3.1 Perception and reading

Nobody can read this panel from where they use it. The panel is operated by an employee standing at arm’s length, but it is commanded by the employer from across the laboratory, typically from beside the vat, typically while…shouting and pointing? From that distance, nothing on the wall is legible: not the stencils, not the needle, not the scope traces. The employer is issuing instructions to an instrument he cannot see, and receiving confirmation by watching whether his employee’s arm moves. The arm is the display. This is not a reading distance for which anything on the panel was sized.

The panel has no “normal.” A good panel can be check‑read: a glance tells you whether everything is where it should be. This panel has one quantitative instrument, two qualitative ones, and no way to tell at a glance whether the system is idle, armed, running, or faulted. Every housing is the same red whether the unit inside is on or off. The only difference between a panel that is doing nothing and a panel that is about to petrify someone is the angle of one handle among several identical handles.

The environment fights the displays. The panel is mounted on a wall of illuminated glass brick, which means the brightest thing in the operator’s field of view is the background, and the instruments are silhouetted against it. The few internally lit elements (oscilloscope, one green flashing lightning thingy, some triangular buttons) are competing with a backlit wall and intermittent steam. State is encoded in lever position, a dim gauge, green light, or in nothing. There is no redundant coding by shape, position, or text for anything that changes.

3.2 Information hierarchy

Everything has the same visual weight. A closed‑circuit television, a power‑input control, and a device that turns people to stone are presented as similar red boxes of identical size with similar stencils, arranged in no particular order. The hierarchy of consequence—the thing that tells a stressed operator which of these to look at first —has not been designed. It has been powder‑coated and danced around.

Spatial memory has been sabotaged. Two controls that undo each other are in different rooms. The controls used together in the laboratory are spread from the ceiling (fluid cylinders, in a chandelier?) to the floor (a handwheel) to the wall (everything else), with a cross‑room walk in between. The layout does not reflect the order in which anything is used, and there is no evidence it was ever stable—the employer moves between panel, vat, platform, and elevator with the confidence of a man who has never had to find a control in the dark.

The motion budget is spent on steam. The largest, most eye‑catching moving element in the laboratory is vapour escaping from pipes near the reactor control, which carries less than no information at all—it obscures the interface. The scope traces in the oscilloscope move continuously whether or not anything is happening. The one moving element that does carry information—the needle—is the smallest and dimmest. Attention througout is being drawn to the wrong things.

3.3 Alerting philosophy

There isn’t one. I looked for a warning tier, a caution tier, and an advisory tier. I found a green lightning bolt that flashes for roughly for a fleeting moment when the most dangerous device on the wall fires. Green. For the weapon. The only escalating signal in the room—the hum that rises with the power of the electromagnet—tracks power, not hazard, and it is heard by everyone in the castle and directed at no one.

Nothing latches. If the operator looks away during the flash, the panel retains no record that the Medusa (note, that is its name, as confirmed by the wall stencil) was ever actuated. There is no “fired” indicator, no count, no armed/safe state. The panel cannot tell you what it did ten seconds ago, which means it cannot tell a new operator anything about the situation they have just walked into.

The nuisance rate, the amount of distracting noise compared to signal, is…and I’ve never had to write this…100%. The oscilloscopes wave whenever they are on and mean nothing in particular. The steam vents whenever the reactor is stepped up and is treated as normal. Employees have been trained, by the equipment itself, to ignore every visible signal the equipment produces. When a genuine fault occurs it will present as exactly the kind of light and vapour everyone has learned to disregard.

Every signal is ambiguous. Not one indication on this panel says what is wrong, where, or what to do about it. The closest thing to a fault annunciator in the building is the monitor going to static, which is the least important device in the room and the only one that admits when it has failed.

