It was written when the cost/benefit calculation of writing it involved "downsizing" thousands of clerks who were doing the job manually.
It can't be replaced because the cost/benefit calculation of replacing it does not involve anything like those kinds of numbers. The benefits of avoiding even a major security incident just don't compare to the costs.
Y2K taught me this. Nothing has changed since.
TSB customers have learned how painful it can be if the migration of a core banking system fails: https://www.independent.co.uk/news/business/news/tsb-it-fail...
And there's the real problem:
The banks will save their money and stick with plaintext if they can get away it.
Put another way, it's an incentive problem.
With a six-digit PIN.
See
- https://www.labanquepostale.fr/ > "Me Connecter"
- https://lcl.fr/ > "Mon Espace"
- https://particuliers.societegenerale.fr/com/icd-web/cbo/inde... > 12345678 > Valider
They did. And Pi factor came in. And budget was cut because those pesky fintech are a threat, and clearly money was better spent on a more modern offer than on those "security" concerns.
And yes, it's possible to add 2FA to the front layer. However, the remnants of COBOL code running on the mainframe for the last 25 years weren't designed to handle passwords length 9 or more, and the last guy who knew how that code was built has retired some 15 years ago.