Live instrument and physical UNI model
How to read this page
Three ways to read this page. Precise is the document itself, exactly as it is written in the repository. Plain and Clear were written for this website to help you meet that document — they are about it. They are not it, and they are not evidence.
A laboratory built around the bacterial flagellar motor. It holds a deterministic reduced model of the motor, analysis of recorded single-motor events, and a cross-study parity programme. Alongside those sit the scientific gates the work has to clear, and independent audits of both the model and the repository around it. The framing throughout is hierarchical active inference.
It is for a reader with a scientific interest, and especially for one who has come to check whether a model fit has quietly become a claim about biology. The laboratory's central discipline is a labelling one: every visible layer carries exactly one class — recorded observation, structural reconstruction, reduced model, or physical teaching analogue — and those classes may not be blended. Behavioural observations of one species are held apart from structural work on another, so that nothing on the page can read as a single measured specimen.
Start with the Living Science Walkthrough, which sets out those classes and the truth contract they belong to. Then the scientific and mathematical contract, then the parity gates, which state what would have to hold before a parity claim could stand.
What it is not: a claim of biological parity. The walkthrough is explicit that the release does not turn a model fit into a biological identity claim, and full biological parity is recorded as false and printed as false. Passing this repository's software tests is necessary here and is not the same thing as agreement with a living motor.
Your browser cannot switch reading levels, so the document itself is shown.
Precise — the source document
This is the document. Rendered from the repository at the commit above, with nothing rewritten for the web. A gate re-renders it on every deploy and fails the build if a single byte differs.
Serial measurement contract
The laboratory reads UTF-8, newline-delimited JSON at 115200 baud. Each complete line is validated before crossing the Markov boundary.
Required fields:
{
"t_ms": 1250,
"ligand_uM": 1.12,
"motor_rpm": 6430,
"rotation": "CCW",
"load_pNnm": 700,
"pmf_mV": 150
}
Optional fields are cheyp_uM, stators, receptor_activity,
prior_angle_deg, and evidence_angle_deg. Unknown fields are retained only in
the raw-frame display and have no model authority.
The browser attaches receivedAtMs independently from device time. A frame
with missing, non-finite or invalid rotation fields is rejected and displayed
as rejected evidence.
Physical mathematical model
The model has eight labeled parts:
- external WORLD-ROTOR;
- inward SIGNAL-CAM;
- PRIOR-GEAR;
- EVIDENCE-GEAR;
- POSTERIOR-DIFFERENTIAL;
- POLICY-GEAR;
- PREDICTION-GEAR;
- outward ACTION-CLUTCH.
A literal boundary plate has only two apertures: observation inward and action outward. There is no direct shaft from the world rotor to the internal belief gears.
Classroom measurement
Rotary encoders or magnetic angle sensors can be mounted in the declared bores. The microcontroller maps encoder angles to prior and evidence log-odds, emits a serial observation, and receives no hidden world state from the browser.
The initial mechanical identity is:
theta_posterior = theta_prior + theta_likelihood
The screen computes the exact Bayesian identity and shows encoder error, backlash and calibration residual. The v1 printed differential is an educational mechanism and requires physical validation before it can be called a mechanical calculator.
Fabrication gates
- Print a bore/shaft/gear tolerance coupon.
- Measure actual dimensions with calipers.
- Record fit, backlash and printer/material settings.
- Regenerate with corrected clearance; never sand away evidence silently.
- Print each gear separately and confirm free rotation.
- Assemble the boundary and verify that no undeclared crossing exists.
- Install encoders and calibrate zero, direction and angle scale.
- Compare physical and on-screen log-odds across the full safe travel.
- Preserve the calibration dataset and declare the valid uncertainty band.
- Use adult supervision for small parts, electronics and rotating mechanisms.
The OpenSCAD file is a conversion-ready parametric starting point. It is not a certified consumer product or a promise of successful printing on every machine.
sha256 98d0e3f49534c5c9 — of the original file, so what was ingested stays checkable.
Plain — written for this website, not the source document
Two things meet here: a real instrument sending readings into the software, and a printed, hand-turned model of the same mathematics.
The first half is a strict contract for the wire. Each line of data is checked before it is allowed to cross the boundary into the belief part of the system. A line with a missing or impossible field is rejected and shown as rejected, rather than quietly used. Fields the contract does not name are kept on screen but given no authority over the model.
The second half describes the physical object: eight named parts, and a boundary plate with only two openings, one for observation coming in and one for action going out. There is no direct shaft from the world to the belief gears, and that absence is the point.
The page is careful about what it is not claiming. The printed mechanism is a teaching object, and it requires physical measurement before anyone may call it a calculator.
Plain · written 2026-08-01 by claude-opus-5 · not yet checked by a person · about the document whose sha256 is 98d0e3f49534c5c9
Clear — written for this website, not the source document
It begins with the serial contract. The laboratory reads one complete line of text at a time from a device, and every line is validated before it may cross the boundary. A required set of readings covers time, ligand, motor speed, rotation, load and a force value, with a few optional extras. Anything the contract does not name is retained only in a raw display and has no model authority. The browser stamps its own arrival time rather than trusting the device clock, and a frame with a missing, non-finite or invalid rotation field is rejected and shown as rejected evidence.
Then the physical model: eight labeled parts, running from an outer rotor turned by the world, through the belief and policy parts, to a clutch that acts back outward. A literal plate divides the two halves and carries only two apertures, observation inward and action outward. There is no direct shaft from the world rotor to the internal belief gears.
Then classroom measurement. Angle sensors can be mounted in the declared bores. A small controller maps the angles to log-odds, emits an observation, and receives no hidden world state back. The screen computes the exact identity and displays the encoder error, the backlash and the calibration residual, so the gap between machine and mathematics is shown rather than absorbed.
Then a numbered list of fabrication gates. Print and measure a tolerance coupon, record the fit, and regenerate with corrected clearance rather than sanding away the evidence. Confirm each gear turns freely, and check that no undeclared crossing exists. Calibrate the sensors, and compare physical and on-screen values across the safe travel. Keep the calibration data with its uncertainty band. And supervise children around small parts.
The closing paragraph carries the page. The printed differential is an educational mechanism, and it requires physical validation before it can be called a mechanical calculator. The design file is a starting point, not a certified consumer product and not a promise of successful printing on every machine.
Clear · written 2026-08-01 by claude-opus-5 · not yet checked by a person · about the document whose sha256 is 98d0e3f49534c5c9