Oh I see, yes that's true. The hash chosen (SHA-256) was picked because it is easy to perform, which aids in verification as well as allows manufacturers to more easily enter the market. Anyone can enter the mining business, since manufacturing ASICS is not as difficult as it would be with other algorithms.
It was hoped that this would lead to greater competition in the mining space to avoid 51% attacks, although I'm afraid that centralized mining is almost a foregone conclusion at this point.
I think hashing (SHA-256, Scrypt) is the most suitable type of workload, which satisfies one important condition - it must be hard to find the result, but very easy to check if the result is correct. That was probably number one reason why scientific computation didn't qualify for PoW as checking for correctness would require significant amount of re-computation on other nodes.
As to the question of mining centralization, it does seem like major players are getting bigger over time, however they are still spread out on the surface of the planet, so the problem might not be as acute as some prefer to paint it.
Who knows maybe Bitcoin is destined to become a new international settlement system and various governments controlling mining farms on their own land is Bitcoin's intended destination. It would still be an improvement over fiat monetary system we have now as long as governments don't collude, and today they are eager to demonstrate some competition (US/Europe/Russia/China). So I would be interested to see how PoW mining scenario plays out in a long term, while keeping an eye on a more lightweight and much more decentralized solutions for end-users.