A different selectivity can choose a different plan inside the same toy model.
estimates can change
Compare two estimates
One estimate chooses hash_join; the more selective estimate chooses nested_loop_join.
choice is model output
Optimizer-choice examples are named toy estimate-model outputs; they are not benchmark results, database-product claims, hardware claims, or real speed predictions.
Keep the honesty note
The shown choice is an estimate-model output, not proof of real speed.
not a benchmark
Summary
Real speed would need a benchmark outside this static lesson.