A label is not only design.
It is a claim about reality.
That does not mean every bottle, bag, box or page must carry the full biography of everything inside it. A label cannot become a book. People still need short names, simple categories and ordinary ways to buy things without performing an investigation every time they are thirsty.
But a label should not be allowed to extract one attractive part of a product and pretend that part describes the whole thing.
That is the rule.
Do not label the orange if the system hides the orchard.
Imagine a future orange soda.
The front says orange.
Maybe it says Mediterranean orange. Maybe it shows a hill, a tree, sunlight, a bottle sweating in someone's hand. The buyer receives a small story before tasting anything.
But the meaningful structure may be somewhere else.
Where did the orange come from?
Was the aroma extracted from fruit, assembled from compounds, or generated through a process that only imitates the memory of fruit?
Was the recipe made by people, by software, or by a system optimizing toward cost, craving, shelf life and market capture?
Who owns the production chain?
Where was it bottled?
What part of the work was local?
What part of the value left the place whose image is being used?
What parts of the system are human craft, automated process, licensed formula, imported input, platform coordination, distribution leverage or brand theater?
None of those questions automatically make the soda bad.
That matters.
The point is not to turn every label into a punishment. The point is to make the relevant structure readable enough that markets do not become machines for hiding their own conditions.
A partial truth can still mislead.
Made with orange can be true while the product's real power sits in the recipe system, the bottling contract, the logistics network, the pricing strategy, the attention budget or the ownership layer.
Local can be true for one step and false for the structure.
Human made can be true at the finishing point while the recipe, inputs, demand signals and production schedule were shaped somewhere else.
AI-assisted can mean a small tool in a human kitchen, or it can mean a system where human taste has become the decorative surface of machine-optimized production.
Those are not the same thing.
A readable market needs better distinctions.
Not because consumers are pure.
Not because every buyer will make the right choice.
Not because disclosure alone fixes power.
Readable information does not make a market fair by itself. But unreadable information makes capture easier. If ordinary people cannot tell what kind of system they are participating in, consent becomes theatrical. Choice becomes a gesture performed inside someone else's structure.
This is where labels become infrastructure.
They decide what a society is allowed to notice at the point of exchange.
If the label only names the attractive ingredient, the buyer sees ingredient romance.
If the label makes production mode readable, the buyer can at least ask a better question.
Was this grown, mixed, designed, bottled, delivered and governed in a way I can understand?
That question will matter more as AI enters food, retail, logistics, pricing, design, advertising and production. The issue is not whether AI food should exist. It will exist in many forms. Some of it may reduce waste, stabilize supply, lower cost, improve nutrition or help small producers do work they could not otherwise do.
But AI-produced and human-produced goods should be able to exist side by side without pretending to be the same thing.
The future market does not need one moral label stamped over everything.
It needs a simple public route from the surface to the structure: what is in this, where did the important inputs come from, how was it designed, who controlled production, where did the value go, and how can a buyer complain, repair or leave?
Those answers do not all belong on the front of every package. But a readable market should know how to make them available without making every buyer become an investigator.
That is the difference between marketing and public readability.
Marketing asks:
What story will make this easy to buy?
Public readability asks:
What structure must remain visible for the choice to mean anything?
The hard case is not the obvious lie.
The hard case is the beautiful partial truth.
The orange is real.
The orchard is hidden.
The label is legal.
The system is unreadable.
That is where the rule begins.
Do not label the orange if the system hides the orchard.
This is not a finished certification scheme.
It is a test.
When a product presents one charming part of itself, ask what larger structure that charm is being used to cover.
If nothing important is hidden, the label can breathe.
If the attractive part is being used to distract from the system that actually produced, controlled or captured the value, the label is not only branding.
It is camouflage.
Source Boundary
This piece is a public-note translation of an internal protocol seed from A Notebook for Society. It does not claim that existing labels are uniformly deceptive, that AI production is inherently bad, or that one universal front-label standard would solve market capture. Any later real-world examples should be source-checked as separate claims.
