Seems like a pedantic article, but I liked his remark that the 'effectively constant' argument can of course be extended to any algorithm where the input size i

Seems like a pedantic article, but I liked his remark that the 'effectively constant' argument can of course be extended to any algorithm where the input size is bounded and therefore, shouldn't really be used to describe your algorithmic complexities, even in real life implementations (except when you maybe use the inverse Ackermann function or something).