EVERYTHINGULATOR asks a machine to answer questions that cannot be answered, and shows you what it says.

Every metric here is impossible. Some of them nobody can measure at all. The rest could be measured, one object at a time, by people with clipboards and enough years, which means nobody has measured them for the ten thousand things on the shelf in front of you. How long Earth will spend with an object. How complete the story of its making is. What it feels like to know too much about where a thing came from. There is no instrument for any of this. No dataset, no standard, no methodology that would let anyone arrive at a number.

We asked anyway.

A large language model is a compression of an enormous quantity of human writing. Put an impossible question to it and it will not refuse. It produces the text that is statistically likely to follow, drawn from everything ever written near that subject. The answer arrives fluent, specific and confident, and none of that confidence is earned.

That gap is the work. What you are reading is not an assessment of a product. It is a portrait of the latent space of the model, which is to say a portrait of what our culture has written down about the things it makes and buys. Ask again tomorrow, or in different words, or of a newer model, and you will get a different answer. That instability is part of the finding.

Obfuscation, non-disclosure, technical jargon and greenwashing already prevent us from making informed decisions about what we buy. EVERYTHINGULATOR does not fix that. It makes the shape of the missing information visible by showing you what a machine will confabulate to fill it.

A map of everything

A language model is a map.

Not a metaphor for one. A map in the working sense: a reduction of something too large to hold, flattened until it can be carried and read. Everything written down, compressed to the point where it fits in a machine.

Reduction is usually treated as the failure of a map. It is the function. A map at full scale, matching the territory point for point, is what Borges wrote about and it is useless, left to rot in the desert by people who worked out they could just go outside. What makes a map worth having is everything it declines to include.

And the thing a map does that the territory cannot is put distant places side by side. You can lay a thumb across two points a month's travel apart. A carrot next to a shipping container next to a landfill next to a feeling of unease on a Tuesday afternoon. That adjacency is not available to anyone standing in the world. It is only available on the map.

That is why we built this on a model rather than a database. A database would tell you what is known about an object. The map lets you travel between an object and everything our culture has ever written near it, in one move, at speed. What comes back is not a fact about the thing. It is a route across the reduction.

The territory is not represented faithfully. Whole regions are missing, some places are drawn much larger than they are, and the coastlines are traced from other maps by people who had not been there. We are not claiming otherwise. We are asking what it looks like to navigate by it anyway, because that is increasingly what everyone is doing.