COBOL

COBOL

COBOL is a programming language from 1959, designed for accounting, payroll, and banking operations. It is considered outdated, but continues to run in banks, insurance companies, and government agencies today, managing enormous sums of money on a daily basis.

COBOL is a programming language, meaning a fixed notation in which humans give instructions to a computer. The name stands for “Common Business Oriented Language.” It was developed in 1959 in the USA, among others by the computer scientist Grace Hopper. The goal was unusual: the instructions were supposed to look like simple English sentences. One command, for example, is ADD SALARY TO TOTAL. COBOL was never intended for games or graphics, but for calculations, accounts, and lists.

Why 65-year-old code still moves money

COBOL seems like a museum piece, but it is very much alive. Estimates suggest that several hundred billion lines of COBOL code are still in use worldwide. Large parts of payment processing run on it: account balances, wire transfers, credit card statements. Insurance companies, tax authorities, and airlines also use the language.

The reason is simple: the programs work. A booking system that has been reliably calculating for 40 years is not something you replace lightly. Every migration carries the risk that money gets booked incorrectly. Several banks have attempted such modernizations and lost hundreds of millions of dollars in the process.

The problem is the lack of new talent. The people who wrote these systems have long since retired. Universities barely teach COBOL anymore. During the Covid-19 pandemic, the US state of New Jersey publicly searched for COBOL programmers because its unemployment system couldn’t handle the flood of applications.

Sentences instead of symbols: the structure of a COBOL program

A COBOL program consists of four fixed sections, known as Divisions. In order, they govern: the name of the program, the hardware used, the data used, and finally the actual computational steps. This strict order seems cumbersome, but it also makes unfamiliar code readable even after decades.

The wordiness is striking. Modern languages write an assignment as x = y + z. COBOL writes COMPUTE X = Y + Z or, in full, ADD Y TO Z GIVING X. The idea behind this was that even merchants without a computer science degree should be able to understand the code. As a result, a COBOL program is often three to five times as long as an equivalent program in a newer language.

A technical peculiarity explains its long lifespan in the financial sector. COBOL calculates with decimal numbers by default, just as one would on paper. Many other languages calculate internally with binary fractions, producing tiny rounding errors. With millions of transactions per day, such errors add up to real amounts. It is precisely this reliability in monetary calculations that is one reason the language has not disappeared to this day.

COBOL in current headlines and in AI projects

You encounter COBOL in everyday life without noticing it. When you withdraw money from an ATM or make a bank transfer, there is a high probability that a COBOL program is running somewhere in the chain. These programs run on mainframes, meaning especially robust central computers that fill entire server racks.

In the news, the language usually appears in two contexts. First, in the case of outages: if a banking system fails for days, a botched migration of old software is often behind it. Second, in politics and administration, when government agencies cannot pay out subsidies quickly enough because their systems date back to the 1980s.

For a few years now, an AI angle has been added. Companies like IBM offer tools that use language models to automatically translate COBOL code into more modern languages such as Java. This is attractive, but tricky: the AI must not only understand the code but also the business rules, which are often not documented anywhere. A common misconception is that COBOL is simply “bad old technology.” The real problem is not the language, but the lost knowledge of why a program calculates exactly the way it does.

Related Products

Subscribe free. Unsubscribe the second it sucks.

High-signal news across AI, business, UX, and tech. Every morning.