The hoverbike makes this easy to see because it is brutally constrained. Mass, surface area and volume are expensive. Heat that cannot be rejected quickly enough is a fault condition. A gas produced by the energy source becomes a storage problem unless the machine can consume, vent or otherwise incorporate it.
Scale the same family of mechanisms into a ship, habitat or station and the judgement changes. Radiator area can grow. Thermal paths can be duplicated. Machinery can be physically separated. Heat that was dangerous on the bike may be useful for water, atmosphere, industrial processes or temperature control before the remaining low-grade fraction is finally radiated away.
Waste is not an intrinsic property of an output. It is a relationship between that output and the system surrounding it.
This does not make every by-product useful. It changes the question. Instead of asking whether a mechanism produces waste, ask where the system boundary has been drawn and whether another process on the other side of that boundary requires the same material, temperature, pressure or energy flow.
The reverse matters equally. A product that earns its place on a mobile machine can become an unwanted liability at station scale. The polonium-bearing CEU lineage is a useful example: a bike has strong incentives to make every output perform another job. A fixed installation can afford dedicated containment, separation and processing. Multifunction is not always elegance; sometimes it is simply what scarcity forced us to tolerate.
This suggests that descendants should not be designed by enlarging their ancestors. The better method is to remove the constraints that shaped the ancestor and see which adaptations remain necessary. A shipboard descendant of an early CEU might retain isotope containment and thermal conversion while discarding almost everything that made the motorcycle version recognisable.
That also gives company histories a plausible way to grow. A power company may own only one branch of such a lineage: thermal transport, sealed sources, conversion hardware, buffering, containment or control. Later products can resemble old experiments without requiring a single clean inventor or a single canonical machine.
The Station should therefore treat outputs as flows before assigning them moral labels such as useful, wasteful, clean or troublesome. Those labels belong to architectures. Change the architecture and the same flow may change sides.
Filed after discussion of CEU scaling, station thermal management, decay by-products and possible shipboard descendants of technologies first encountered under vehicle constraints.