"How to continue making kerosene lamps on the eve of electricity"
Roger Stringer / April 18, 2023
2 min read
David Heinemeier Hansson (DHH):
The recent and rapid advance of AI has rightfully giving many in software real doubts about the future of their profession. I'd probably still wager that the fears are overstated – that we also got prematurely euphoric about the imminent prospects of self-driving cars – and that AI generating code is different from it evolving existing systems. But I wouldn't want to bet the house on it. This might just be The Big One.
That uncertainty, whether we truly are on the cusp of a total transformation, is a surprising novelty to me, despite thirty years in software. We've been through so many breathless predictions about how this or that advancement in software would simply "Change. Everything. Forever." that a certain pessimism would be perfectly reasonable.
I've seen, for example, at least three hype cycles of No Code movements come and go over those thirty years, and none of them ever fulfilled their grand promises. Even the two properly big transformations we've had in that time, the internet and mobile, hasn't made the kind of programming most people do today look all that different from the kind of programming done in the 90s.
I like to imagine that all of us in software development, as it looks today, are busy making beautiful kerosene lamps in all shapes and sizes. We're improving the burn efficiency. We're finding ever clearer forms of glass to let the light through. We're tinkering with a formula that's been around for a long time. But as we do, Edison and Tesla are busy inventing electricity in the other room.
I'm sure it wasn't clear at all, even to Edison and Tesla, when exactly the switch would be flipped on that final lightbulb moment. And meanwhile, the world still needed to see in the dark, so the lamps of old were still necessary. That just might be us, right now, lighting the path to the future with the ways of what could soon be the past.