This is a good write up, but for me seems to miss the mark. I agree with the author that different algorithms require differently organized data structures. But what if your problem requires 2 different algorithms, which each work best with a different data structure? Either you pick one and run both (which is not ideal) or you run the first algorithm with one structure, convert it, and the run the second algorithm in the second structure.

And once you can do that… why not have every algorithm ensure it runs in its best structure, and convert where necessary (and possible) on the way in? Yes, there’s absolutely a performance or storage cost… but if the algorithm is that much faster, it should be worth it.

Basically a beefier version of sorting your data before searching in it. If an algorithm works best with a specific model of a graph, then let that be part of the algorithm.