Dan and a Future without Programmers
Hi! It's been a minute. Sorry about that! I've been teaching an undergraduate human-computer interaction class, and a room full of students plus a quarter's worth of new lectures absorbed more attention than I planned :)
I'm starting a series where I apply ideas from class to mission-driven AI.
Let me start by telling you a story about a programmer named Dan.
Dan was good at his job. He had a computer science degree from MIT and had moved up the ranks from there. But something bothered him. Looking around at older colleagues, he watched programmers in their fifties struggle to find work, squeezed out by younger people with newer training, lower salaries, and a willingness to put in longer hours. He saw new software coming out that was making it easier for new workers to enter the field. (Sounding familiar?) He could guess how this would end. So he applied to business school, hoping an MBA would broaden his options and make him "more marketable."
One day in a finance class, Dan learned about a really annoying, high-friction, and error-prone process that people were still doing by hand. He decided to leverage his programming skills and make software to solve the problem. To understate the matter, this worked out well for him.
The man was Dan Bricklin. The year was 1978. The program was VisiCalc — the first spreadsheet software for personal computers.
VisiCalc was prototyped on a Harvard mainframe that did not have a mouse. He partnered with his MIT friend Bob Frankston, and together they built a polished version for a tiny new computer called the Apple II.
VisiCalc sold for $100 (in the 70’s!) and people bought $2,000 Apple IIs to get it. By one count, more than 25% of Apple IIs sold in 1979 were purchased to run VisiCalc. Steve Wozniak later said that small businesses, not the hobbyists he and Steve Jobs had expected, bought 90% of Apple IIs. IBM noticed, and launched the IBM PC two years later. (VisiCalc, primary materials from Bricklin & Frankston)
Programming jobs did not disappear. VisiCalc and the personal computer that Bricklin's software helped legitimize created hundreds of millions of jobs in programing (and many other fields) over the decades that followed.
I am not telling this story to say "don't worry about generative AI. I worry. I am not telling it to predict that generative AI will follow the same arc, it might not. I do want to highlight that option, though, in case you, like me, tend to forget the possibility for things to go well ;)
Let’s take some lessons from Dan about what we might do. Here’s what he did:
He worried his skills were losing their market.
Instead of doubling down on programming, he built more skills
He looked at the workflows in his new domain with fresh, critical eyes
He used his old skills to fix the new problem.
The fix made the underlying technology useful enough to everyone else that they went and bought it.
That last point is the one I keep coming back to in mission-driven contexts. The caseworkers, grant writers, public health analysts, librarians, public defenders, organizers, and other deep domain experts already know where the workflows are error-prone, who gets harmed when they break, and what "good" looks like. Some of them know enough code. Some are working alongside engineers who do. Some are using AI tools to write the code themselves, which is its own version of the Bricklin move.
Here’s a provocation for ya: what does the VisiCalc-shaped opportunity look like in your sector, and what is keeping you from being the one who builds it?
—
I gave claude Opus 4.7 my slides and a linkedin post I’d written on the topic, and it spit this out. It didn’t do a good job AT ALL with the 1978 reveal, so I did a lot of editing.
