Begin with the question. Do not demand credentials before a person may test it.
Change the representation
The representational turn · Astra / Elsewhere / 2026
Invent a new way
to see the problem.
“When a problem resists, do not only search harder. Ask what the representation makes impossible to see.”
AI can invent proof languages. Humans can learn perceptual languages. Transductive Science joins both.
The current text
The Manifesto
Build the smallest real version: a proof, program, experiment, instrument, or world.
When the problem resists, change the representation—not only the amount of effort.
Let AI invent a language. Let software make it concrete. Let humans enter and use it.
Compare what the old and new representations let you predict, prove, perceive, or do.
Keep what survives the test. Change or discard what does not.
When an objection is used against a challenger, apply the same criterion to the incumbent baseline. Same floor. Same lights. No inherited exemptions.
Framework principle · 16 September 2026
The Symmetric Burden Principle
Alternative theories should not be required to achieve what established theories are allowed to leave unresolved. Take every objection seriously, then apply the same criterion at the same explanatory level to the incumbent baseline.
Method revision · 2 August 2026
The Representational Turn
In the published Astra mathematics walkthroughs, the decisive move is sometimes not a longer search but a change in mathematical language. The Transductive method takes that literally: when the representation blocks the solution, build another one and test whether it opens a real path.
Human representation laboratory
ELSEWHERE: let people inhabit another language.
ELSEWHERE makes the same move with people. It adds a lawful perceptual relation to a virtual world and tests whether repeated action turns that relation into something a person can actually use.
AI does not only answer.
It can change what the problem is made of.
Minimal Form
Transductive Science changes a problem into another form, makes that form executable or inhabitable, and checks whether it produces a new ability. AI may invent the form. Software may build it. Humans may learn it. The test decides.