[Cryptography] Paper check security

John Levine johnl at iecc.com
Fri Oct 2 00:16:25 EDT 2015


Re US check usage: American businesses still write over 10 billion
checks per year.  I encourage all my clients to pay me electronically,
but I still get more paper checks than ACH electronic payments.
Consumers vary a lot, older ones write lots of checks, younger ones do
everything with debit cards.

> In theory, in the old days, the bank could check Alice's
>signature on the check against the signature card on file.  In practice ... that was almost never done.

Depends what you mean by "old days".  Until the 1950s, checks were
processed and posted by hand, and accounts were filed alphabetically
by the account holder's name.  In 1950 the Bank of America hired SRI
to automate check processing, which SRI did brilliantly.  First they
invented account numbers, then they invented MICR coding to print the
numbers on checks, and a magnetic reader for them, and then they built
a prototpe machine to read and sort them.  Word got around, banks all
over the country were desperate for them, General Electric got the
contract and built a somewhat improved version with transistors and
core memory rather than tubes and drums.  The commercial version
called ERMA shipped in 1959 and was a huge success.  The GE ERMA team
included the pre-AI Joe Weizenbaum.

Personalized checks were not common, so for the most part, the only
way to tell who wrote a pre-ERMA check was to look at the signature.
My parents had an account at a small bank owned by my mother's uncle
and well into the 1960s the only concession he made to modernity was
to print the bank's routing number on the checks so other banks would
clear them.

Appparently check fraud increased a lot in the 1960s, which would make
sense since that's when banks first could clear checks without anyone
actually looking at them.

Historically,
John


More information about the cryptography mailing list