Hacker News Reader

Comment by sgt | Hacker News Reader

Ask HN - should I build my new startup around COBOL? Or should I use a less proven language such as Python, Go, Rust ? /s

View on HN

Your big problem with COBOL will be staffing - everyone knows that scarcity plus demand means that COBOL programmers routinely command 7-figure salaries, and even if your funders can pony up the required cash, getting those guys away from their private islands that they purchased with their Y2K booty is going to be tough.

I placed second in a statewide (Texas) COBOL programming contest in high school. This was the late 90s. While in college (which I started early), I kept having people track me down to offer mid-six-figure salaries. It was very tempting. It was pretty surreal (again, this was pre-web boom, and starting tech salaries were mid-5-figures) as an 18 year old having people offer me $400k/year.

I call bullshit.

Being offered $400K a year is a whole lot different than actually getting $400K a year.

Just saw this now. Not sure you caught the context, which was implied by the poster above. This was 1998, when loads of major corporations were terrified of "the Y2k bug" which was talked about in very apocalyptic terms. There was a scramble to find COBOL programmers, because it was feared that the majority of the bugs were in long-running COBOL systems, but most competent COBOL developers at that point had retired. It's true that I didn't follow up with those folks (other than to say, "no thanks"), but being contacted by multiple recruiters promising six-figure salaries was very much a thing.

That was always my impression. But some thread on HN from yesterday a lot of people were saying COBOL salaries were very low, so which is it?

I know a guy who used to write COBOL at IBM. He was apparently quite good. He now drives trucks for a landscape supply company. I told him that he should look into COBOL jobs again so he did some research. It seems that the actual number of available jobs is a lot lower than the hype would suggest.

my understanding is that companies with COBOL are big instituions with people not much interested in "career" but in "stability". So those old programmers stay there until they retire (nothing wrong with that). So they keep up the COBOL system. Once they get older management thinks about replacing the COBOL system with something new. If it's really expensive, a new system is built around the COBOL system with new technologies : report by report, DB by DB, etc. until then old system can be unplugged. It takes 10 to 20 years to do so. Nothing unbearable for a big corp.

What is high paid is not the ability to write COBOL, what is being highly paid is understanding the 40 year old application building the foundation of the insurance or bank. If you understand all the implications of code changes there, you are worth a lot. If you need oversight to not break anything (hardly any tests etc.) you a worth a lot less.

Spot on. I see people talking about learning COBOL or not, but as you say that's not so much the point. You really need to know how ancient mainframes work. A programmer can pick up the basics of COBOL soon enough, but understanding the ins and outs of zOS and the likes? Much harder, especially the more recent your tech knowledge is. Approaches to software and systems design has changed dramatically since the IBM mainframe days. If you want to make money working on old software just learn Java EE or .NET Framework or something!

The way I’ve always heard it is that it’s somewhat bimodal. You got people with many decades of experience who can, as independent consultants charge an impressive price for there time. Then you get people who were maybe bait and switched or took what they could find working as a low paid contractor or developer for a company that doesn’t pay the best.

If you're an expert in CICS, JCL, DB2 then you'll get a job, but if you just hello_world.cob you won't get a job.

Kinda like its somewhat easier to get a react.js angular.js node.js job than a generic javascript job.

I believe the post was sarcasm.

My first job was as a COBOL developer 10 years ago. My annual salary was around $4000

The only high-level language more proven is Lisp, but most of the procedures are written in lower case so I don't think it's serious enough to be a business oriented language.

Also, business people don't like things to think for themselves, so Lisp would be risky in that respect.

Not Fortran, in which you originally needed to SHOUT, though you couldn't go SHOUTING? It was good enough to implement one of the earliest RDBs, after all, which sounds business-like.

You have a sarcasm tag, but I've been waiting for this release!

Just think about it, I can run my PI servers on COBOL without the intricate, ever-failing dance of dependency from externally sourced, crufty packages, whose security holes eventually get opened wide by the latest DawnSpreader or Mousy Trick-Track exploits. /s

Wait... maybe there is something there after all...

You absolutely should. Make COBOL ON COGS a reality and post about it on HN for PR and, more importantly, our amusement.

I mean, if your startup is a COBOL consultancy for big banks and gov't orgs…

We have a bank in South Africa called Capitec, and they were sort of considered a "startup" bank a few years ago, doing things better and more efficient. But they are using COBOL regardless (+ other cloud tech of course as decorations)

Tools that make reading and understanding legacy COBOL code would always do well.

Reply on news.ycombinator.com