There's nothing particularly difficult about Cobol, it can be learned, and with enough effort, all the undocumented existing code can be understood and reverse-engineered enough to be able to make changes. It's not pleasant work, but it's doable.
If there was value in it, salaries/consulting rates would reflect it and people would be queuing up to learn it and make good money. That isn't happening, so it seems like Cobol devs aren't actually that valuable to these companies.
There was a comment on HN a while back to the effect that the well-known stories of COBOL devs being dragged out of retirement to earn massive consulting payments was a myth. They were actually being paid good, but not extremely good, rates for their knowledge of the business logic embedded in old code-bases. The implementation language was largely irrelevant.
> What if you are one of the only few who know how to maintain a Cobol system for a bank? Aren't you very valuable?
The bank would see the situation (only one person knows the Cobol system) an extreme risk and do whatever it takes to mitigate that risk. The short term answer is to keep you hired but the long term answer is to either find a steady supply of Cobol programmers or move away from Cobol to something where expertise is more available.
People get overly excited by some "last man standing" success stories of a COBOL developer based in US of A. My only encounter with COBOL developer was a freshly graduated girl based in post Communist country, her earnings were something around 25k EUR annually. I pointed out that her career might be something of a dead end but she was more like "a job is a job". Oh, and they were using some dinosaur version control as well.
I've seen a post on local job board in Poland, IBM was looking for junior cobol engineer. Any student with some java knowledge will fit, they said. I'm sure they will not pay anything more than 25k eur anually.
Also if there is some legacy cobol system, many managers are incentivised to gather a new team to write a replacement with some modern technology stack. Cobol is definitely a dead end for the career.
> legacy cobol system, many managers are incentivised to gather a new team to write a replacement with some modern technology stack.
Absolutely not in outsourcing centers like this. You're not supposed to show any initiative or god forbid - design anything. You grunt the COBOL until budget for the project is zero.
Do you seriously suggest that banks outsource the maintenance of the mainframes that process their money transfers? If so I'd like to see a source, because the only articles I have read about on this topic (e.g. [0]) tell me the exact opposite: it's all in-house the risks are too high. For larger banks it's practically a national security risk. Maybe the software they use is from abroad (usually ancient IBM shit), but that's not the same.
Yes, they do; I won't name names because it's from personal experience, but I've seen all maintenance of the core banking system outsourced (e.g. to IBM), I've seen the development/maintenance of the core banking system (which many banks buy from outside vendors, most banks do some customization but many have a commercial system as the base, not some unique fully home-grown solution) go to heck because the vendor moved most of their maintenance from the existing (presumably expensive) teams to cheap overseas engineers which weren't very effective for various reasons; I've also seen banks doing offshoring without outsourcing (i.e. make up an IT department in a 'cheap' country where they don't really do banking) and outsourcing without offshoring.
It's definitely not the case where it's all in-house. Some organizations do, some don't, and there are so many banks (and so many different types&sizes of banks) in the world that every option is represented.
I should have been more specific: When I wrote "maintenance" I was thinking more of "the people in charge of keeping things running on a daily basis and who solve crisis situations". Of course I don't believe that banks build all of their computer systems from the ground up themselves.
But even after narrowing it down to that your stories make me think I had too much faith in the banking system.
Again, I would like to see a source. I'm not saying you're wrong. Maybe you're talking about small banks in the US or something, of which I know basically nothing. I am thinking of examples like Nordea, the bank in the article I linked, which has a market share of around 20% in Sweden[0] and handles the government's bank accounts. Especially that last part makes me extremely skeptical that the Swedish government would be OK with handing over the keys to their economic kingdom to cheap digital labor abroad so a company might save some money.
> Especially that last part makes me extremely skeptical that the Swedish government would be OK with handing over the keys to their economic kingdom to cheap digital labor abroad so a company might save some money.
Well, to begin with, the Swedish government seems to have known for a rather long while that Nordea has been doing quite a lot of other shady shit abroad, and still stuck with them... From your own WP cite, a bit further down: https://en.wikipedia.org/wiki/Nordea#Scandals .
Not that the Swedish government is much of a guarantor against sending sensitive data abroad in the first place: They were quite OK with the Swedish Transport Agency letting a contractor (Oh look, IBM again!) hand over data, including higly secret (protected identities, military vehicle registry, security van routes and timetables) to sub-contractors abroad with no Swedish security clearances: https://simple.wikipedia.org/wiki/Swedish_Transport_Agency%2... (very cursory article; much more in the Swedish-language version). So them trusting Nordea... Isn't much of a recommendation for Nordea either.
The link in your earlier post seems dead, but if we're talking about Nordea specifically (coincidentally, I have worked there many years ago), partial IT outsourcing has always been an option, it has had thousands of external IT consultants, it has used a number of core systems from external vendors in its many banks (in such multinational organizations it's often not that simple to draw boundaries where a department or region or acquisition is often doing things differently with different systems or processes), it had a very large strategic outsourcing deal to IBM in 2017-2019ish, etc.
If you're extremely skeptical that the Swedish government would be OK with handing over the keys to their economic kingdom to cheap digital labor abroad so a company might save some money, then you need to re-evaluate your mental model, because that is definitely happening in reality. They may impose some legal limitations - the exact same limitations as towards any other bank - but they generally act as a normal minority shareholder, demanding the company to be cost-efficient and maximizing profit even if it means outsourcing and offshoring.
Well my brother worked in BNY Mellon India center. They have thousands of people and work core banking processing. BofA have their own India development center with thousands of employees. And this is outside of 10s of thousands outsourced to Tata, Infy etc.
You think core banking process like transfers/ trading is some kinda crown jewel. It may be in terms of messaging but most of core IT part is just a cost center.
Customer data and all need to be in their own secure data centers for legal reasons but all the processes can be designed, developed and executed from anywhere at lower cost.