Thread regarding IBM layoffs

AI won’t magically fix COBOL modernization. The real challenge is decades of business logic embedded in those systems. But it can reduce the human effort needed for the mechanical translation part. If that work becomes faster and cheaper, it could pressure the legacy maintenance ecosystem around companies like IBM — and probably reduce the number of people needed for that work.

I struggle with the fact about how people are gullible to give any credence to the words of id1ots, useless pieces of sh1t, like Rob Thomas. The guy most likely did very little in the technical side (productive) of IBM, if any. He is just a scam artist, no different than any other snake oil salesman.
So, there is no need to waste time reading and thinking about what he says.

@m8
DOA
dead on arrival

IBM is where software goes to die. Once IBM a purchases (not develops) a platform , or a successful software product , and it passes through the IBM doors, the slow erosion and degradation begins.
No amount of AI can save it from demise. It will Only buy some more time until it is sold off , or tossed on the scrap heap.

@ez excellent insights and comments and honesty thank you.

This is typical IBM FUD (fear, uncertainty, doubt.)

He does tip his hand a little bit in the last sentence. It is absolutely true that customers have been trying to re-platform since the 1990s. And they have been doing so for very good reasons. The most important reason is that the pricing model is very beneficial to IBM and very costly to the customer. The great idea of IBM pricing has always been to make IBM's revenue as high as possible while not actually driving customers to "bite the bullet" and invest heavily in re-platforming. In addition to the financial reasons, there are technical reasons to re-platform. Despite what IBM tells you, this really is a geriatric system that shares little with modern hardware and software practices. Operating this system requires personnel with bizarrely recondite knowledge. It is very difficult to entice anyone born in this century to pursue the fine buggy whip skills required for this platform when they could master modern automotive driving.

The FUD presented in the post held a certain currency for many of the years since the 1990s. One of the reasons for that is that the COBOL language itself does encourage a very inscrutable encoding of the "business logic" mentioned in the post. There are a number of reasons for this. Probably the most important reason is the famously unconstrained control flow in COBOL programs. So many "perform" and "goto" statements would, today, be function or procedure calls but the nature of these calls is often obscured to even skilled human software developers. The fifty or sixty year old concept of BCD (Binary Coded Decimal) to do simple integer arithmetic is problematic because there are so many possible bit patterns that are not legal and also not diagnosed anywhere along the way. Also, the somewhat addictive (to a COBOL programmer) EXEC THIS and EXEC THAT in a COBOL program means that a single source file actually has several programs (a COBOL program and a THIS program and a THAT program.)

Let us ignore the IBM claim that this new AI initiative is about simply "translating legacy code." Concomitant with this claim is IBM's Elysian dream that the translated code will continue to run on the platform (and attract giant revenues for IBM.) Suppose that modern AI can clarify actual control flow logic (including logic that would normally be procedure/function calls) and meaningful integer arithmetic. Also, suppose that modern AI can separate SQL logic from COBOL logic. Now we are close to actually being able to abstract out business logic, Furthermore, we can determine which programming concepts are fundamental to all programs in the enterprise and which programs might be eligible to be re-platformed one at a time.

The recent news about AI processing COBOL code doesn't at heart have all that much to do with COBOL. But it is a likely (or at least possible) path to practical re-platforming. And, apparently, even IBM recognizes that this is a threat.