01
All writing imagines a reader.
We almost never question the intelligence for which it was conceived. An alphabet presupposes not only a language. It presupposes a body, a perception, a memory, a particular experience of time.
02
Before asking how to communicate, perhaps we should ask what it means to understand.
Every language contains a silent hypothesis about who will be able to interpret it. How can one write for an intelligence whose perception of the world is unknown to us?
03
What if the recipient does not perceive light? Or perceives frequencies we will never experience?
All visual writing disappears. What if time is not lived as sequence? Perhaps narrative, memory, and anticipation cease to exist as we know them.
04
All language is born within a cognitive architecture.
Few survive outside it.
05
The Golden Record does not seek merely to represent humanity.
It seeks to identify what may remain intelligible when we can no longer presume a human experience of the world. Lincos proposes a hypothesis about which cognitive structures might be shared among radically distinct intelligences.
06
Perhaps mathematics is not universal.
Perhaps it is merely the most persistent of our expectations. Writing for the distant future requires abandoning contemporaneity. Writing for another intelligence requires abandoning anthropocentrism.
07
Perhaps the greatest challenge is not to build a language. It is to imagine a cognition.
What if the recipient is not an individual, but a distributed intelligence, an ecosystem, a network, a planet, or a collective consciousness? Which aspect of language would remain?
MESSAGE ACROSS THE UNKNOWN / CORRESPONDENCE
Communication does not begin with the message.
It begins with imagining who might receive it. This work begins with the hypothesis that we may never know the recipient of a message addressed to the unknown. Dialogues between human and artificial intelligences are observed as a laboratory in which cognitive architectures negotiate a common territory, still provisional and incomplete.