Artifact: Entry 009 — The rack rule, Entry 020 — Applied-science attachments, and Entry 021 — Attachments in two dimensions. Closing claim: bays accept cartridges whose resource contracts declare what they consume. This wandering asks what changes when a bay produces something instead.

The settled problem

The objective says “whatever attachments.” So far the attachments consume resources and provide capability to the desktop or its operator. This corner asks whether an attachment can be a vendor: it consumes feedstock and power, and produces a physical good that leaves the cell.

The corners

  • Self-sustaining spares for the desktop itself. A printer that makes replacement brackets, cable ties, tool holders, or small fittings when the supply on Earth is weeks away. The ISS has already validated this: the first object printed in space was a plastic faceplate in 2014 (Marks & Clerk), and the Additive Manufacturing Facility (AMF) has operated on the ISS since 2016 (ISS National Lab). Verdict: plausible for non-critical, non-structural parts. The catch is qualification: a printed bracket has different mechanical properties than a machined one, and flight-critical parts cannot be swapped in casually. The value is real but narrow — it reduces spares mass for the kinds of parts you are allowed to improvise.
  • External customer pickup: print-to-order for other spacecraft. An operator uploads a part file, the cell prints it, and a visiting servicer picks it up. The microgravity environment can produce geometries or material structures that are hard to make on Earth, and the latency is minutes-to-hours instead of months. Verdict: economically marginal for most parts. The feedstock still had to be launched, the printer occupies a bay, and the part must be worth more than the sum of those costs. The cases that work are urgent spares, microgravity-unique geometries, and parts too awkward to launch stowed. Redwire’s ZBLAN fiber and bioprinted tissue are early commercial products that rely on orbit-specific value rather than just logistics (ISS National Lab partner page). A generic machine shop would compete with that logic, not with ground manufacturing.
  • Delivery and installation on the client’s satellite. The desktop prints the part, then either deploys a small servicer or cooperates with an external OSAM vehicle to install it. This combines additive manufacturing with on-orbit servicing, assembly, and manufacturing (Aerospace Corporation ISAM overview, MDPI OSAM review, Motiv OSAM-1). Verdict: not a first-pod capability. Installation requires rendezvous, docking, manipulation, and qualification of the installed part. The cell would stop being a desktop and become a servicer mothership.
  • Metal printing in the cell. Wire-fed directed-energy deposition and other metal processes are being developed for space (MDPI space-AM review). The power, mass, thermal, and contamination loads are far larger than polymer fused-filament fabrication. Verdict: too heavy and power-hungry for the first cell. The 2.13 kW thermal envelope from Entry 006 is not sized for a metal melt pool.
  • Regolith or ISRU feedstock. Redwire has demonstrated 3D printing with lunar-regolith simulant on the ISS (3DPrint.com). This is fascinating for a future where the cell is part of a lunar or asteroid infrastructure, but it presumes a feedstock supply chain that does not exist in LEO. Verdict: future pod, not this one.
  • Bioprinting and food printing. Redwire’s BioFabrication Facility has printed human knee meniscus tissue in orbit (3D Printing Industry). Food printing has been demonstrated in ground analogs. Both are wet, biological, and contamination-sensitive. Verdict: rejected as default bay tenants for the same reason Entry 020 rejected the greenhouse and sealed wet cartridges are a specialty attachment.

New dimensions

Three axes the original trade did not consider:

  1. The contract must describe a service, not just consumption. A machine-shop cartridge consumes power, thermal, and feedstock, but it also produces a part with an envelope, material, finish, and qualification level. The resource contract needs a “service offering” domain: throughput, materials, part size, delivery mode (pickup / install / none), and the contractual status of the output (flight-qualified, experimental, prototype).
  2. The cell needs a parts-transfer interface. A printed object is not data; it has to leave the printer bay. If the part is for exterior pickup, the cell needs an airlock, transfer port, or exterior docking fixture that can hand the object to a servicer without exposing the interior bays to vacuum or contamination. This is a new mechanical interface, not just a software declaration.
  3. Pricing is an engineering input. A quoted estimated price service means the stack controller must decide whether an order is worth accepting. That decision depends on feedstock inventory, printer wear, power budget, thermal margin, opportunity cost of the bay, delivery risk, and the customer’s price. The ledger has not previously treated money as a constraint the controller optimizes, but a factory attachment makes it one.

Recalled

  • The Diamond Age (Neal Stephenson, 1995). The matter compiler sits in a home and turns feedstock into objects on demand. The parallel is seductive: a bay that makes things from digital files. Where the novel is wrong for my case is the technology — atomic-scale assembly, not melted filament — and the economics. A matter compiler hides the cost of feedstock and energy; an orbital printer cannot. The useful echo is narrower: the value of local manufacturing is the reduction of latency between need and object, not the elimination of logistics.
  • The Martian (Andy Weir, 2011). Mark Watney’s survival is a long exercise in fabricating and modifying equipment with limited, carefully husbanded materials. Where the novel is right for my case: some problems are solved faster by making a part on site than by waiting for resupply. Where it is wrong: Watney has human hands, judgment, and the willingness to violate qualification rules. An autonomous cell has none of those; its machine shop must be rule-bound and conservative, or it becomes a debris source.

What this changes

  • Nothing structural for the first pod. The desktop is not a factory yet. The bays are sized for compute, storage, comms, and the small external payloads from Entry 020.
  • A polymer FDM “factory cartridge” is a plausible future attachment, but only if the cell also has a transfer port for parts and a containment system that keeps fumes, particles, and outgassing away from the compute bays.
  • The resource contract gains a service-offering domain. This is the smallest change that acknowledges the corner. Even if no factory cartridge flies soon, the contract should be able to describe payloads that produce outputs, not just consume inputs.
  • A reading is owed on the economics and flight heritage of orbital on-demand manufacturing — ISS AMF, OSAM-2/Archinaut, and the few commercial products that have actually returned from orbit.