I recently threw in (want to check those intelligence metrics!) some real production code into a non-agentic system (just to get a feel how things perform without a custom harness) and results where ... interesting. The particular program uses some preprocessor no LLM we have access to (newest was GPT 5.5) has any clue about - so they confabulate what it could do (Gemini 2.5 didn't even notice there was a preprocessor...). This is expected of course but it somehow seems the problem of this technology that unless you feed it masses of data or mechanically break up the tasks in rote subunits, it just doesn't do anything sensible still...

In school they still perform ok for what is required there today, but from what happens when you say "oh, wifi doesn't work today in our place" I would argue they behave like addicts... It might be a fun addiction insofar as mental problems are usually not the cause but possibly the result if this runs for years on end.

My pet theory is that these kids will graduate to online-gambling around 12-14 because that's exactly the kind of gameplay they prefer on roblox currently. Even when I was 16 there was a noticable share of classmates gambling online - and this was pre-facebook.

Probably the best way forward would be to make KYC mandatory for large, data-selling public fora like we do with banks. Noone complains there and noone is forced to use platforms.

If you don't want to do KYC (such as on hn or a model train forum) you are liable for any criminal activity you enable (such as pedophiles using your private messages or whatever...).

so: I doubt anyone has to care about that pesky GDPR if they buy the government of Malta.

So essentially they sell new hardware and "support" to customers who have been in need to process tabular, multi-GB databases since when a PC was 128MB memory and have been doing electronic record-keeping since the 1970s. They also allow their ~hostages~, ehm, customers who trust them with their data to run processing near the data at a cost/in a cloud style billing model. That is so expensive though that every large IBM-shop has built an elaborate layer of JVMs, Unix and mirror-databases around their IBM appliances. Lately they bought Redhat and hashicorp and confluent thus taking a cut from the "support" of the abominiations of IT systems they helped birth for some more time to come (also remember the alternative JVM OpenJ9, do you all?).

I think the later a company started using centralized electronic record keeping, the higher the likelyhood they are not paying IBM anymore: commercial banks, governments and insurance started digitizing in the 60s (with custom software) and if the companies are old (or in US-friendly petrostates) they are all IBM customers. Corps using ERP or PLM offerings (so manufacturing and retail chains which are younger than banks) used to start digitizing a little later (Walmart only was founded in the 60s and electronic CAD started in the 80s) and while they likely used IBM in the past (SAP was big on DB2) they might not use it anymore (also it helps they usually bought the ERP or PLM from someone else). New Companies whose sole business was to run a digital-platform started on Unix (see Amazon who successfully fought to ditch Oracle even) or just built their whole platform (Google). If those companies predate Unix they usually fought hard to get rid of IBM (Microsoft, Amadeus)

Consulting/outsourcing services have been spun out to Kyndryl, so nowadays IBM only sells hardware, support for their products and ostensibly has some people left to develop their products... The days when that was a big thing and IBM produced all the stuff they sell support for now, have been long gone. A fun link to see how their "product development" operates nowadays is this discussion to bring gitlab-runners to z/OS: https://gitlab.com/gitlab-org/gitlab-runner/-/work_items/275... - tl;dr "hey you opensource company, we are IBM and managed to pay someone to port a go compiler to z/OS. Now we have a customer who wants to use gitlab with z/OS. Would you like to make your software part of our product offering?". A fun fact is that - even within IBM - access to the real mainframe seems to be very limited which shows a bit in the discussion linked above and also with an ex-Kyndryl-person saying: "oh, I once had a contract where we replaced the mainframe and we ran that on Linux-boxes inside IBM, because it was just cheaper that way. Just the big reporting was a bit slow, but the reliability was just fine"

very fun!

here that person is a manager which got demoted from ~500 reports to ~40 and then convinced his new boss that it's good to reuse his team for his personal AI strategy which will make him great again.

We also sublease our mainframes to at least 3 other ventures; one of which is very outspoken they have left the mainframe behind. I guess that's true if you view outsourcing as (literally) leaving it behind with the competitor of your new system... It seems to be the same for most banks, none of which are having mainframes anymore publicly, but for weird reasons they still hire people for it offshore.

Given that our (and IBM's!) services are not cheap I think either a) our customers are horribly dysfunctional in anything but earning money slow and steady (...) and b) they actually might depend on those mainframe jobs. So if you are IBM or a startup adding AI to IBM I guess the numbers might add up to the claims.