Show HN: How are Markov chains so different from tiny LLMs?