Let $S$ be the surface of a convex body, polyhedral or smooth, 
embedded in $\mathbb{R}^3$.
For a point $x \in S$, let $F(x)$ be the set of furthest points
from $x$, measured by shortest paths on the surface $S$.
Let $f(x)$ be the length of those shortest paths:
$|x y|$ for $y \in F(x)$.

It seems natural to hope that

> **Hypothesis**: For any $x \in S$, $f(x) \ge \tfrac{1}{2} \mathrm{diam}(S)$.

Here $\mathrm{diam}(S)$ is the maximum distance between any two points
on $S$ (again measured by shortest paths on the surface of $S$).
Suppose, for example, that $\rho$ is a diameter-realizing geodesic.
Then for any $x \in \rho$, $f(x) \ge \tfrac{1}{2} |\rho|$, just
tracking along $\rho$. 

A non-comprehensive literature search has
failed to uncover a relationship between $f(x)$ and $\mathrm{diam}(S)$.

<hr />
Itoh, Jin‐ichi and Costin Vǐlcu. "Criteria for farthest points on convex surfaces." *Mathematische Nachrichten* 282, no. 11 (2009): 1537-1547.
[Journal link](https://onlinelibrary.wiley.com/doi/abs/10.1002/mana.200610811).