Thread by @el33th4xor

FEI dropped down to $0.136. In the process, it should have taught everyone a few lessons about stablecoin design and, perhaps, crypto investing.

A thread.

FEI/TRIBE was a two-coin algorithmic stablecoin, with a twist. The twist was flawed from the start and it should have been possible to predict that this idea would not work.

In a typical two-coin algorithmic stablecoin, you have one coin, $FEI, trying to maintain the peg, while the other one is used absorb the volatility. We wrote about this structure in our stablecoin taxonomy paper.

arxiv.org/abs/1910.10098

Algorithmic stablecoins work very well when the demand for the coin is so high to be over the peg: you just mint more FEI to bring the price down to the peg. The actual challenge lies in what to do when demand is lagging and price is low.

At some point, demand might be so low that the price goes far below the intended peg, rendering the coin useless. We call this zone outside the "feasibility envelope." In essence, the coin has diverged so far from the peg that it has failed.

Now let's examine the FEI twist. The thinking goes like this: Why do stablecoins go lower than their peg? Because people sell. So let's punish people who sell FEI when it's trading below the peg!

The lower the price, the greater the punishment, aka penalty. If you sell FEI when it's trading below its peg, the recipient receives fewer FEI than you transferred, taking some of the transferred coins out of circulation. The thinking is that this should boost the price.

But now, instead of just doing the obvious, single step of logical inference ("let's penalize transfers when below the peg"), let's look ahead a bit.

Every sale has a buyer and a seller. If you punish the sellers, you're also punishing the buyers.

The penalty mechanism in FEI not only makes the supply disappear, it also makes demand disappear. It punishes both sides, and thereby **narrows the feasibility envelope for the coin**.

Suppose we would consider a regular stablecoin to have failed when it's trading at, say, $0.50 instead of $1. FEI is designed such that it, effectively after the penalty, trades at $0.50 when the actual price is only slightly off of $1.

The challenge with an algorithmic stablecoin is to come up with force fields that cause the pegged coin to tend towards the peg. Create more supply when the price is high, and create more demand when it's low. $FEI's twist damages the supply AND demand when the price is low.

Making the effective price lower when the price is low is exactly the opposite of what you want in an algorithmic stablecoin!

The design of algorithmic stablecoins is very much uncharted territory, scientifically, and so far, this was a fascinating experiment by a well-intentioned and good team. What happens next depends on what the contracts allow, but there are countless ways to recover from here.

Let me add that I expect the coin price to recover, but I don't expect the coin to be viable, as is, in the long term. There are two reasons for this.

First, the team can boost the price back up simply by buying their own coins. They can do this as long as they are capitalized, exactly like a central bank. But knowing the game will have to end at some point, users will not trust the stability of the coin.

Second, if the coin is widely perceived to have failed, then people will sell with the penalty and exit at pennies on the dollar. The mechanism will actually work! But at what cost?

Users who exit at such a huge loss will probably not come back. The game theory essentially turns into a game of chicken where two people in two different cars drive towards each other, and the first to steer away loses.

Users are currently reticent to sell, because they believe there are other options (e.g. option 1). When the price is down in the dregs and only when they believe that the system is out of options, will they take the penalties.

The exit sales will restore the price. But those people will likely not come back. Just as important, the token as it is right now has such a narrow feasibility envelope -- transfers essentially stop whenever demand is under the peg -- that no one in their right mind will use it.

For long term success, it feels like some on-the-fly coin engineering is necessary to change the dynamics and game theory. But it has been an amazing experiment.