There is a temptation to tidy history from the viewpoint of the present. An old mechanism is labelled obsolete, an old page is rewritten to explain what replaced it, and an old laboratory is converted into an archive. The result is orderly and almost useless as a place.
A physical station behaves differently. If a better laboratory is built elsewhere, the old one does not transform itself. Its notes still describe what was current when they were written. Components remain on benches. Procedures remain pinned to walls. A machine abandoned halfway through disassembly remains halfway through disassembly.
Rejected by what?
The same rule applies to technology. Rejection is usually contextual. The hoverbike is an unusually severe environment: mass, volume, vibration, heat rejection, rider safety and transient power all compete inside one mobile structure. A mechanism that loses there may still be perfectly competent somewhere that can afford more space, more shielding, thicker thermal paths or a slower operating cycle.
This is why an old CEU branch should not be recorded merely as a failed generator. The useful record is that it failed to justify itself under a particular set of vehicle constraints. Remove those constraints and the branch deserves to be inspected again rather than automatically discarded.
Scale changes the question.
A mobile machine often has to make every kilogram perform several jobs. Waste heat may be turned into thrust or electrical recovery because radiator area is scarce. A stationary or shipboard installation can separate those concerns. Heat can first serve water, atmosphere, industrial processes or inhabited spaces before the final remainder reaches a radiator. A fragile thermal path can become a large reinforced manifold simply because there is room to build one.
The inverse is true for by-products. Something useful on a vehicle may become an unwanted stream in a station installation and require dedicated containment. Neither case makes the substance intrinsically useful or wasteful. The surrounding system gives it that role.
Technology has descendants.
Old designs are therefore better treated as ancestors than verdicts. A later ship generator may descend from an early compact CEU without resembling it closely. Perhaps only the containment method survives. Perhaps the thermal architecture grows enormous while the original conversion stage disappears. Perhaps a power company licenses one useful branch and takes it somewhere the vehicle programme never intended.
This creates another way to read the Station's technical history: not as a straight sequence of versions, but as a branching ecology. Some branches die under one set of constraints, some migrate into different machines, and some remain dormant until a new problem makes them legible again.
Do not curate the ruins.
If Lab 01 is eventually superseded, it should not be updated into a museum of Lab 01. Build the next laboratory and go there instead. The old room remains accessible because it exists, not because it has been prepared for visitors.
That distinction matters. A curated archive tells us what the present thinks was important. An abandoned laboratory preserves what the past thought was important before the outcome was known. It can therefore contain errors, dead ends, forgotten assumptions and trivial fragments that later become unexpectedly valuable.
The Station can acquire genuine old parts only if they are allowed to become old without being continually translated into the language of the present.
This is a documentation rule as much as a piece of lore: preserve superseded technical pages when practical, record why work moved on elsewhere, and avoid rewriting old rooms merely to make them agree with newer conclusions.