Me too, and I just drafted a simulation according to (my understanding of) @robf 's ideas (about a first stage to go), described in another thread: The coding of longer sequences in HTM SDRs - #27 by complyue
I believe runnable code / simulation should clarify things faster, and running that with Gitpod is just a few clicks away with a browser. Please have a look, and let us continue the discussion and drive our ideas / understandings with simulations.