CONCEPT
Full House: Variance Over Mean
The distributional analysis revealing that apparent progress often reflects expanding variance from a constrained starting point rather than directional movement—the right tail extends while the mean barely moves.
Gould's
Full House (1996) framework reframes progress as a statistical artifact of distributional change. His paradigm case: the disappearance of .400 hitting in baseball is not evidence of declining skill but of compressing variance as the entire population improves. Mean batting average stayed constant around .260 for a century, but variance declined—highest averages fell, lowest rose, distribution compressed toward mean. Ted Williams's .406 in 1941 was not superior absolute performance but greater distance from his era's average. Modern players are more skilled overall, which is why outliers disappear. Applied to complexity in evolution, the same logic: life began simple (left wall—minimum organizational complexity), variance expanded in all directions, leftward expansion blocked by wall, distribution appeared to move right. But mean complexity barely moved—bacteria remain dominant. The right tail (multicellular organisms, consciousness) extended but the tail is not the trend. The trend is expansion, not direction. For AI, the framework predicts: democratization drops the left wall (who can build), variance expands leftward (novices gain capability), right tail