James Q. Whitman has an interesting new book about the development of the concept of reasonable doubt in the Anglo-American legal system. An odd feature of that system is its quite explicit refusal to produce anything like a formal definition of this supposedly crucial concept. How much doubt equals reasonable doubt? Most jurisdictions in the US forbid judges to give probabilistic instructions to juries when explaining the concept, so they’re limited to very general platitudes about “moral certainty” regarding the defendant’s guilt and so forth.
Needless to say, what constitutes “moral certainty” ends up varying wildly, depending on all sorts of circumstances, some of which have nothing to do with probability and everything to do with things like race, socio-economic status, etc.
The basic problem is that most criminal trials end up adjudicating issues in a gray zone between factual guilt (the defendant actually committed the crime) and legal guilt (it’s possible to assert with some unspecified but extremely high degree of confidence that the defendant committed the crime). A good criminal lawyer walks a tightrope between arguing his or her client is factually innocent and reminding the jury they’re not supposed to convict unless guilt is established “beyond a reasonable doubt.” (For fairly obvious psychological reasons criminal defense lawyers almost never base their arguments on the explicit distinction between factual and legal guilt).
I’ve never been on a criminal jury, but I think I can imagine the enormous pressure people must feel when they’re fairly certain of a defendant’s guilt but still harbor some “reasonable” doubts. Our system deals with that pressure largely through avoidance and denial, by not defining reasonableness in this context, and rarely enquiring into what actually happens inside the black box of the jury room.