currency generated goes to those users in proportion to their stake. This is the fundamental basis of all Proof-of-Stake systems, and while it might be novel to you it's not particularly new; government bonds operate on a very similar methodology.

On Steem the new currency generated goes to users in RShares distributed as voting power. This is the profit from the stake, not from vote-selling. Part of what this means is that, in spite of perception, vote-sellers aren't making profit when they sell their votes. They are, generally, selling on their RShares for less than their nominal value, and taking a loss in return for not having to put effort into using those RShares themselves. It's possible to sell those RShares at a profit at some times through some bid-bots, but that profit comes directly from a loss taken by the vote-buyer.

It is not possible for both seller and buyer to make a profit on the vote sale. It only sort of looks that way because of the way Steem obfuscates the staking rewards. The staker makes a profit from staking and then is selling some of that profit to others at a discount.

2. Investor Confidence

This is tied into #1, because if you think Steem is going to go up in price and are going to hold it anyway, the external cost of locking that Steem up in a stake goes down. That means that the actual cost paid in lending the system money lowers while the absolute return remains the same, increasing ROI. That doesn't matter for the vote sale itself but affects the stakers' willingness to discount their rewards.

Investors who perceive themselves as having a higher ROI will be more willing to discount some of it in order to receive it as cash.

3. The Broken SBD Peg

But you say that's responsible for just about everything!

Well, it kind of is. If something weird is happening around here, look to the broken SBD peg first, because it's probably doing something unexpected. In this case, when SBD is above $1, it inflates the value of all votes substantially: by a percentage of (SBD - 1)/2.

With SBD at $1.76, votes are worth 38% more in real money than the system thinks they should be. This is entirely profit put into the system by traders who are overvaluing SBD, and it's responsible for a lot of the extra money floating around. It scales with the price, so puts us in an interesting spot where some of the biggest self-voters and vote-buyers and middlemen with the most capital ought to be devoting some of it to pumping SBD. Since there are far fewer SBD than Steem out there the price is very vulnerable to manipulation, but I don't think they've noticed.

(In practice the effective percentage is somewhat lower because of curation not paying out 50/50. 38% is the instant-vote value.)

4. Steemit Inc.

The @steemit account, as well as several related ones like @steem and @ned, decline a lot of their rightful stake reward RShares by allowing their accounts to sit at 100% voting power much of the time, not voting. When this happens RShares that would have been rewarded to them for their stake simply aren't; this inflates the vote values of everyone else in proportion to their stake. Since @steemit holds 44 million SP, this is kind of a big deal. Between them these accounts hold at least 25% of all of the staked SP, and the total value held in inactive accounts might be as high as 40%. This is reciprocal: sitting on 25% of SP makes everyone else 33% more important in the voting pool, and sitting on 40% of SP makes everyone else 67% more important.

With Steemit not using their stake, the value of everyone's votes is increased by at least 33%. This is profit being injected into the system by Steemit Inc. Crucially, this multiplies with the profit from the broken SBD peg. We're beginning to see why there's a lot of money floating around. Even if we use the 25% of SP number the combined effect is a 84% increase in the real value of everyone's vote. If we use the 40% number it increases to 130%. It's interesting that somewhere in the middle there is the immediate ROI of self-voting.

So what does that mean?

I'm not necessarily willing to draw a lot of conclusions from this just yet. But it looks to me like if the system were working efficiently in a mechanical sense, with SBD at $1 and everyone voting their stake, self-voting and its cousin vote-buying would be very close to simply distributing the RShares at their nominal value. Any profit in buying votes would be tied directly to the effort saved their original owner in not spending their RShares, and vote-selling would mostly be used at its most human-useful and system-useful points: allowing people to continue using their stake when for life reasons they weren't able to vote it.

However, if SBD stays high and @steemit keeps staking and doing nothing, there's going to continue to be a ton of profit available for someone. It will doubtless spread out in the system depending on vote supply vs. demand and competition between the middlemen. We're onboarding a lot more customers and almost never a new whale, so expect the price of votes to keep going up.

Epilogue

All of this, of course, comes down to the core question of Steem, which is what all of this conflict ends up being about at its base: do you feel you're getting value for your RShares when using them to vote on other people? You're getting less direct financial return for your stake than the people who are using theirs to self-vote or selling them. But if you feel like the non-financial value of your voting is worth the nominal value of your RShares, then you're actually making more than the vote-sellers are.

If you feel like the value of your voting isn't worth the nominal value of your RShares, then you should probably be doing something with your money other than staking it here and refusing to sell your vote. You're losing value just by being here.

Personally, most of the time I feel like the value of my voting on other people is worth more to me than the value of my RShares. That's part of why I like buying RShares from other people at a discount, so that I will have more of them to use later. The times my vote isn't worth more it's my fault for not having built up my reading list sufficiently and managed my voting power sufficiently well. (Although if you really want to change something, please, post more on Wednesdays.)