I've been reading your posts and have the repeated impression that with one exception you're mixing work and sport! BTW thanks Ina for the great posts, you've verbalised my own reflections.
I only wanted to draw you attention to the fact that most csv can easily recognize which tasks are authentic and which are not. This may have important implications for the training methods and explain some of the difficulties.
We used to have a truly working csv (without a pedigree and looking more like the first csv) that just by accompanying my husband in his field work trained herself to act geological resarch dog. Gradually, human teams relied more and more on her senses: she clearly signalled if a fresh wall in a quarry was approachable or not, she was able to 'gather' the team back after they worked individually in a difficult area (no mobile phones in the 90-ties!), when given the order she could easily find rocks croppin out in very dense forrests, etc, etc.
At home lazy sofa dog, in the field was turning into another animal: quick, exteremely rational and reliable, brave, fit and smart. Still, she was very independent and logical even if this was not human logic - this way she saved at least three people if not from death than surly from serious injuries. When completing a mission she didn't run after wild animals; deers and rabbits could walk next to her nose when she was trailing. What's more she loved that kind of work and when 'retired' at the age of 11-12, we had to give her tasks, otherwise she was getting depressed.
BUT, whenever we wanted her to show off in front of our friends how smart she was - she'd often pretnd a dummy and thus taught us modesty!
Lorka is not even two-years old, so I wouldn't dare to speculate what kind of dog she'd be when she fully matures, but she too, is able to recognise authentic, genuine sitautions from the pretended "mock" ones.
I suppose this is connected with am average very high level of empathy - we simply cannot fool our dogs!