> the algorithm for floating point addition is very complicated

Ah yes, of course. I think it is slightly misleading that the article is mostly about large integer multiplications, while that remark holds for floating point operations.