The idea of those barcodes is to maximize the reliable, storable information, regardless of format. If you make the information bytecode for a programming language, people would be able to get mobile phone apps from soda cans without needing to program in an esoteric programming language.

Using something like piet doesn't maximize reliability (error correction) and thus isn't good for that domain.

yet.

No.... Not at all.

Sorry if I'm misunderstanding your reply due to its brevity, but again, error correction is the primary concern of these barcodes, and given that they are good error correctors, you can put what data you want within them, be it an underlying visual programming language or text.

If by "yet" you mean your proposal is to expand on these visual languages by making a programming language + error correction system, you're not adding any value that didn't exist before, it is the same thing as what already exists.

And besides, it's better to just give the user a URI and then they can download the application from a more robust location.

By "yet" i meant "at this moment in time you may be correct, who know's what the future will bring".

I agree in principle with your points, but my point was "one day I can see something like this replacing the space where the 3D barcodes are on cans now"

That's actually not such a crazy idea.....that may be an idea with legs.

I didn't think it was crazy. Although for complex applications Piet needs bigger and bigger images... but then most things start that way.

That will make barcodes a whole new medium to exploit software vulnerabilities - I can't wait to see the first barcode virus.