I am going through our video again - and I now realise there is one way in which I differ from Molly's teachings.
The question in the video was: "What if each small spot is 90% success rate, but when you put the whole piece together each time you fail in different places, so that the total success rate is 50%?"
Molly's answer is: "First of all, you need to get a 100% success rate for each spot, otherwise you'll never be solid in the whole entire thing."
So far, I am totally with her.
But to fix the problem, she recommends lots of random practice, particularly "performance testing using random practice methods".
And for me, that doesn't work.
If I can play the spots correctly, but I make mistakes in playing through, one of the main causes is that I don't have the piece connected well enough in my head. It is as if I every time need to get into the right track, and sooner or later I miss a track. And yes, I do vary my sections. They can be shorter or longer, and the transitions can be at the beginning, middle or end of a section.
For me, instead of more random section practice, I need to play through - particularly "performance testing - play through". Then I make a notice of the mistakes that I make, or, the mistakes that I almost made, fix them, and once again, a "performance testing - play through".
I do this "performance testing - play through" also in the early stages, when I still play a piece at a very slow tempo. In this way, I learn what comes next on a much more profound level than when I do random sections.