Nokia had the budget, talent, and urgency to fight the iPhone. One hidden bottleneck shows why AI won’t make companies faster unless they rethink the work.
Thanks, MidJourney. Good enough.
Here’s the passage I found suggestive:
Compiling is where the code that a programmer writes gets turned into something the phone can actually run….The 48 hours were for one programmer’s piece. The full build, which required gathering code from every team and turning it into a working version of the operating system, took up to two weeks—two weeks to learn whether a change had worked or whether it would break something.
Working on the IBM mainframe at my undergraduate college in 1962 operated on a similar schedule. But for a a mobile phone company, an engineer probably expected a snappier turnaround. I am okay with a two-day cycle, but I am a dinobaby. In 2007, the developers wanted hours maybe? Nope, I am wrong. The Inc. articles dispenses the gospel:
At Google, a comparable build would take under 20 minutes. And Google’s engineers were complaining that 20 minutes was too slow.
And we know what Google has achieved with its mobile phone, don’t we?
The write up includes a useful statement of what I call “workflow time”; to wit:
So why do big companies feel so slow? Usually, it’s not because the people are lazy. We saw already, it’s that everyone all waiting on someone else. And the wait never ends because the parts of a company run on different clocks:
Finance closes the books four times a year.
R&D plans its bets five years out.
Customer support hears from customers every day.
The online store learns something new every second.
Those clocks never line up. But they aren’t meant to.
My view of this anecdote and the comment about “clocks” is that rapid information flows have different impacts on each work task. Online allows the finance department to eliminate handwritten ledgers with spreadsheets. Computational chemists can use software to identify, model, and discard lots of potential chemical combinations.
What happens when smart software becomes available for the discrete work tasks? I have observed three impacts:
The shared understanding that emerges from old fashioned and slow work is eroded. The result is that errors can persist simply because no one is around to provide comments or criticism. The “manager” and leadership are useless because the old slow work methods allowed information to diffuse, some drifting upwards to the top dogs.
The speed with which outputs are delivered overwhelm many people, yes, even those who are MBAs and lawyers. The output from a simple online query (forget AI for a moment) means that most people look at the first few outputs and say, “Hey, I understand.” Sure, these people do.
The arrival of AI adds two additional factors: [a] “Normal” workers don’t have to think, ask colleagues, or doubt what the AI says. It’s expensive. The boss loves it. Go with it. And [b] top flight workers see that AI can allow them to push out the less intelligent colleagues, maybe take credit for something and get a promotion to a cushy job, or start up a side gig because of the free time the AI delivers.
Consequently the Nokia case illustrates that work tasks can prevent innovation. The present day’s struggles with AI help explain some of the perils of AI.
Net net: Organizations will make decisions based on what AI says. Many of those decisions will be be bad, and the impact of those decisions will be evident more quickly. It took time for Nokia to lose its zip. Today those organization-killing effects will surface more and more quickly. Look around and ask:
Why are air line companies struggling with schedules, fares, and customer service?
Why are customers unable to reach a human at many firms?
Why do mixed messages about health flow from different governmental agencies?
Why do those self-driving taxis end up in Cow Hollow?
Go fast and break things means that nothing will work as expected. Just like at Nokia years ago. Toss in AI and the failures will continue to skyrocket. No problem. Just blame the different clocks for the disease.
Stephen E. Arnold monitors search, content processing, text mining
and related topics from his high-tech nerve center in rural Kentucky.
He tries to winnow the goose feathers from the giblets. He works with colleagues
worldwide to make this Web log useful to those who want to go
"beyond search". Contact him at sa [at] arnoldit.com. His Web site
with additional information about search is arnoldit.com.
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
