This is a fascinating and complex topic. You make some interesting points. Off the top of my head, let me respond to your challenge that fraud doesn’t necessarily destroy wealth.
Every time someone is a victim of fraud he is marginally less likely to participate in the market in the future. Fewer market transactions, less wealth. In a free market it is a transient effect. The first time someone mixes sand with his wheat to get a dishonest profit the market is disrupted. If every one starts doing it, the market would adjust, prices would drop and things would get back to normal (except the end user would be stuck with the expense of separating sand from the wheat). Fraud is worst in a regulated market. Say the price of a basket of wheat is fixed at 10 moneys. Now when people start mixing sand with wheat everyone gets poorer and stays poorer.
This well may be, but (i) one has to differentiate the existing wealth and wealth creation, and (ii) the model is incomplete: one has to model explicitly choices other than abstention from the fraudulent market (the victim typically goes to some other market when suffering a setback in a given one --- witness the real estate bubble after the burst of the DotCom bubble --- in which case even creation of new wealth is unchanged).
I agree with you when you say that this point fascinates many. My personal take is that it is because of the "common sense" perpetrated by media and uneducated folks: fraud is equated with robbery, hence loss. People do not understand, of course, that the loss is NOT macroeconomic: there is a countervailing gain to the robber. But, if a robbed one feels a loss, the folks reason, there must be a loss.