Earlier quoted context omitted.
He claims to work for google as search "liason". He's a PR guy. His job is to make people think that google's search system is designed to improve the internet, instead of it being designed to improve google's accounting.
I actually work for our search quality team, and my job is to foster two-way communication between the search quality team and those outside Google. When issues come up outside Google, I try to explain what's happened to the best I can. I bring feedback into the search quality team and Google Search generally to help foster potential improvements we can make.
Your job is not to disseminate accurate information about how the algorithm works but rather to disseminate information that google has decided it wants people to know. Those are two extremely different things in this context.
I work on these kind of vague "algorithm" style products in my job, and I know that unless you are knee deep in it day to day, you have zero understanding of what it ACTUALLY does, what it ACTUALLY rewards, what it ACTUALLY punishes, which can be very different from what you were hoping it would reward and punish when you build and train it. Machine learning still does not have the kind of explanatory power to do any better than that.