The only thing that should be included in a report are facts.
It's nice to believe that facts are readily available to investigators, but I get the impression from reading John's posts that investigators rely on interviewing witnesses in order to figure out what the "facts" are. Sometimes those witnesses (from what John describes) have an incentive to be dishonest... which can make the determination of "facts" a little complicated.
It seems John's point is that investigators have to occasionally inject their expertise to sort through the witness testimony, try to determine which "facts" are reasonable and which aren't. If two witnesses give conflicting testimony, the investigator will inevitably have to go with one of them... and by default suggest the other was lying.
This is how I interpreted what John described.