Very interesting. I first wondered how nature can randomly generate such a pattern, and then realized we are just falling for our "built in" pattern recognition: it would feel much more "natural" for the stop sequence to be the encoding of some specific protein without any clearly recognizable pattern... But it would actually be more unlikely to appear/survive mutation than "any long-enough sequence of A".

I also like how it is established that this has an effect on replication, but that as far as I understand we do not understand the underlying process. Humbling.