Meet COBOL: The 60-Year-Old Language Holding Civilization Together.

There’s probably no programming language in history that’s quite as awkward as me.

For over 60 years, I’ve been quietly powering 43% of the banking systems worldwide, handling a mind-blowing $3 trillion in daily transactions.

In the US, where information systems developed earlier, I’m even more dominant: 95% of ATMs and 80% of banks still rely on me.

Today, 220 billion lines of my code are still running out in the wild.

That’s right — I’m COBOL.

Let’s have a quick look at me in action. Here’s the classic “Hello, World” example:

IDENTIFICATION DIVISION. 
PROCEDURE DIVISION.     
    DISPLAY "Hello, world". 
END PROGRAM HELLO.

Simple enough that it needs no explanation, right?

Now, here’s something a little meatier — calculating a worker’s pay with overtime (time-and-a-half for hours over 40):

COMPUTE-GROSS-PAY. 
    IF HOURS-WORKED > 40 THEN 
        MULTIPLY PAY-RATE BY 1.5 GIVING OVERTIME-RATE 
        MOVE 40 TO REGULAR-HOURS 
        SUBTRACT 40 FROM HOURS-WORKED GIVING OVERTIME-HOURS 
        MULTIPLY REGULAR-HOURS BY PAY-RATE GIVING REGULAR-PAY 
        MULTIPLY OVERTIME-HOURS BY OVERTIME-RATE 
            GIVING OVERTIME-PAY 
        ADD REGULAR-PAY TO OVERTIME-PAY GIVING GROSS-PAY 
    ELSE 
        MULTIPLY HOURS-WORKED BY PAY-RATE GIVING GROSS-PAY 
    END-IF

Reads almost like plain English.

That was by design. Back in the day, the U.S. Department of Defense wanted a programming language that used the simplest English possible so that more people could describe problems to computers without needing a Ph.D. in computer science.

I ended up with over 400 keywords — compare that to the few dozen in C or Python — and some folks find me clunky and wordy. Others love it, saying I’m self-documenting code: just read it out loud, and you’ll understand what’s happening.

Where I shine is handling structured data — reading records from files (names, addresses, Social Security numbers, phone numbers, you name it), processing them (like calculating paychecks), and writing them back to files.
That’s why you’ll find me everywhere in banking, insurance, and government systems — the real “serious business” stuff.

But here’s the catch:
The folks who know how to program?
They’re mostly retired, chilling at home, spoiling their grandkids.
Universities don’t teach COBOL anymore, either.

The result?
COBOL programmers are going extinct.

Because of my age and so few people understand me, my codebase often gets labeled a “spaghetti mess.”

During the pandemic, when unemployment systems were overwhelmed — New Jersey alone saw 320,000 new users quickly — the state urgently needed COBOL programmers to fix critical systems.
The problem was there were none to be found.

Governors even went on live TV to publicly beg for COBOL expertise.

People have tried all kinds of creative solutions to deal with the COBOL shortage:

1. Bring the Retirees Back

Take Bill Hinshaw, for example.
He started programming in the 1960s and was ready to retire — until desperate clients called him for help.
So he founded COBOL Cowboys, a consulting company made up of veteran COBOL developers.

His team of 20 part-timers has a “youngest” member over 50.
These seasoned pros can patch critical systems, write documentation, and bridge old and new platforms — for about $100 an hour.
Sure, it’s pricey — but compared to rewriting an entire banking system from scratch?
It’s a bargain.

2. Rewrite the Code

Some organizations have tried to ditch COBOL by rewriting their systems in modern languages.

For example, in 2012, the Commonwealth Bank of Australia partnered with Accenture and SAP to replace its COBOL core banking system.
The project took five years and cost over 1 billion AUD (USD 750 million). Along the way, critical payment system failures occurred, like 600,000 lost transactions in 2015. Ouch.

In 2018, TSB Bank in the UK attempted a similar migration and ended up with multi-day outages and a staggering £330 million in costs.

Moral of the story:
Trying to “kill” COBOL is like replacing a plane’s propeller with a jet engine — in mid-air.
Good luck with that.

3. Let AI Handle It?

Enter IBM.
Since most COBOL code runs on IBM mainframes, Big Blue couldn’t just sit on the sidelines.
They introducedwatsonx, a tool that uses AI to assist in converting COBOL code to Java.

Sounds magical.
Well… not so fast.

IBM later admitted that watsonx it was just a coding assistant—not a magic wand.
The AI-generated Java code still needs human engineers to tweak and validate it.

In other words:
AI can speed things up, but it’s not replacing COBOL programmers anytime soon.
(For now, I can breathe easy.)

Even DOGE (yes, that DOGE) announced plans to rebuild the Social Security Administration’s 60 million lines of COBOL using AI — But everyone is rather pessimistic about the news.

Honestly, it’s poetic:
Our modern world still runs on a 60-year-old who barely understands the COBOL code.

As the veteran COBOL engineers gradually fade into retirement, the big question looms:

Who’s going to keep civilization running next?