1.  I heard about a config optimization study, and I believe it was a grad student project at Stanford in the late 70s / early 80s, where the object was to maximize range factor with a fixed structural weight.  The student set up the code and ran it, starting with a canard configuration, and the code kept making the canard bigger and bigger, and the wing smaller and smaller, until a conventional config was arrived at.

2. The lift of the canard creates downwash on the wing, at the wing root where the chord (and area) are greatest.  At the same time, the canard's tip vortices cause upload on the outboard wing, which increases bending moments on the wing.  It also makes it hard to get anywhere near an elliptical lift distribution on the wing without resorting to twist, cuffs, etc.  So the canard's lift ain't for free.

3.  At the end of the Beech Starship debacle ($500-800M spent in development; 20-40 built; Beech buying them back and incinerating them), Beech spokespersons said that the .85 scale Rutan prototype proved completely worthless as a pathfinder to a full-scale machine.  I've put a link to an interesting article below.  It essentially proves what Fred says, that there's an awful lot in the details of the execution.  Beech somehow lost 90knots, 3,000lb, and $1M of unit cost in the translation from the prototype to the full-scale bird.