Kerem Albayrak
Written alongside a small group at Anières.
Field notes on the record, the people around it, and the questions that keep returning.
Exposure as a network property, not a per-subject fact
A person's exposure is largely a function of who else's exposure they sit next to. The single-subject report was always a convenient fiction.
Wholesale rails and retail rails carry different signals
The interbank pilot and the consumer wallet are two different data sources. Reading them as one flattens the useful distinctions.
Reading central-bank digital rails as a public dataset
Whatever the political argument, the pilots are producing structured records. Reading the records is a different exercise from having an opinion on them.
Programmable payments change what a monitor is watching for
Conditions attached at the settlement layer are metadata. Denser metadata makes standing reports easier to write and harder to bluff.
The upgrade nobody noticed that changed what a payment carries
A payment used to be an amount and two account numbers. Now it is a small structured object. Structured objects are exactly what our work reads.
Stablecoin activity read as counterparty exposure
A wallet is not a person. A pattern of settlement across chains, exchanges, and off-ramps is closer to one than any single record we would otherwise pull.
Where digital value leaves the network is the part worth reading
On-chain analytics gets most of the attention. The off-ramp, the exchange, the bank, the country, the hour, is usually the more useful record.
The correspondent network read as a working map
Who clears for whom, in which currency, through which hub. The map is older than most compliance stacks and reads better than any of them.