An inky thought on programming

and a big challenge

I was talking this week with someone who has spent the last decade or so helping people learn how to code. A very good thing to do - help folks get onto the lower rung of a lucrative ladder, start their career in tech.

Except…is it still? While we were talking, it occurred to me that the practice of writing code might, to some degree, be compared to the practice of calligraphy and transcription in the era before the printing press. Back then, it was super valuable to be even able to write at all, so training and supporting someone to slowly copy out a book, accurately (and with ornamentation) was a useful thing. It was hard to put letters on the page, if you could do it well, you had value.

And then…what mattered when the printing press arrived suddenly wasn’t putting letters on the page, it was putting ideas on the page. Letters were the easy part (well, sort of - I suspect with AI we are still at the "printer’s devil" stage more than than the desktop publishing stage, where you are pretty much covered in ink if you want to dabble in printing at all). Suddenly there weren’t monks transcribing things (bad for monks) but there were a whole bunch of new ideas, political movements (like the reformation and all that followed), universities and so on. Knowledge (and even reading!) went from being something very few people had (and most people had no use for reading) to a fundamental part of our social and political lives.

Something similar will likely start to happen with code, coding, and the idea of programming at all. The idea of human readable source code might start to seem as quaint as a handwritten book. There will be a lot more code being produced, by a lot more people, and the skills needed to propagate it will get simpler and simpler (from lead cast type, to lithography, to desktop printing). And look at this - I am creating some text and sending it everywhere, no paper or printing involved. The code equivalent is that code is produced and used without anyone really understanding that it is, unless they really stop and pay attention.

In the near term, this leads to some real challenges. Right now, advanced models can usually do the work of entry level engineers (and I’ve heard from lawyer friends, the same is true in the legal world - models are a fair replacement for first- and second-year graduates, who aren’t usually that valuable until the third year). It’s tempting not to hire for those roles (no need for ink grinders now!) but we need them to train folks for the later stages of the career.

It’s an open question how far this analogy goes. Do we eventually lose the “transcription” class (monks, nee coders) and get a more specialized printer skill that fewer people have? How do we bridge the gap as new capabilities emerge? There were plenty of “print entrepreneurs”, starting with Martin Luther, and the penny newspapers in the 19th century, and so on. So it’s likely that there will be lots of ways to take advantage of this shift.

Or is code more robust than print? Is the analogy false?