The public record is six registered refractory alloy compositions — AV-001 through AV-006. All are computational. None have been synthesized. The scoreboard is the prediction ledger.
Stage Zero is six falsifiable composition predictions and a retrieval index over published literature. It is not a spectroscopy-to-fabrication pipeline, not a printed part, and not a validated alloy.
AV-001, AV-002, and AV-003 were descriptor-screened and founder-selected. AV-004, AV-005, and AV-006 were model-proposed. A variational autoencoder is the stated surrogate for the model-proposed set only. It did not produce all six.
These are refractory alloy compositions. They are not “six high-entropy alloys.” AV-004 (Mo5Nb30Ta20W45) has configurational entropy below the conventional high-entropy threshold; it is a medium-entropy / complex concentrated alloy.
The ledger holds six. AV-001–003 are shown here as the descriptor-screened set; AV-004–006 complete it as the model-proposed set. Falsifiers and status live on /predictions.
Medium-entropy / complex concentrated alloy — not a high-entropy alloy under the conventional entropy threshold.
Full prediction ledger — falsifiers, status, AWAITING SYNTHESIS →
The live scoreboard is /predictions. Entries below keep the existing homepage anchors used by that ledger. Content is corrected to match it.
AV-004 (Mo5Nb30Ta20W45), AV-005 (Mo15Nb25Ta25Ti10W25), and AV-006 (Hf20Ta10Ti20W30Zr20) are the model-proposed set. A variational autoencoder is the stated surrogate for this set only. AV-004 is a medium-entropy / complex concentrated alloy, not a high-entropy alloy under the conventional threshold. Together with AV-001–003, the ledger holds six. None have been synthesized. Status and falsifiers: /predictions.
AV-001, AV-002, and AV-003 were descriptor-screened and founder-selected. They were not produced by the VAE. They are three of six registered compositions, not the complete set. All six remain [COMP]. Refutations will be published with the same prominence as confirmations. The scoreboard is /predictions.
A longer program discusses onboard spectroscopy in and a fabrication spec out; edge-GPU inference at the point of capture; a laser-sinter spec; treating orbital debris as inventory, including a roughly $2,500/kg launch-cost figure; the line “the best part is already in orbit”; and an Orbital Alloy Foundry Swarm (OAF-Swarm) concept. None of that has been demonstrated. It is not the Stage Zero result. It is not the mapping a preprint reader should take as shown.
Stage Zero is the six computational predictions, the retrieval index, and the furnace campaign that has not run.
Research — the six registered compositions, their labels, or the planned arc-melt campaign.
Programs — we will say what is computational and what is not. Nothing here is validated hardware.
Corpus access — the literature retrieval index at /corpus.