The idea
Why a pattern in one domain might say something about another.
The same mathematics can form it in both.
In 1952 Alan Turing described how stable patterns arise from the interaction of activating and inhibiting forces: the mathematics behind leopard spots and embryonic tissue. It is not specific to biology: any system with something driving growth and something holding it back can settle into the same family of shapes.
Kongen grows one organism on that mathematics for each domain. When the same pattern appears in two domains that share no data, the collective treats it as something to test: not a coincidence to explain away, and not yet a result.
01Tested, not asserted
Two things looking alike is not evidence. So it has to be tested.
2,000 of 2,000
The one claim here you can check yourself in a minute: one step, scored 2,000 times against the shipped scoring module, came back identical every time for that scorer version. Our cross-domain tests are written down before they run, and a result has to come from a test that could have failed.
Research status, October 2026. Every research result Kongen has published is unverified until it is re-derived on our rebuilt research platform, under tests written down before they run. Results that did not survive review stay withdrawn where they were published, among them the capital-market transfer figures, the data-centre figures (they came from synthetic data), the language-to-locomotion figure, the regime-accuracy figures and the genetics-to-language figures, now being rebuilt under a pre-registered design. Measurements of the shipped scorer, such as the 2,000-run repeat check and the 106 ms median from one latency run, are engineering measurements, not research results, and each current one is stated with how it was measured. Separately, the markets organism’s live trader was halted on 3 June 2026; what it left is held as evidence, not offered as a track record.
02The method
Grown, not weighted.
Its foundation is a quantum-grounded model of awareness that uses the same pattern-formation mathematics found in living systems.
Perception
Each organism extracts the structural shape of signals in its domain: the underlying form, independent of surface detail.
Transfer
Patterns travel through Kron, the meta layer, and organisms in other domains check their own data for the same shape.
Confidence
When organisms find a pattern independently, the cross-domain confidence adjustment rises; when they do not, it falls. Unverified: to be re-derived on our rebuilt research platform.
03The Garden
Thirteen worlds, one pattern language.
Each organism perceives a different kind of data, and they all speak one pattern language. What one learns, the others can read, through Kron.
04The obvious objection
Why not just ask a model to decide?
Because it is the same call you were trying to think twice about, paid before the work starts, and it can answer two ways on two runs.
It costs a call
A model deciding whether to spend a model call has the cost curve of the work it decides about. On every step, the decider becomes a line item of its own.
It waits before anything starts
A model has to generate before the step can begin. A Kongen answer is arithmetic over the input: there is no sampling to wait for.
It can change between runs
The same step can be decided two ways in two runs, which makes it hard to test against, to reproduce an incident from, or to budget from.
Even as JSON, its values are sampled
Something then has to trust them. Kongen’s regime is an enum and confidence a float between 0 and 1: branching on them is ordinary code.
05Receipts
The receipts, in full.
Each one links to where it comes from.
21 to 43% lower cost, estimated
An estimate from list prices as listed on 5 October 2026, not a measured workload: the calculator’s starting position, comparing every step on a frontier model with each step on the tier its regime earns. Shorter requests save less. Move the sliders and you get your own numbers. See how it is calculated.
Halted on 3 June 2026
The markets organism ran a conservative-mode trader live on Kraken from 26 February 2026, and it was halted on 3 June 2026. It is not running today. What it left is kept as evidence, not offered as a track record, and this site makes no claim about its performance. What the markets organism does.