The author recounts their long-standing skepticism about ELIZA's significance in AI history. They first encountered ELIZA in the 1990s and found it unimpressive, believing it was only slightly better than basic if/then logic. While acknowledging some interesting natural language parsing, the author never thought ELIZA could pass the Turing Test. They were particularly doubtful of claims that people used ELIZA as a virtual therapist. Having recently re-tested ELIZA via a web recreation, the author provides a transcript of their interaction. The conversation highlights ELIZA's repetitive and nonsensical responses, particularly its grammatical errors. The author concludes that people who view current chatbots as friends are misguided, but those who saw ELIZA as a confidant were severely mistaken. They emphasize the profound disconnect between ELIZA's capabilities and the roles some users attributed to it.