Your code is just a cache
We keep hearing software engineers cry wolf: software engineering is dead, and coding is being solved. I don’t think that’s the right way to look at it.
The truth is, it was never about the code. Source code does not matter that much on its own. We glorified it, praised it, and built palaces of rules around it. We even took pride in constructing castles of abstractions from design patterns that we gladly “borrowed” from other industries to formalize code and give it structure. But what, essentially, is code?
The snapshot
Code has always been an artifact: a snapshot of reasoning captured at a particular moment in time. Think of all the times the business moved on, or your perception of it changed because you learned a new paradigm or deepened your understanding of the business domain. How did the code feel then? Uncanny. Misaligned. As if something were wrong. We called it a code smell, technical debt, or simply drag. That feeling nudged us to change things and make them… better? I wouldn’t call the result better. I’d call it a closer match, or simply, more aligned. Alignment was the measure all along.
But there are great repositories out there
Yes, you might now make the But there are great repositories out there argument. And clearly, there are! They preserve valuable snapshots of reasoning, reflect extended thought processes, and express ideas in particularly interesting ways. Do you know what has similar qualities? Books! Their authors stand on the shoulders of giants, building new levels of understanding. Yet what gets captured is not the reasoning itself but just a snapshot. A picture taken at a particular moment in time. Creating that picture can take a year (hopefully no more!), but it is still just a snapshot.
The new era
As reasoning becomes faster and cheaper every day, the value of these snapshots decreases over time. It may still be worth keeping some of them, referencing them, or using them as starting points. Eventually, though, if reasoning capabilities become powerful enough, why bother considering these snapshots at all? Perhaps they remain useful as part of a training set, helping improve those reasoning capabilities, but beyond that? They are just snapshots, fading with each passing day.
The code? The code is just a cache.