From porters to protocols

Prior to about 1850, New York City banks settled their accounts by sending a porter to visit every other bank, every day.

At each stop, the porter presented checks his bank had received from customers drawn on the bank he was visiting, and collected the amount due in gold.

“The route of the Porter was then very long,” financial writer JS Gibbons wrote in 1857. And potentially perilous. Porters carried as much as $50,000 of gold and bank bills in a pouch slung over the shoulder — with “a chain enclosed within the strap, to foil any attempt that might be made to cut it.

“Several of the banks were in the habit of sending a guard with the Porter,” Gibbons added, “and in one or two instances, he was armed with a loaded pistol.”

“Happily, no case required its use.”

Gibbons reported there was only one known attack on a bank porter in mid-19th century New York (which makes me think Gangs of New York may have been a little exaggerated).

With or without a guard, the porters crossed and recrossed one another's tracks, hurrying to get their banks' accounts settled before the end of the working day (they were responsible for closing up, too).

“The system had the simplicity of Indian camps in which each tepee had a path leading to every other tepee,” a Fed history wrote. “But as the number of banks grew, these paths became a tangled web." 

Around 1850, the volume of payments was such that daily settlement in gold became impractical. The banks therefore agreed to exchange only checks each day and settle the resulting balances once a week. 

On Friday mornings, the city's bank porters would gather on Wall Street — outdoors — and attempt to untangle a week’s worth of payments between dozens of banks.

Gibbons described the process:

“Thomas had left a bag of specie at John's bank to settle a balance, which was due from William's bank to Robert's; but Robert's bank owed twice as much to John's. What had become of that! Then Alexander owed Robert also, and William was indebted to Alexander. Peter then said that he had paid Robert by a draft from James, which he, James, had received from Alfred on Alexander's account. That, however, had settled only half the debt. A quarter of the remainder was cancelled by a bag of coin, which Samuel had handed over to Joseph, and he had transferred to David.” 

The Porters’ Exchange was filled with "confusion, disputes and unavoidable blunders," Gibbons added.

In 1853, a bookkeeper suggested a simple solution: The banks should settle their accounts through a central clearing office. The city's banks quickly agreed and the New York Clearing House (pictured above) was established by a group of cashiers in the basement of 14 Wall Street.

52 banks participated in the first day of central clearing, exchanging checks worth $22.6 million.

At the end of the day, each debtor bank made a single payment to the clearing house in gold equal to its net balance. The clearinghouse then distributed the total among the creditor banks.

“By the aid of the clearing-house,” a 1902 encyclopedia explained, “each bank can settle all of its relations to the banks of the city by a single payment, instead of adjusting its relations with each bank separately.” 

Genius!

Besides saving the porters from miles of daily walking, the new arrangement saved the banks from having to keep so much money on hand.

Whereas previously banks had to keep enough gold in their vaults to meet every check presented to them, they now needed only enough to meet the difference between what came in and what went out.

This made the banks of New York City approximately 19x more capital efficient.

“In the first year of the New York Clearing-House,” the encyclopedia added, “average daily clearings of $19,104,594.94 were effected by average daily payments of $988,078.06.”

Modern clearing houses do even better.

CHIPS

Today, the Clearing House Interbank Payments System (CHIPS) settles $26 of payments for every $1 that banks set aside for settlement.

The improvement comes from running payment instructions through a “liquidity-saving mechanism” — a combinatorial algorithm that identifies groups of offsetting payments, minimizing the amount of money that actually needs to move.

It’s essentially the New York Clearing-House, but with better math — and running continuously. 

Whereas the Clearing-House pooled a day’s payments and settled them in a single, end-of-day batch, CHIPS searches continuously for groups of offsetting payments that can settle immediately.

Banks submit payment instructions to CHIPS, which looks for combinations that can settle using the liquidity available. When it finds one, it releases the payments, updates each bank’s position and starts searching again.

Most payments settle in less than a minute. And money that would otherwise sit idle can be put to work elsewhere — making loans, financing trades, or underwriting a merger, say. 

In 2025, the CHIPS network resulted in an estimated $5.5 billion in economic savings for participating banks. 

Crypto might do even better.

Cycles

For all its optimizing wizardry, CHIPS has the same limitation as the original New York Clearing-House: You have to get everyone inside the same centralized clearing system before their obligations can start cancelling out.

Not many make the cut. The New York Clearing House began with 52 participating banks. CHIPS — intended for very large payments — has just 43 financial institutions in its network. 

Cycles founder Ethan Buchman thinks it doesn’t have to be that way: He wants to open up the club to everyone by replacing centralized clearing systems with a blockchain.

Buchman says the inspiration came from an academic paper on obligation clearing. His eureka realization, as the Cycles website puts it, was that “the clearing that banks had guarded for centuries did not have to stay private, and that the same mechanism could be opened up and extended to businesses and individuals.”

Like CHIPS, Cycles has an algorithm that searches for offsetting obligations. Unlike CHIPS, anyone can submit an obligation.

As described in a white paper, participants would submit mutually agreed obligations to the Cycles protocol. Cycle’s privacy-preserving solver would identify where they might cancel out against other pending obligations. The resulting net balances and remaining settlement flows would be posted back to the blockchain where they would settle in a single transaction.

There’s a lot of math involved: graph theory, network-flows theory, max-flow problems.

And blockchain cryptography: zero-knowledge proofs, trusted execution environments, Merkle-tree state transitions.

But underneath all the complexity is the same basic idea a New York bookkeeper had in 1853: Before moving any money, first figure out how much of it doesn’t actually need to move.

Then let everyone not move it.

— Byron Gilliam