The article by Yifeng Liu addresses the complexities of using data model classes in software development, particularly in service-to-service communication. It emphasizes that these classes should only contain basic functionality, specifically getters and setters, without embedding any business logic. A key point is the dilemma between coding these models directly in the codebase or relying on a shared data model from the contract owner. Liu warns against the latter, labeling data model dependency as a significant pitfall. When teams depend on shared data models, they face one of three possible scenarios. The ideal situation is that teams consistently update the dependency as the contract changes. However, this rarely occurs. More often, teams either fail to update the models, leading to inconsistencies, or they end up duplicating the model in their own codebases, which can lead to maintenance challenges. The article makes a strong case for reconsidering how dependencies are managed, suggesting that duplication, in certain contexts, may not always be a negative practice but rather a necessary strategy to ensure agility and responsiveness in software development.