3.4 System state and modes

The mode annunciator is a person. The panel supports at least five distinct operating modes—monitoring, vivification, magnetic retrieval, freezing, petrification—and no element of the panel indicates which is active, whether they are mutually exclusive, or what happens if two are engaged at once. As it is, mode transitions are announced by the employer shouting the name of the next mode across the room. When the employer is not shouting, the current mode is unknowable. When the employer is singing and dancing, it is worse.

There is no “I have the controls.” Over the course of the inspection, four different people operated this panel: two employees, the employer, and a houseguest with no training and, at that point, no shoes, no protective gear, and no, uh, clothes. At no time was there any indication of who was in control, any handoff protocol, or any lockout to prevent one operator’s action from stacking on another’s. Aviation solved this with three words and a rule about who says them. This laboratory seems to believe they “solved it” with an nod.

Trust is uncalibrated in both directions. The panel displays nothing about what the equipment perceives or intends: not what the magnet is about to pull, not who the Medusa will resolve to, not what the reactor is even doing. As a result, operators have no basis for confidence, yet use the equipment with total confidence anyway, I guess because…it has always worked? This is the trust profile of someone who has never seen a failure, which points to a lack of training that is unfortunately not part of the scope of this document.

3.5 Controls

Prior reviewers have covered mapping and switch selection in detail and I refer the employer to their notes. Two system‑level findings remain.

Negative transfer is designed in. The panel borrows from at least four control lineages — theatrical lighting, automotive shifters, telegraphy, and the slot machine—and every one of them brings habits that are wrong for its neighbor. An operator who has driven a car will treat the guided channel as a gear shift, a mode selector. An operator who has run a lighting board will treat the knife switches as a persistent state, when one of them is momentary. One of those control attitudes will be wrong at least half the time. Conventions were not chosen so much as randomly distributed.

No operator has been qualified. The panel is used by anyone who happens to stand nearby. The houseguest operated the monitor on first contact without instruction. Nothing on the panel distinguishes controls an untrained person may touch from controls that literally petrify people. The envelope runs from floor to cupola, the operators range from a man on his knees to a man in platform heels on a ramp, and no one has asked whether any of them can reach or see the things they are expected to reach and see.

3.6 Failure and degradation

A dead panel and a live panel look the same. With the scopes off, the wall is a set of red boxes with handles. There is no power‑on indication for the panel as a whole, no indication that any individual unit has lost supply, and no way to distinguish “off” from “failed”.

Every switch displays commanded state, not actual state. A handle in the down position tells the operator the circuit should be closed. It does not tell them that it is. A welded contact, a broken linkage, or a burnt‑out unit will leave the handle exactly where it was and the operator none the wiser. There is one feedback loop on the wall that closes on the effect rather than the input. It is a dial associated with the Reactor Power Input, and that is wholly unlabeled. More on this below. 

Load and limit are invisible. The entire panel appears to draw from a single reactor whose only readout is a needle on an unscaled dial with no red zone, no maximum, and no indication of what is drawing from it. The employer stepped it up “three points.” Three points toward what? He did not answer when I asked. He just sort of glared intensely.

The effect lags the action. The magnet reaches full power and then does nothing until a fourth, separate, unmarked action is taken. Delay between control and effect is how operators learn to overshoot, and overshooting is how you wind up with  a wheelchair through a wall and the full force of the Americans with Disabilities Act shutting down your laboratory.

3.7 Regulatory and evaluation

No mandated indications are present. There is no high‑voltage marking, no energized‑equipment marking, no emergency stop, and no means of de‑energizing the panel from the operator’s position. I am not aware of a standard symbol for petrification, and I would settle for a standard symbol for anything here.

Nothing has ever been tested. The panel was, as far as I can determine, built to be looked at, and the only evaluation it has received is an audience. I found no evidence of occlusion testing, of any attempt to measure what an operator can read in a single glance, or of a single rehearsed failure scenario. The one unplanned failure I observed—the specimen leaving the vat under its own power—was handled entirely without reference to the panel, which is the most honest review of the panel anyone has yet produced.