criticality

Seems sure to become material for my favorite problems

"Misaligned agents will lose" by Ruben Laukkonen

Claude explains criticality to me:

systems near criticality maximize dynamic range—the range of stimulus intensities a network can discriminate. This is directly a computational advantage: a subcritical network can only detect strong signals, a supercritical network saturates and can’t distinguish anything, but a critical network responds discriminately across orders of magnitude.


Seems related:


Richard Gabriel writing about Christopher Alexander in Patterns of Software


the experience of a text is produced by the relationship between what is written and what is not -- its negative space

though the writer depends on their sense of it to write, they cannot control it as they do the text, as each reader provides their own sense of the negative space