An important role of top management in a large corporation is to make sure that projects are managed according to contractual priorities. One way project managers deviate from priorities is by pursuing features not expressed in the contract with the argument that they are trying to make the product better. This approach would be fine if there were infinite time and resources. Since projects have to operate within hard constraints, a manager has to first satisfy the minimum requirements, otherwise the project will be late and engineers will be confused and demoralized.
In my earlier years, I thought that when managers are reminded of priorities, they would quickly change their way of doing business. To my dismay, I saw that this is sometimes not the case, with catastrophic results. When you say to anyone "a project should follow priorities", you will get a "yes" response 100% of the time. But implementation can vary widely. After living in fantasy land, the second most common reason of project failures are not technical or budget deficiencies but wasting time on low priority tasks while many high priority ones were still waiting for attention.
Top management should from time to time have informal meetings with people one or two levels down the project manager and ask them whether they think the project in managed according to priorities.