> The Bank of New York Mellon in 2012 found it had 112,500 individual COBOL programs, constituting almost 350 million lines;
Could that really be true? Can anyone provide some context to make that make sense?
And Cobol programs are _verbose_. Have a look at the examples on Rosetta code. C# (which isn't particularly concise) can do in 40 lines what Cobol needs 250+
Where we (modern programmers) would write a program with a large number of modules and imports, most Cobol programmers probably wouldn't. There is a lot of copying-and-pasting going on there.
That's an average program length of 3000 lines. Imagine your code base consisted of 112,500 files, each of about 500 lines (including HTML, CSS, SQL statements and "real" program code).
That's about what it is equivalent to.
Plugging that equivalent number into a Cocomo calculator, it says "20 years of 800+ programmers".
> Plugging that equivalent number into a Cocomo calculator, it says "20 years of 800+ programmers".
Of course, it's been 50+ years, but I doubt Bank of New York had 300+ programmers working for them all those years?
It would make sense if copy-paste were a lot of it.
Which makes me think if you did want (or more likely have no choice and need) to make some serious changes to the codebase, some kind of automated assist could be very helpful. I wonder what the state of COBOL tooling is. I wonder if it's going to get better as the technical debt comes due in a hard way.
As a bank, you also do all this stuff internally to pay your employees, contractors, create audit trails, document things, keep track of office equipment and supplies, etc.
All the things that other companies might use commercial software for, or just use Google sheets or email or another generic solution, this company has a long-established COBOL program for.
Are all 112,500 programs used? Probably not. But are 10% of them used at least once a year? Probably.