"Normal" phishing is a bad example, because many phishing emails notoriously use misspellings and bad grammar on purpose because scammers don't want to waste their time on people with half a brain, they only want people dumb enough to respond in spite of the ludicrous misspellings. But in this Verizon case, that logic doesn't apply, because there was only a single targeted recipient.
With respect to "If your defense against phishing relies on your adversary not knowing what a legitimate request looks like - it's not a very good defense", I wholeheartedly agree. My point was only that the fact that Verizon responded to this fake subpoena despite the reddest of red flags makes me think that they must have horrible procedures generally for verifying these types of requests.
Many phishing attempts are shockingly bad like this - but that doesn't make it difficult to not have mispellings and grammar mistakes and come from a vaguely plausible domain (gmail?). If your defense against phishing relies on your adversary not knowing what a legitimate request looks like - it's not a very good defense.