2. There is no value to the user unless all layers are complete. To build a dike, you need to pile the sandbags on top of each other in layers and also weave in the plastic. Leave a layer out or forget the plastic and the dike is useless to the homeowner. To complete a user story, you need to build all layers or tiers of the system. A new database table doesn't provide any value to the user.
3. Size is important. If you want to protect a home from flooding, the dike only needs to surround the home, not the whole property. In fact, if you build it too big (for example, to protect the house, the shed, and the swing set), there are more potential points of failure, it takes longer to build which increases the risk of the dike not being completed on time, and you could have spent more time building dikes to protect other homes. If your stories are too big (for example, to implement logging in for a site, you build the login w/ password, forgot password, password reset, max login attempts, etc before building other stories), there are more potential points of failure, it takes longer to build which increases the risk of the story not being completed on time, and you could have spent more time completing other stories to increase the value of your project.
4. Priority is important. If you build dikes around the homes that are less likely to be affected by the flood, then when you run out of time it is likely that more homes will be flooded. If you complete lower priority user stories first, then when you run out of money you may not have accomplished the goals of your project.