Suppose the ETI couldn't keep everyone — that preservation had a cost, and someone had to choose who's real and who becomes an NPC. Pick the criterion. Watch who disappears. Then notice who it spares.
✦ Speculative · the selection problemEvery criterion you can name does two things at once: it quietly deletes huge swaths of humanity — usually the poor, the past, the undocumented, the inconvenient — and it just so happens to keep the kind of person doing the choosing. The canon never tells us the ratio of preserved minds to NPCs (is it 1%? 10%? 99%?), and that silence is the point: once preservation is rationed, there is no defensible rule for who counts as real. The only criterion that indicts no one is "everyone."
The Cognitive Husbandry "Selection Problem" asks who the ETI actually preserves as a conscious mind versus who is rendered as an NPC. The unsettling result is that every proposed criterion fails the same way. Require literacy and you delete roughly 15% of living adults — and essentially all of human history before ~200 years ago (Homer was illiterate). Require formal education and you erase most people who ever lived. Require no violent record and you hand the decision to incomplete, unjust, culturally-relative archives. And each line is, suspiciously, drawn just wide enough to include its author.
The tool is a thought experiment, not a canon ratio — the percentages are illustrative of scale, not exact demographics. Pairs with the Kinetic Probe (which branch the ETI chose) and the Feasibility Calculator (note: on energy, the ETI never actually needed to ration — which makes any selection a choice, not a necessity).
Further reading: Pronin, E., Lin, D.Y., & Ross, L. (2002). "The Bias Blind Spot: Perceptions of Bias in Self Versus Others." Personality and Social Psychology Bulletin, 28(3), 369–381 — on the near-universal tendency to identify bias in others while exempting oneself; directly relevant to why each proposed selection criterion "just happens" to include its author. Bostrom, N. (2014). Superintelligence: Paths, Dangers, Strategies. Oxford University Press — includes discussion of selection criteria under an optimization agent that is indifferent to human moral categories, mapping to the ETI's non-human evaluation logic.