BlinkerSpawn wrote:Maybe instead of looking at variance between neighbor counts, look at variance between neighborhoods within neighbor counts?

For instance, a relatively orderly B2 rule would likely experience a lot more B2a or B2e than other B2 transitions.

Yes I tried to include orientation into account but it multiplies the number of variable by 45, and consider this is a O(n^2) complexity..... It's not clear how much symmetry we should impose on this scheme. The less symmetry in the prior, the harder the computation. So I guess we have to impose some symmetry to reduce the computation power required.

Bullet51 wrote:

It turns out that chaotic rules have weak tangling, and complex rules have moderate tangling.Maybe we are going off-topic...

If we just use the crude estimation, rules with glider usually has an average tangling around 0.25-0.20.

Yes it's quite off topic I guess we should switch to the "smoothness" post.If we just use the crude estimation,