A very short and easy question. Why do people write "locally constant constructible sheaf" (e.g. everywhere in SGA) instead of just "locally constant sheaf"? A constructible sheaf is by definition locally constant on a stratification, hence locally constant implies locally constant constructible. And locally constant constructible implies locally constant. So the two notions should be the same. Have I misread the definitions?

1$\begingroup$ Apparently "constructible" has a second meaning: mathoverflow.net/questions/69422/… $\endgroup$– Qiaochu YuanFeb 24 '14 at 7:40

1$\begingroup$ Indeed, if we impose finiteness in the definition of constructibility, then "locally constant constructible sheaf" is a bit shorter than "locally constant sheaf of finite groups". On the other hand, the Stacks Project seems to be using "finite locally constant", which sounds nicer. $\endgroup$– Piotr AchingerFeb 24 '14 at 7:45

$\begingroup$ If I recall it correctly constructible does require locally constant (with respect to a stratification) of finite rank (local system on each stratum)! $\endgroup$– Oliver StraserMar 26 '14 at 9:51
Locally constant means locally constant on the whole space, not on a stratification, so constructible doesn't imply locally constant (or locally constant constructible). Depending on which SGA you are talking about, constructible imposes some finiteness conditions, so locally constant doesn't imply constructible.

1$\begingroup$ But locally constant constructible does imply locally constant! $\endgroup$ Feb 24 '14 at 8:10