I think the general contention is that the process you describe is just using someone who's going to write sensible HTML not SEO. The shady side of actually modifying code not for easy-to-read, browser-efficiency, etc. is SEO.
I have to disagree. I consider part of my job to be some SEO (not my primary task and neither the most important, but I take care of it at my company) and its definitely the first part of what you sketch. And trying to clean all the second part stupid stuff someone else may have done before.
I don't see it as a game against Google... I try to make our sites as useful for humans as possible, and hope Google agrees with a decent enough ranking so humans can visit it and enjoy the content. But when it doesn't, then the game is between me and the "other SEOs" who may not see the point of making something useful for readers, not machines.
Writing sensible HTML is optimising for machines. The Google algorithm runs on machines, so they'll like it and benefit from it. Browsers, screen-reader, future colleagues will also like it and benefit from it. Calling this search engine optimisation seems a bit weird, though it'll certainly help Google process your website properly.
If you read the OP, his mean concern is the tiny subtle changes in the algorithm that drive him mad. He calls optimising for them SEO. It is considered shady and evil by some, myself included. (Ultimately it's about the content, not about who best reverse-engineered Google's algorithm.)
