Section 00
The proposition, and the test
A position was put: that the next advance in cross-border payments is not another rail but an intelligent layer that routes across the rails that already exist. This is what was proposed, what would have proved it wrong, and where the evidence sits.
What was proposed
Payments as infrastructure argues that the payments market does not suffer from a shortage of rails but from fragmentation; that bank transfers, instant-payment systems, card networks, stablecoins and blockchain settlement will coexist; and that the competitive advantage therefore lies in orchestrating them intelligently — evaluating cost, speed, liquidity, regulation and resilience per transaction, and making the complexity invisible.
It is a coherent position and it makes claims that can be checked. That is worth saying plainly, because most strategy documents do not.
Three claims that can be tested
| The claim | What would prove it wrong |
|---|---|
| Stablecoins are cheaper — settlement across a chain undercuts correspondent banking and the FX leg. | A corridor-level cost model in which the stablecoin path, including both fiat legs and the conversion spread, costs more than a competent bank-and-FX route. |
| Stablecoins are always-on — 24/7 settlement removes the weekend and the banking calendar. | Evidence that the constraint is the funding window at each end rather than the rail, or that the domestic rails being routed to are already continuously available. |
| Orchestration is the differentiator — choosing the best route per transaction is where the advantage lies. | A measurement showing the gap between the best and the next-best route is small against a competent operator, or that buyers do not pay for route selection. |
How this was tested
32 research documents, each following the same template: a stated question, conclusions with a confidence level attached to each, a table of numbers, the variance and tail risk rather than just the averages, the questions left open, and an explicit statement of what would change the conclusion.
The part that matters most is the last column of every numbers table. Each figure carries the basis on which it is known — whether somebody measured it, whether a named party published it, whether we derived it from things that were published, or whether it is simply a practitioner's judgement. That distinction is usually invisible in strategy work, and it is the difference between a finding and an opinion.
So: 67% is sourced to a named party, 13% is our own arithmetic with the working shown, and 2% is judgement with nothing behind it. That last figure is the one to hold this work to.
Where a number could not be obtained at all, it is recorded as a gap rather than filled with an estimate. There are nine such gaps and they have their own section, placed deliberately before our conclusions rather than after them.
One of the tests fired
Several of the research documents wrote down, in advance, the number that would prove them wrong. One examined the closest historical precedent for this proposition — an earlier wave of companies that also promised to route intelligently across payment networks — and registered a falsifier: if the gain from choosing the best available route is under twenty basis points, the thesis is in trouble.
The measured gain against a competent operator is thirteen to seventeen basis points. That is the strongest single argument on this site against the position it was built to test, and the sections that follow should be read with it in hand.
How to read the rest of this
Any sentence with a small number after it is a claim resolving to a specific research finding. Click it and a panel opens showing what that finding says, how well it is known, why it might be wrong, and what would change our mind about it. A claim that rests on one of the open gaps is marked with a triangle.
The evidence sections come before either interpretation, deliberately. They are laid out so they can be read against the three claims above and judged directly. Section 07 then places two independently reached readings side-by-side against the same evidence. Nothing before it depends on either conclusion, and the disagreement is exposed rather than reconciled away.
The analysis pack
All 32 research documents, the synthesis, the open-gaps register, both independent review passes with a point-by-point comparison, and the extracted claim graph — the complete input set, for anyone who wants to work with it or dispute any part of it.