Disclaimer: This thread is a record of thoughts.
Discussion Given a compact set.
Do mere neighborhood covers admit finite subcovers? $$C\subseteq\bigcup_{i\in I}N_i\implies C\subseteq N_1\cup\ldots N_n$$ (The idea is that neighborhoods are in some sense fat.)
Application
Given a locally compact space.
Every compact set has a compact neighborhood base: $$C\subseteq U:\quad N\subseteq U\quad(C\subseteq N^°)$$ (The above would give clues how to prove this.)
Counterexample
Consider the compact set: $[0,1]\cup[2,3]$
Cover it by neighborhoods: $$(x-\varepsilon,x+\varepsilon)\cup\{x+2\}\quad(0\leq x\leq1)$$
Then it admits no finite subcover!
Discrepancy
Consider the compact set: $\{1\}\cup\{-1\}$
Cover it by neighborhoods for all points: $$(1-\varepsilon,1+\varepsilon)\cup(-1-\varepsilon,-1+\varepsilon)\quad(1-\varepsilon,1+\varepsilon)\cup\{-1\}$$
Then it admits a bad finite subcover!
Essence: The power of opens lied in them being neighborhoods of its point.