A tech startup has developed a new Large Language Model designed to assist with creative writing tasks, such as generating story plots and character descriptions. To assess the model's usability, the development team proposes an automated evaluation method. Their plan is to measure the similarity between the model's generated text and a large dataset of classic novels, using a computational metric. They argue that a high similarity score will indicate high usability, as the model's output will be stylistically close to established great works. Critique this evaluation plan. In your response, identify at least two major flaws in this approach specifically concerning the assessment of usability, and propose a more effective, human-centered evaluation strategy.