...
Generating An Infinite Number Of Digital Products? Not Quite Impossible...

Friday • October 28th 2022 • 11:01:07 pm

Generating An Infinite Number Of Digital Products? Not Quite Impossible...

Friday • October 28th 2022 • 11:01:07 pm

“Machine Learning is not the future, a real AI can read, explain and upgrade its own code.” – Abraham Lincoln


Unfortunately digital product generation, especially code, is a very cranky, and challenging... challenge.

It is too early to have any good news, but please read on…

Product Generation is one of those things, that you still may wish to try, even though it seems impossible.


Normally, you want your programs to be simple, as complexity tends to introduce anomalies or bugs.

But there is an exception, when you write a complex generator…

That generates simple, beautiful, and bug free programs.


A code generator, never leaves the computer it runs on.

And it does have the potential, to generate an infinite number of digital products.

But the pathways to get there, are tough as heck.


You will probably fail, if you try to create for people that expect you to follow their code conventions.

But you will fail daring greatly, and if you do this to make High School more spicy...

Then this failure, will also make you a powerful programmer.


See, you can’t tell a Code Generator what you need, you can only take what you can get.

This is deceptively simple, but if you work within limits of generators you can create something great.


A good and easy to understand example, is generating Bootstrap Cards.

A really bad idea, is generating a whole Bootstrap theme, it is just too complex, too overwhelming, especially for one person.

And you have to follow programming conventions for humans, a machine has it own way of creating Themes (similar to Bootswatch)...

And this would require that you generate, an alternative to Bootstrap.

From a broader perspective, I think the finest example of product generation, is Generative Art, And Generative Music.

A png or jpeg, or an mp3, is pretty much good for 5 decades.


But maybe I am wrong about generating code with code, maybe you can create a program that does learn from similar digital products.

Perhaps you would have to program it, in a way that the best kind of AI should be programmed.

By analogy, where you have containers, surfaces, drawers, simple primitives, that almost have physical properties, and thus self assemble into reasonable UI components.

Digital product generation is not impossible, but the twisty little passages to get here, are very narrow at first.

Artwork Credit