There's a difference between geometric constraints (GC) and dimensional constraints (DC), and each has its place. When I do a sketch, I will apply the GS first according to my "design intent". Then I'll apply the DC to control the size. I think this makes the sketches simpler, and easier to understand and modify.
Additionally, I never use the 'fixed' GC, and very seldom use the 'point on curve' GC. These should be avoided like the plague IMHO. I also find it's also easier to understand the GC when you don't rely on the 'inferred' constraints. In-fact, the only inferred GC I use is the 'coincident' GC, and manually apply the others as I see fit...
YMMV
Regards,
SS
CAD should pay for itself, shouldn't it?