There is a well worn analogy in software that writing code is like wood working. We’ve all been crafting these chairs by hand for decades. This analogy predates generative AI, and it’s been a cliche that software developers retire and inevitably pick up wood working as a hobby, but I digress.

With the advent of AI coding agents there is now lots of chatter about how it’s industrializing software development. Far faster and cheaper than an artisinal chair, one can now go out and buy one of the millions of IKEA Poäng chairs that whiz out of the factory every year! The analogy is comfortable and feels familiar, but the problem is that it is wrong… sort of. It’s right, but it’s also wrong.

It’s wrong because the nature of software is different than the nature of chairs.

Yes, it’s atoms versus bits again

Let’s look at a second order effect. Industrialization of chairs made the world of interior design more uniform. Before industrialization every chair was different even if you didn’t want them to be. After industrialization there are now more chairs, but many of them are copies of one another.

Now let’s consider software. The internet has already made software very uniform because of the classic refrain that the cost of replicating and distributing software is virtually zero. It’s just digital information after all. Looking at the security landscape, one could even argue that software today is too uniform, but again I digress.

Now with AI tools there are vastly more lines of code being generated (organically or mechanically) globally every day. The industrialization analogy feels okay when only looking at output rates, but the differences start to appear when one considers that many more people now are generating code than there were before. A lot of this code would not have been written without these new tools for various reasons. Essentially I think we may be at an inflection point where we start to see the uniformity of software decrease. Is it just a Cambrian slopsplosion? Maybe, but it still looks very counter to what happened with physical industrialization.

And that’s the weird wrinkle in all of this that I realized while making coffee this morning. In some respects, yes, generative AI is sort of like the industrial revolution, but in other ways it’s not. It actually might do the opposite and result in more “hand-crafted” software not less. So if we do accept this analogy we also need to accept that software is a bit peculiar. If anything it’s always been peculiar.