Re: semantics for F_{sub,rec} ??

Thomas, you wrote: 

"if it turns out as built on sand (in the sense that it lacks a good
semantics for the target language used in these translational

Would you mind illustrating this "building on sand" portion of your
letter with, oh say, one or two practical examples? 
 - How would programmers suffer from a lack of a classical semantics?
 - How would programming language researchers suffer? 
 - What can go wrong? 

Thanks -- Matthias