Blurt Brainstorm: The Economic Landscape, A Personal Perspective

I'm starting this new series of posts using the #brainstorm tag and encourage other members to contribute their ideas. I recall that Steem once had a wealth of talented individuals who shared their own branch of expert knowledge to discussions about the construction of the whole system. That seemed to fall away as it became obvious that all decisions were taken behind closed doors and, most importantly, away from articulate dissenting voices. I hope we can stimulate such discussions on Blurt.

One part of brainstorming is to think the unthinkable, to indulge in blue sky thinking and play out all the "what if" scenarios. But every silver lining comes with a cloud, so we also need some devil's advocates. Even ideas can be stress-tested with logic and a few hammers.

Let us not forget that Blurt's code is open source, so anybody can look inside to see how the magic actually works. There's a great film about this issue: The Prestige. The revelation is not so much in knowing where all the strings are, it is in watching what happens when you pull them.

We can use many analogies, but an economic system behaves most like a force field - like an electromagnetic field or even a terrestrial landscape with hills and valleys, deep wells and billowing geysers. So what are the forces that shape our economic landscape? What are the actions a user can perform to change his immediate environment and hence affect the whole system?

Economic Force Fields

Here it is worth noting that any economy has two main points of view: the local and the global, better known as the microeconomy and the macroeconomy. I think Steem's macroeconomic system is very resilient, with feedback loops between the reward pool and activity (recent claims) having been tested under some seriously stressful conditions! Blurt has inherited much of this dynamic while also trying to simplify the underlying machinery.

I have witnessed three major changes in Steem's so-called reward curve and none of them - none of them - have seriously affected the vote yield that a user can generate. There have been short-term changes after hardforks, but what I found incredible is that they were actually short-term. Sure, rewards may have sucked for some user groups more than others, but from a macro perspective all those changes were mitigated by the dynamic response of the reward pool. There are a few issues that I think will make the Blurt macroeconomy even more responsive and I shall describe those in further posts.

However, what most users experience is the microeconomics - the effects on their own account, vote values and earnings. I hope my articles will show that such local effects are a result of, and reflect back onto, the broader macro parameters. But worth noticing one thing now: that all the mathematical machinery in the background is there to support one thing - votes.

That's it! All the discussions and arguments about reward curves and curation splits and earnings and upvotes and downvotes and coin price and... well, anything, are constrained by one thing: votes. Somehow, the expectation is that we can resolve every issue by messing around with one of those levers. But every change in algorithm affects everybody all the time. And yet, people expect the system to make subtle distinctions in behaviour when the only available action is a vote.

I feel we need more functions that users themselves can access and hence help them guide the economy. We already have one important extra action available: voting for witnesses. Without delving into the issues surrounding DPOS here, I feel we need more such levers at the user level. I once heard so much about the "gamification" of Steem, yet nothing happened. I shall go into what could be available in further brainstorming posts.

Action and Reaction and Complexity

Now, let's move on to expectations, actions and reactions; one example shall suffice for now. We've heard many times the following argument: "Users are earning too much, inflation is too high, therefore let's decrease post rewards!" So what happens if we actually do that? Well, we initially see author reward drop because fewer rshares are created per vote, hence fewer recent claims are created. This then means that the reward pool starts to increase, and this then also increases the amount of rewards distributed per claim. Depending on the formula, it only takes 2 or 3 weeks for the macroeconomy to be back exactly where it was before such changes.

This is what I mean about investigating second-order effects - and sometimes even third-order consequences. The reward pool acts like a carburettor in a car, adjusting the flow of coins into the economic engine based on changes in activity. However, a carb has no idea if the car is actually moving or not! Similarly, I would like to see some feedback inserted into Blurt so that the reward pool reacts to both changes in activity and changes in the vested BP that generates that activity.

So, the macroeconomic machine in Blurt, as in Steem, is already pretty good at managing the stability of the whole system. What it currently lacks is the micro-machinery to manage effectively complex human behaviour. Now, complexity is a subtle concept; it does not mean chaos or disorder or making a mess of things. Complexity can emerge from very simple systems - so long as they are allowed to interact. As an aside, this is a huge change in how mathematics itself is done. Horrendous algebraic monstrosities have been replaced with computational algorithms and a computer. We can now iterate towards solutions that previously yielded mere approximations - see the above image. Those iterations are what every user experiences as part of the whole Blurt system.

The one and only example I can see within the Steem system of complex behaviour arising from simple formulas is that short voting window known as the "reverse auction". Within that shrinking period hide two algorithms: the curation rewards curve and the "curation loss" line. One curve is essentially a square-root while the other is linear with a negative gradient. It looks, on paper, to be not very interesting, but add people and there develops a battle to maximise curation rewards. The resulting graph of curation yields is surprising and fascinating; change one vote and the whole graph changes. That brief window has created a microcosm of complexity that I have found nowhere else within the code.

This is an area I wish to investigate further: adding simple user functions that, with feedbacks spreading through the community, can start to correspond to human actions. Running a central bank is far easier than constructing a functioning algorithmic economy. Most people seem to have been brainwashed into monetarist theories without even noticing it: inflation rises, increase interest rates; inflation falls, decrease them. We are now seeing the catastrophic consequences of such naive policies.

The point I wish to make here is that economists have it easy because they rely on humans to react and interact. Within a cryptoeconomy, such interactions have to be encoded, they don't just happen - no code, no action.

All we need are experts in mathematics, economics, psychology and coding. Easy!

---

images: wikimedia, European Space Agency (ESA)

---