neural architecture searches must be really hard
i keep finding papers where people say they uncover repeated, highly non-random circuit motifs in brain connectivity data
but the circuits they find are like, just three neurons connected to each other in simple ways
i can't tell whether this is just incredibly hard, or whether people were just trying to publish something
because if they changed their assumptions for what's considered 'random' even a little bit, i doubt a lot of these circuits would be statistically interesting
vespertine ash