Analytics have their place but it's folly to place too much emphasis on them. Analytics in hockey are in their infancy and new models are popping up every day. Just a few years ago Corsi was the stat to determine how good a player was and now it's more or less accepted as useless at worst or very limited at best. The men and women touting their models are often times asking fans for feedback one day and the next day using those same models to make declarative statements on how good a player or team is.
There is a reason that quant funds don't win all of their trades in the stock market and that's because algorithms, machine learning, and data still need humans to piece everything to make decisions. There is a human and behavioral element to sports that analytics will never be fully able to capture, just like there's a human irrationality component to the markets that quant-only funds can't fully capture.
And I'm not one of those "if fans knew the game they'd be in it" kind of guys but I do think there's an element to it when it comes to the analytics community. I haven't come across anyone in the analytics community with any type of playing background and while that certainly doesn't preclude them from having an opinion, or an informed opinion or the right opinion, I do think that dynamic lends itself to people compensating for a lack of understanding the game by relying too heavily on spreadsheets.
And since we're on the quote train, there's always this gem, "There are three kind of lies: lies, damned lies, and statistics." - (widely but inaccurately attributed to) Mark Twain