TL;DR: I want to find the minimum path from a node $X$ to a node $Y$ in a directed graph. But the edges aren't directly given: There are some rankings (ordered lists of nodes), and there is a directed edge from the node $A$ to node $B$ if and only if there is a ranking where $A$ is placed above $B$. Is there a algorithm for that, or a name of this kind of dense graphs, with "ranking-like" edges? Is this a known problem? Are there good heuristics?
Context of the problem + What I've been thinking: suppose that were was a game, skill-based, where a lot (around ~100k) of participants take place in each round, and at the end of each round there is a ranking with all participants sorted based on how well they performed on it. Not all players of this game take part on each round - Actually, just a small fraction of them.
Naturally, there are some players who are really good at it (as it isn't luck-based), so they always take the top positions. I was wondering how, given $A$ and $B$, I could find a chain $C_1, C_2, C_3\ ..., C_n$ of participants such that $A$ has performed better than $C_1$ in some round, $C_n$ has performed better than $B$ in some round, and $\forall\ 1 \le i < n$, $C_i$ has performed better than $C_{i+1}$ in some round. Additionally, I wanted that chain to be the smallest possible. But I can't just do a standard graph searching algorithm as a BFS on it, because the number of edges would be too huge - $O(K\cdot N^2)$, where $K$ is the number of rounds and $N$ is the number of participants in each round, because each ranking would lead to $O(N^2)$ edges.
I didn't had many progress towards it, mainly because I don't know the terminology to that stuff. One of the things I thought was a heuristic algorithm: I assign a "ELO" to each participant based on their mean performance. Then, I find a path from $X$ to a higher ELO, and from there, it should be easier to get to $Y$. But I'm interested in knowing better heuristics (not sure if deterministic algorithms would be fast enough). This is really, really close to six degrees of separation, but the main difference is the amount of the edges involved, I think.