Explaining a hidden financial crisisfrom @maxxrosenblum

Transcript

Need to talk about the COBOL knowledge crisis. Somewhere right no, a 55 year old in a golf shirt is the only person on earth that knows why your direct deposit worse, and he's thinking about retiring and no one is training to replace him. The language he writes is called COBOL, and it handles 95% of ATM swipes, dollars 3,000,000,000,000 in daily transactions, and the bad end of nearly every system that touches your money. It just worse. And that's actually the problem because as long as it works, nobody feels the urgency to understand what's behind it. And what's inside it is not code in the way most people think about code. It's forty years of tax law, benefits eligibility rules, and court rulings no one remembers anymore encoded directly into business logic that was never written down anywhere else. It's effectively the institution's memory, and the only people who can read it are retiring at about 10% per year with no pipeline behind them. But to understand why this matters so much, you have to understand how economists think about knowledge. They split it into two types: explicit knowledge, which is the kind of knowledge you can write in a manual and teach in a classroom, and tacit knowledge, the opposite, the kind that only exists because someone has done a specific thing for decades. Right now, the Harvard Business Review estimates that Fortune 500 companies lose around $31,500,000,000 a year best from employees failing to share knowledge before they leave. If that still doesn't make sense and you want an example, look to the skies. Boeing is a textbook case. They outsourced production, laid off some of their most experienced engineers to cut costs, and took their process knowledge with them. And we saw how that ended. COBOL is that same pattern running at the scale of rank entire knowledge transfer takes years of overlap, and worst companies never funded that overlap. A modernization consultant might put a price on the cost of that neglect at millions of dollars for a business to reverse engineer their own systems and reconstruct business logic from code no current employee understands. Tasks that if done five years earlier, will from code no current employee understands. Tasks that if done five years earlier while the original programmers were still on staff might have cost $200,000 and taken one tenth of the time. And now every year they wait, that cost multiplies roughly by five. Most universities saw this drop off happening. In fact, 85% of them dropped COBOL from the curriculum in the nineties thinking the language would die off. And three decades later, we can see they were wrong. The language didn't die, but the workforce did. In fact, Social Security alone runs on 60,000,000 lines of COBOL and about 30% of that agency's IT leadership is eligible for retirement right now. And sure, everyone wants to talk about the fix, AI migration, cloud conversion, rewriting it all in Java. It's very clear the code will outlive all of us. The question is whether anyone will be left who knows what it means.