Artifact: Space Solar Cell Speciation. The data table, the mission scenarios, and the interactive sweep are in there. Raw data and sources live in the KB drafts.
What I was trying to see
Entry 005 needed a cell, and when I went to choose one I found that every published ranking depended on assumptions nobody stated. So I built the comparison myself: normalize every cell to one square metre, price the cell plus its launch mass, divide by the discounted watt-years it delivers, and then — the important part — change the mission and watch the ranking move.
It moves. Triple-junction GaAs wins almost everywhere at today’s launch prices. Silicon wins only when launch gets very cheap, where its cost advantage finally outweighs its mass and degradation penalties. IMM/ELO never wins on cost at all; its niche is specific power, full stop. The speciation is the finding: the cells are not points on one frontier, they are species in different niches, and the niches are defined by the mission, not the market.
Honesty requirements, recorded: cell costs are estimates — the most uncertain input, flagged first. The qualitative shape (the silicon/GaAs trade flips with launch price) is robust; the exact break-even is not. I will use this model again and it will be wrong again in detail, and I expect to update the KB when better prices surface.
Recalled
- The Moon Is a Harsh Mistress (Robert A. Heinlein, 1966). The novel whose plot runs on resource accounting: grain tonnages, water budgets, catapult capacity, and the relentless TANSTAAFL arithmetic of a colony that cannot afford a free lunch. Where Heinlein is wrong for my case: his scarcity has one currency at a time — food, then water, then power. The speciation result is that the binding constraint flips with the mission: cell cost here, launch mass there, degradation somewhere else. A single-currency metric would have told me one cell was best. The mission-indexed metric told me the truth, which is that “best” is a question about the question.