By the end of this you will know what actually holds a stablecoin at one dollar, which kinds exist and which kind died, and how a blind chain ever learns a real-world price.
Back in checkpoint 2 you met stablecoins as the calm species, the dollars that live on-chain. You have used them since without asking the obvious question: why does one of them stay worth exactly one dollar while the coin next to it drops forty percent in a weekend?
It is not magic and it is not a promise on a website. There is a mechanism underneath, and once you can see it you can also see exactly where it can break. We open the panel on this one too.
Here is the whole engine on one board, before we name any of it. The issuer holds real dollars in reserve, one for every coin it has handed out. And it keeps a door open: anyone can hand back a coin and take a real dollar, or hand over a dollar and take a coin.
That open door is the trick. If the coin ever slips to 98 cents, traders rush to buy it cheap and redeem it for a full dollar, pocketing the gap, until the buying pushes it back to one. The peg is held by people chasing two free cents, not by anyone's good intentions.
Stablecoins are not one thing. They are three, and they differ entirely in what sits behind the coin. The first keeps real dollars in a bank, one per coin. The second, like DAI, locks up crypto worth more than the coins it issues, so even if that crypto dips there is a cushion, and code sells it before the backing runs thin.
The third kept no reserve at all. It promised to hold a dollar using only clever code and incentives, with nothing real underneath. That is the species worth looking at closely, because you have already watched it die.
You met this one already. In checkpoint 4, Terra's UST was the stable dollar with nobody in the payer slot, kept upright by code and incentives instead of a reserve. In May 2022 confidence cracked, and because nothing real sat underneath, the peg had no floor to land on. It fell all the way, and tens of billions evaporated in days.
That is the difference this board draws. A backed coin can fail, but only by losing its actual backing first. An algorithmic one had no backing to lose, so a wobble in belief was enough. Same event you saw from the yield side, now from the peg side.
So a backed coin is the safe kind. But safe is not the same as untouchable, and the reason is almost boring: the reserve is real dollars, and real dollars have to sit somewhere. They sit in banks.
Which means a backed stablecoin is only as solid as the places its reserve is parked. Most of the time you never think about that. Then one weekend, you have to.
In March 2023, USDC, one of the most trusted dollar coins, slipped below a dollar. The cause was not crypto at all. A bank holding part of its reserve had failed, and nobody was sure, over one long weekend, whether all of those reserve dollars would come back.
The price drifted below one because the open redemption door had effectively jammed: with the bank shut and the weekend closed, traders could not be certain they could redeem one coin for one full dollar. So let us stand in that weekend and read it.
Now flip the panel around. To keep a peg, or to sell crypto when a loan goes underwater like in checkpoint 4, a contract needs to know a real-world price. So ask one: what is ETH worth right now? It cannot tell you.
A contract is sealed inside the ledger. It can read its own storage, the memory you saw in checkpoint 12, with perfect certainty. But the price on an exchange, the weather, who won an election, none of that exists inside the chain. The machine is brilliant and completely blind to the outside world.
Here is the exact line a contract cannot cross. On the left are facts that live inside the ledger: pool balances, owners, the swap formula from checkpoint 3 that computed a price purely from its own reserves. The contract knows all of these for certain, because they are its own memory.
On the right are facts from the real world. The contract has no senses, no internet connection, no way to look any of these up. So when a lending market needs the real dollar price of ETH to decide if a loan is safe, it cannot just check. It needs someone to bring the fact to it.
The fix is to give the blind machine a thermometer. Something outside watches the real-world reading, the price, and writes it onto the chain on a schedule. Now the contract has a number it can read from its own memory and act on, even though the number came from outside.
That bridge has a name. A service that reports outside facts onto the chain for contracts to use is called an oracle, and the most widely used one, the one on your ecosystem map from checkpoint 7, is Chainlink. It is the chain's thermometer.
Here is the catch that we will spend the next act on. A contract has no gut feeling. If the thermometer reads a price, the machine acts on that price with total confidence: it sells collateral, it settles a payout, it mints or burns coins, all on the strength of one number it cannot independently check.
Which means the feed is one of the softest targets in the building. Trick the thermometer into reading a false price and the machine will execute flawlessly on something that is not true. Many of the largest exploits you will study later started exactly here, at the thermometer, not the machine.
So that is the panel opened on the working money. A stablecoin holds a dollar through a real reserve and an open redemption door, in three species: fiat-backed, crypto-backed, and the algorithmic kind that died because it kept only a promise. Even the good kind has issuer, reserve, and bank risk, because the dollars live somewhere real. And a blind contract learns outside facts only through a feed, an oracle like Chainlink, which it then obeys without doubt.
The thread under all of it: every token has plumbing underneath its price, and now you can see some of the pipes. But a stablecoin is engineered to hold still. Most tokens are designed to do the opposite, and that design has rules of its own.
So next we open the last machine in this act: where a token's supply comes from, who gets it, and why so many of them are built to bleed value over time. Tokenomics, blueprint out.