A better question might be “what happened to IBM?” Most ascribe competition and cloud to their decline, and those are primary factors, but the main reason for their inability to effectively compete was an incompetent CEO.
If you ever heard Ginni Rometty speak at an interview or conference, it’s clear she was in way over her head. She had no real understanding of anything AI or cloud, only that she needed to have AI and cloud and a ready offering for any new buzzword: blockchain, quantum computing, etc.
Under her watch, IBM acquired a cast of companies and tried to assemble a cloud computing arm with little time to do it. While some of Watson’s core tech was born in house much was bolt on acquisitions and hodgepodge marketing to push the narrative of a cohesive whole. The developers gave it a chance, and it mostly sucked. It wasn’t aimed at them after all—-it was aimed at their managers.
I was at IBM at the time when Ginni took over. While I'm sad that she didn't have what it took to save IBM, the seeds of IBM's downfall were sown over the preceding decade by her predecessor, Sam Palmisano. This Forbes article sums it up better than I could [0]. In brief, Mr Palmisano decided to focus his tenure on delivering unsustainable earnings-per-share numbers to investors, while ignoring existential threats to IBM's business from cloud computing and other advances in technology. He was handsomely rewarded for this focus. While Amazon built AWS, IBM bought back shares and squeezed cash cow business as hard as it could. The shareholders were happy with that, for a time.
Even at the point of Palmisano's departure in 2012, he had committed IBM to the infamous (for IBMers at least) Roadmap 2015, which promised that IBM's earnings-per-share would rise even higher by 2015. It left increasingly little money for arresting the decline in competitiveness of IBM's products. It is to Ginni's credit that she did eventually abandon Roadmap 2015, though not as soon as she should have done.
When Ginni took over, I remember feeling a sense of relief if anything and maybe a little renewed hope. IBM finally got its head out of the sand and started trying to compete again. The Watson badge that got slapped on anything vaguely related to big data, machine learning and analytics was one of those efforts. It's a shame that it didn't work but it was probably too late by the time Ginni took the job.
> Under her watch, IBM acquired a cast of companies and tried to assemble a cloud computing arm with little time to do it.
IBM's problem was that consulting and legacy products (think mainframes) are still making money and what the company is known for. These two products aren't attractive for the type of engineer you need to build a world class cloud. This, coupled with less than stellar compensation meant that the critical talent required simply went somewhere else.
What they should have done is invest in cloud computing startups, keep them separate entities and act as a sales channel for these products, for legacy customers only.
Watson was more of a Marketing name and concept than a product. The actual Watson tech, while not bad, was not a general purpose AI - all it did was assemble a bunch of inferences from a large corpus of text.
At the time, it was better at that than most systems, but it still produced a fair number of howlers when reviewed by humans. (I briefly looked at Watson as an alternative in evaluating technology for a new learning system some years ago.)
IBM ended up buying The Weather Company (Weather Channel, Accuweather) around the time when I was there. Lots of things were getting the "Powered by Watson" thing slapped on them around that time.
Lots of things said "powered by Watson," but Watson was not a single system, nor did IBM actually add value in most cases (see: Watson Health, which was a catastrophe).
Not surprised to learn. Approx. 5 years ago we talked to the team at IBM to see if they are able to replicate and improve historical results we had achieved in trading our quantitative algorithm using our proprietary data set. Not only could they not improve it, they couldn't even come close replicating the real results. They became so desperate that they were willing to send an entire team from the West Coast to the East Coast at their expense in order to identify the reason.
Former IBM employee here. When I joined IBM, Watson was all the rage. I made sure I got on a Watson team, which happened.
During the training phase, I quickly realised I was only learning things that were specific to Watson. I really hate the idea of learning things that are not transferable, so I quickly backed out. I dodged some blockchain projects as well, and got me on some good old full stack projects.
This comment is not meant as an answer to the article title (it's behind a login/pay wall, so I couldn't read it), but I can imagine there are some other fantastic concepts out there that people are not willing to learn, because they have to think about their own career as well, not just about the success of their company.
The sad thing is: I was really good at the Watson thing. I just didn't want to invest in it myself.
What's "transferrable" is fungible. If you think you're only learning things specific to Watson, you're not looking at a big enough picture.
I spent nearly a decade at Microsoft on the Windows team. I was working in a private codebase with C++ frameworks that no one else in the industry had exposure to, using technologies like Win32 and kernel debugging that were either out of fashion or were internal technologies other companies didn't have access to. For a long time it was using proprietary source control, proprietary bug/project tracking - you get the idea.
When I switched companies, none of that mattered. Technology is boring and easy to teach. Career skills are important. I was learning to evaluate business needs and customer requests, to triage and plan issues, to design feature, to mentor and lead junior engineers, to think about things like accessibility and internationalization. These are the important things, and they all transfer.
I went on an interview loop, got several offers, and took one at a company working mostly in Java (which I hadn't used before joining that company) doing server-side backend work (which I hadn't done before joining that company). It didn't matter; I was up to speed on the basics in a couple of weeks, fully productive within a couple months, and a valued member of the team by my first performance review.
ex IBM intern here
Watson is just a marketing name for a bunch of out of the box ai solutions, nothing too fancy, just chat bots, CV, and nlp apis and such
it's not bad either tbh imo
“We thought it would be easy, but it turned out to be really, really hard,” said Dr. Norman Sharpless, former head of the school’s cancer center, who is now the director of the National Cancer Institute. “We talked past each other for about a year.”
If you ever heard Ginni Rometty speak at an interview or conference, it’s clear she was in way over her head. She had no real understanding of anything AI or cloud, only that she needed to have AI and cloud and a ready offering for any new buzzword: blockchain, quantum computing, etc.
Under her watch, IBM acquired a cast of companies and tried to assemble a cloud computing arm with little time to do it. While some of Watson’s core tech was born in house much was bolt on acquisitions and hodgepodge marketing to push the narrative of a cohesive whole. The developers gave it a chance, and it mostly sucked. It wasn’t aimed at them after all—-it was aimed at their managers.