The ancestry of a shipyard

The daughter

Before departure, the shipyard tests the machine that will make the next shipyard's machines. A cutting head moves along a polished guide. Light measures its path. The machine is heated on one side, cooled and measured again. After each cycle, the head returns to the same line within the tolerance its destination will require. The vessel is allowed to leave.

Its cargo contains a small power supply, mining robots and furnaces that fold into the transport frame. Several cases hold processors and sensors made elsewhere. There is enough reserve equipment to survive failures and start a daughter yard, but the manifest cannot hold every machine that daughter will eventually need. Most of the mass of the new industry must come from the body on which it settles.

The polished guide was made locally. So were the furnace that treated its metal and the structure carrying the cutting head. Their ancestry reaches back through successive workshops to a much smaller machine brought from Earth. Following that ancestry reveals how an industrial seed grows: each generation must produce something that makes the next generation physically possible.

The intelligence travelling with this cargo has a similar ancestry. It has learned how to turn uncertain minerals into dependable machines. At its destination, it will have to decide which parts of that knowledge still apply.

The imported ancestor

Years earlier, the parent expedition anchors to a metal-rich asteroid. The first robots work slowly, gripping the surface while they cut, because a heavy machine has little weight here. Deployed solar panels power the first workshop. Shelters intercept fragments and shade the equipment from abrupt changes in sunlight.

Its AI organisation begins by characterising what it has landed on. Samples vary in composition and in the inclusions carried through them. Refining consumes energy; moving and rejecting material consumes machine life. The yard needs a sequence in which the equipment it can build today supplies the materials and power for what it wants to build next.

One early expansion uses large reflectors to concentrate sunlight into heat engines. Much of a reflector can be made from local material with modest precision. A turbine is less forgiving. Its rotating parts must clear the stationary housing through startup and operation. A small permanent distortion can make them rub. The imported workshop can make demonstration parts, but production at the intended scale requires larger machine tools.

The first local tools cut their own successors' frames and guideways. This is where expansion stops. A newly finished guide passes inspection, then changes shape after the machine has run through several cycles of heating and cooling. Parts made on it inherit the error. The yard can produce tonnes of metal and cannot yet produce a dependable larger version of the equipment in its first cargo.

The mission's human directors postpone the first vessel. They assign its metal and workshop time to a second generation of experimental machines. The departure date moves away while the asteroid's small settlement of furnaces grows.

A surface that will not stay finished

Residual stress is a familiar manufacturing problem. The founding models know about annealing, temperature control and the distortions released when a tool removes metal. They have already allowed for them. What they cannot explain is why guides with similar bulk composition and apparently adequate heat treatment accumulate different permanent errors under the same operating cycle.

A metallurgical team suspects local differences hidden by the bulk measurements. Another group questions the measurements themselves: the workshop's inspection frame also expands as it warms. Recalibrating a good guide and remelting a bad one are very different responses. The yard needs to separate movement in the object from movement in the instrument before it can trust either explanation.

Researchers build a test assembly around a surviving imported reference. They reverse specimens within it and measure from independent directions, borrowing established methods of separating instrument error from part error. They also reproduce the heating pattern of an operating machine while applying a controlled load. The experiment is awkward and slow. It occupies equipment intended for production, and each specimen has to complete many cycles before its permanent movement becomes clear.

The inspection frame accounts for some of the apparent drift. The rest remains in the metal. Sections cut from failed specimens show that constituents segregated during casting produce regions with different responses to stress. During uneven heating, small regions yield while the surrounding metal stays elastic. A treatment sufficient for one casting leaves another able to move again. Average composition and final temperature have concealed differences that determine whether a surface stays where it was made.

For months, the institution follows the consequences across refining, casting and machine design. Its strongest reasoning has to remain available at the end of that chain, where a change in a furnace schedule may remove the need for a heavier machine frame, yet introduce a different failure during startup. Every answer creates work for another discipline. The organisation learns to evaluate those dependencies together; restarting each team from an abbreviated manufacturing history would repeatedly restore the very assumptions the experiments are dismantling.

The family acquires a method

The yard develops a manufacturing plan around the thermal and mechanical history a part will face. For critical guides, it changes the casting geometry and heat treatment, then cycles the unfinished component through a deliberately demanding load before making the final cut. Measurements during that conditioning tell the machine designers whether the part has settled sufficiently. Material that fails goes into structures with less demanding duties.

The yard can now determine which machines can be built from a particular batch of material, through which process, and for which operating conditions. Its researchers develop models that reason about that whole relation. A furnace trial now changes the set of machines the organisation knows how to make.

The industrial plan changes with them. A group formed to reconcile metallurgy and precision measurement becomes responsible for qualifying each new manufacturing route. It develops specialised models from the experiments, including the misleading measurements. New researchers join with the ability to interpret a specimen's response during conditioning and choose a further test. Their first assignment can begin beyond the point that cost the founders a year.

The resulting tools manufacture the turbine components. More power permits larger batches and better refining. The improved material supports larger, more stable tools. At each step, the yard has to decide whether to spend its new capacity on output or on improving the equipment that will produce later output. It can now explore those choices through a physical account of the dependencies, including the uncertainties that still need experiments.

Eventually, most of the departure vessel comes from the asteroid. Its processors still draw on imported stocks, and some instruments remain beyond local manufacture. Those limits determine how many descendants the yard can provision. The first daughter carries extra experimental equipment at the expense of payload: its destination's minerals are less well characterised, and the parent has learned what ignorance can cost.

An unfamiliar inheritance

The daughter settles near a different asteroid's pole. Long shadows change the power schedule. Its ore contains much less free metal, and the first refining trials leave impurities the parent's process seldom encountered. The familiar heat treatment produces specimens that appear stable during conditioning but crack when loaded at operating temperature.

The inherited organisation initially searches within its most successful explanation. It varies the conditioning cycle and casting geometry. That improves dimensional stability and leaves the cracking. A model specialised in the parent's materials keeps recommending more elaborate versions of the same treatment. Its acquired skill has become an expensive preference.

The daughter retains a separate ability to question that preference. It can examine a manufacturing route as a chain of assumptions, and it knows how to build a comparison that changes one disputed link. Researchers divide a batch, refine part of it more thoroughly and give both portions the same thermal history. The cleaner portion survives. Microscopy directs the investigation towards impurities concentrated at grain boundaries. The fault belongs upstream of the operation they had been trying to perfect.

The local organisation revises its refining process and the models that select treatments. It keeps the parent's measurement methods and the machines capable of applying controlled loads. Most of the inheritance remains useful because it includes the ability to investigate why a process works. The daughter has arrived with enough developed science to locate a new problem inside an unfamiliar industry.

Its report reaches the parent after the first corrected batch is in production. The parent applies the new impurity tests to reserves it had considered too difficult to use. An investigation at one destination has opened a material resource at another. The two yards begin preparing a third seed whose researchers will start with both bodies of experience.

The granddaughter's cargo

By the next departure, no founding model is left in charge of the whole undertaking. The lineage can now invent local industry, which none of its original components could do. A descendant receives that continuing organisation along with its equipment.

The third vessel carries a smaller version of the load-testing apparatus and a furnace adapted for a wider range of feedstock. There are fewer spare structural parts. The manifest instead makes room for the means to discover how unfamiliar parts should be made. In a cargo frame, a polished guide waits for the first temperature cycle at its destination.