June 2019. Facebook announces Libra: a global payment currency, backed by a reserve of real assets, governed by an association of large companies, aimed at a potential user base in the billions.
If you had been reading industry predictions for the previous five years, this was the thing they described. Real backing. Real distribution. Real money behind it. Not a startup with a whitepaper, but the largest consumer platform on earth with a payments consortium attached.
It never launched.
Payment partners withdrew within months under political pressure. Central banks and legislators objected to a private currency operating at that scale. The project was renamed Diem and narrowed repeatedly. In January 2022 the association wound down and sold its intellectual property and technology to Silvergate Capital for $182M. A former Meta executive later characterised it as a political decision rather than a technical or commercial one, which is a self-interested framing and also broadly consistent with the public record.
Nothing failed technically. Nothing ran out of money.
The best-resourced attempt anyone has made at the exact future everyone was forecasting was stopped by institutions that were not in anybody's model. Whoever was forecasting from capability, adoption curves or engineering timelines had no term for the variable that actually decided it.
Every course like this ends with a chapter about the future, and most of them are worthless, because they are a list of things the author would like to happen written in the present tense.
We are going to do something narrower and more useful. This lesson is mostly a method, because the method survives and the predictions will not.
Note the frontmatter: this lesson is marked low confidence, and that is deliberate. It would be strange to write honestly about the future and then claim to be sure.
First, separate two kinds of sentence
People move between these inside a single sentence and it is the most common way a claim about the future gets smuggled past you.
Observably true now. Something you can go and check today. Stablecoin volume, the existence of spot ETFs, what a given jurisdiction's law currently says, how many chains process what.
A forecast. A claim about a state of the world that does not exist yet.
"Stablecoins are being used for settlement, so within five years they will replace correspondent banking" contains both. The first half is checkable. The second half is a forecast wearing the first half's credibility.
Whenever you meet a claim about where this is going, split it at the seam before evaluating either half.
What is observably true, as of this writing
Short list, deliberately, and every item is checkable rather than argued. Anything here may have reversed by the time you read it, which is the point of the review cadence on this lesson.
- Stablecoins are the largest real use. Most value moved on-chain moves as stablecoins, not as volatile assets. This has been true for years and has grown throughout.
- Regulated access wrappers exist and are used. Spot ETFs and similar instruments now let institutions hold exposure without touching custody, which S212 covers properly.
- Regulatory frameworks have moved from absent to partial. Several major jurisdictions now have specific rules rather than applying old ones by analogy. Partial, inconsistent between jurisdictions, and no longer a vacuum.
- Fraud has grown alongside legitimate use. F106-01's figure of a record $17B estimated in scams and fraud in 2025 is the same trend line seen from the other side, and any honest picture of the direction includes it.
Notice that none of those is a prediction, and none is about price.
What is genuinely uncertain
Also short, and stated as questions because that is their honest form.
Does anything achieve mainstream use that is not trading or stablecoin transfer? Many attempts, little evidence yet, and this has been "about to happen" for roughly a decade.
Do institutions custody directly or only hold wrappers? These lead to very different worlds and the answer is not in.
Does regulation converge or fragment? Fragmentation is the current state, convergence is frequently predicted, and Libra is what happens when you forecast this wrong.
Does self-custody stay a minority practice? F102-02's uncomfortable question. If the technology's central promise requires a skill most people decline to learn, the outcome is custodians, which is what already happened once.
We do not know the answers. Anybody who tells you they do is describing a preference.
Take a claim you will meet in some form this year: "Tokenised real-world assets are going to be huge."
As written it cannot be evaluated, and worse, it cannot be wrong, which means it was never information. Fix it in three moves.
One. Add a threshold. Huge compared to what? "Tokenised treasury products exceed $50B outstanding."
Two. Add a date. "By 31 December 2027."
Three. Name the source that settles it. "As reported by a named public tracker, using its stated methodology."
Now it is a claim: "Tokenised treasury products exceed $50B outstanding by 31 December 2027, per that tracker." You can attach a probability to it, write it down, and find out.
Then do the part almost nobody does, which is decide what you would do differently at each probability. Suppose you put 40 percent on it. Ask: is there any action I would take at 40 that I would not take at 20? If not, the forecast is entertainment, and you can stop.
And a warning that F102-02's P10 makes compulsory. Even a well-formed forecast that comes true tells you nothing about whether an investment in it makes money, because the price already contains everybody else's forecast. Being right about the world and being right about the trade are separate problems, and S206 and S207 are where the second one is treated seriously.
The scoring is the whole exercise. Write down five of these, with probabilities, with dates. Put a reminder in your calendar. When the dates arrive, score yourself. Almost nobody does this, which is why almost everybody believes they were roughly right about the last cycle.
Why the institutional constraint binds hardest
The Libra autopsy generalises, and it is the single most useful heuristic in this lesson.
Technical problems in this field get solved, slowly and imperfectly, but they get solved. Throughput improved. Custody improved. User experience improved.
Institutional and regulatory constraints do not get solved. They get negotiated, and they can simply say no, without a technical reason and without appeal.
So when you are assessing where something is going, the productive question is rarely "can this be built". It is:
- Who loses if this succeeds, and how much power do they have?
- Which existing rules does it break, and who enforces them?
- Does it need permission from anyone, and what is that party's incentive?
Libra was technically fine and comprehensively answered "yes" to all three in the worst way. Every roadmap that ignored those questions was reading the wrong variables.
The technology is inevitable, so the details are just timing.
"Inevitable" is the word to distrust most in this industry, and the reason is structural: the claim can absorb any evidence. Adoption grows, it was inevitable. Adoption stalls, it is early. A claim compatible with every observation is not a claim about the world.
Two specific corrections. Technologies fail at the adoption stage routinely, and being technically superior is not close to sufficient. And "inevitable eventually" is not a useful statement, because everything is inevitable on a long enough horizon and nobody can act on it.
There is also a self-interested pattern worth naming. Inevitability is what people say when they hold the asset, because it converts a forecast into a fact and removes the burden of defending it. Notice who is saying it, and what they own.
The honest version, which is smaller and true: a specific set of properties exists that did not exist before, some real uses have been found, more may be, and the outcome depends on institutional decisions nobody can currently predict.
Closing the course
You now have what F102 was for. You know what money does and that a deposit is a claim. You know the four arguments that survive and the four objections that survive. You know why the two largest systems are shaped the way they are. You know what makes a stablecoin stable and where each kind breaks.
And you know that the thing that determined the biggest attempt at this so far was not on anybody's roadmap.
F102-L, the lab, is where you find out for yourself: price a $200 remittance to three countries across a bank wire, a remittance service and a stablecoin transfer, with the full cost, the time and the failure modes including the off-ramp. We have told you already that the crypto rail does not always win. Do the lab and find out on which corridor, and by how much, because that specific number is worth more than any forecast in this lesson.
Then F103, where we stop asking whether this is worth it and start on the part that determines whether you keep any of it.
Split every claim about the future at the seam between what is observably true and what is a forecast, because people cross that line inside a sentence. A forecast is only usable with a threshold, a date and a source that will settle it, and only worth making if you can name an action that changes with the probability. Institutional constraints bind harder than technical ones, because technical problems get solved and institutions can simply say no, which is how the best-funded attempt at this exact future ended with its assets sold for $182M without anything failing. Distrust inevitability, notice who is asserting it and what they hold, and score your own predictions on a date you set in advance.