But it felt like IBM Research was totally disconnected from the actual business of IBM. The switch away from pensions also seemed to make IBM way less attractive for new hires.
To have the second largest super computer down the hall, an electron tunneling microscope, and beauty of the Hudson Valley seems like a bygone era. The innovation, patents, and persona were not enough to move the needle vs Silicon Valley.
Total comp is what matters. A meager DB pension is worth less than RSUs from a competing tech company earning tons more profit.
I vastly prefer 401k and mega backdoor Roth capabilities than any DB pension. But IBM’s problem was not just getting rid of DB pensions, it was not paying competitive total comp in the first place. And also not having offices where their target employees would want to work.
The pre-internet / big web tech days were vastly different. Tell a student today that you and everyone you were working around at the time got the chance to work at a computer company. Most don't even have a framework that would allow them to conceptualize what a computer company that isn't Apple and doesn't sell phones or throwaway junk at Best buy actually does. And especially they can't understand what it's like when there isn't a single, incredibly low margin, cheap, and slave labor supported supply chain ready to ship you parts at a moments notice, and those parts will be for a hardware architecture and small number of very similar operating systems that happen to also account for 90%+ of all the serious computing capacity supporting the world economy.
The fact that all the computer companies died and/or were outsourced, and the technologies were all turned into commodities, is something we don't really talk about all that much.
(hope this doesn't come off as condescending, I'm including these extra details for any non native English speakers because it's kind of a slang word)
IBM's approach was indeed heavy-weight, and they loved complexifying things—but so did a lot of their customers in finance, aerospace, automotive, consumer goods, yadda yadda. IT operated by an entirely different business model and set of preferences in 1960–1990 than it did in 2000. It was the bespoke, pre-open standards, we'll handle all the edge cases in this lovingly homegrown solution era. The products were expensive, but in many cases solved technical operational problems that nothing else did. "Best of breed" was a common rallying cry; interchangeable parts and the virtues of utter standardization only came into prominence slowly.
The world's very different now, and there was enormous tension inside IBM, and between IBM and its customers and competitors, evolving to where we are today. (The same was true of almost all the legacy competitors.) However real that tension, what worked in the Before Times was the cause, not that IBM only employed doofuses or brought Mongol Hordes and crappy technology to bear.