Shmuel
- read just the abstract/statement,
and try to fill in gaps yourself
(at least, this is what he does)
(this way you actually understand it)
try to make sense of results
- know lots of examples
know *representative* and *extreme* examples, rather than (just) the axioms
(Obviously of course, try basic examples)
Geometrically, a set of points determine their convex hull,
and often more simply than this set's defining equations.
This is *opposite* to Grothendieck's "deduce everything from axioms";
it can be a complementary approach
(following Drinfeld: know both),
or may be more or less suited to a given field.