Random variables partitioning a space into chunks over which they are constant
$\iff$
Capturing all information provided by the random variable.
Cf:
[[Statistical Inference#Minimality and Sufficiency]]
[[Conditional Expectations]]
Consequently, the $\sigma$-algebra created by such partition represents the **information** carried by the RV.
- In particular, the trivial $\sigma$-algebra $\{ \emptyset, \Omega \}$ provides no information: for any RV $X$ over $(\Omega, \mathcal{F}, \mathbb{P})$, $\mathbb{E}[X \,|\, \{ \emptyset, \Omega \}] = \mathbb{E}[X].$