* I think it "should" use that index based on trying to follow that exercise.
* The part about changing the collation was an idea in the course of trying out different things.
* enable_seqscan is off, and the sharedmem and temp_buffers are set so high that most things happen in RAM.
I wonder what it that the other gentleman, Merlin, found out in the documentation and if he would share that.
I've also tried this on another table I have, with and without other indexes, but no success :-(
Wondering ...