A Theory of Mechanism

Theory of Novelty / Theory of Ideas / Theory of Motifs

In[]:=
ParallelTable[ResourceFunction["AdaptiveCellularAutomaton"][<|"InitialRule"->{0,4,1},"MaxSteps"->1000,"AdaptiveIterations"->4000|>,"FinalState","Plot",RandomSeeding->32424+i],{i,100}]
Out[]=

,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,


What is the novelty metric?

Songs have “new ideas” when 5-note patches are different ...
[ Suno ] [ https://soundpatrol.com/ ]

What is the minimal program|computation that builds a piece of the pattern?

What are the patches in which one can jump ahead?
Cf. snakes & ladders

What is the semantic space of the patterns?

These words get their “embedding-level meanings” through the contexts in which they occur

Core ruleset?

Inside a pocket of reducibility, there are probably fewer rules used .... Or at least the rules are used in the same relative configurations [“rule operon” | “ruleon” ] (e.g. R110 background)
E.g. if there’s a patch of fixed color, this is obviously true. Same with stripes
Rule density mask ... how many distinct rules are being used in this neighborhood
Does there exist a permutation of the rule cases such that the interesting reducible pieces of this are visibly blobs?
For each point, there is a growth rate of the number of rules one sees at distance r
How small a rule works in the local neighborhood?

A map of the diversity of rule use in a region....

In addition to counting the rule cases that are locally used .... one could ask how the set of rule cases used changes as one moves across the pattern