From Programs to Source Code, and Now to Prompts

Table of contents

Shareware years

In the early internet years, shareware was everywhere. A single developer could make a small tool or game, put it online, and let people try it first. If you wanted the full version, you paid for a registration code. Sometimes payment was optional and the author just left a donation link.

What you got in that model was a binary program. You could run it, maybe rely on it every day, but the source code stayed hidden. For most users, software was something you used, not something you could inspect.

Open source years

Then the center of gravity moved. Open source made the code itself part of the product. Sites like GitHub lowered the cost of sharing, reading, and reusing software. Instead of downloading a finished black box, you could open the repo, see how it worked, change it, and send the changes back.

That was a real change in how software circulated. The value was no longer only in the executable. It was also in the ideas, structure, and implementation details sitting in plain text.

Prompts as a new layer

Now I keep seeing a third pattern. Some repositories barely contain code at all. They are mostly prompts, maybe a README, maybe a few examples. The “artifact” being shared is no longer the program or even the source tree. It is the instruction set that can regenerate the program on demand.

That feels new, even if it is still rough around the edges. You can think of it as a JIT (just-in-time) way to make software. Instead of distributing a binary or a repository, you distribute a recipe and let a model produce the code when someone needs it.

One last question

We used to turn ideas into software through a chain of tools, files, and manual work. Now some of that chain is collapsing into a prompt box. I do not think code is going away any time soon, but I do think the unit of sharing is changing again. The part I keep coming back to is simple: if the prompt becomes the product, what exactly are we shipping?