Data on police: a great step to reform
I should be able to go to a website and type in a police department and find out what percentage of the officers of that department denounced fellow officers in a given year and what percentage of those denunciations resulted in disciplinary action. I should also have much more data: overall crime rate, cost of department per capita, comparison data to similar jurisdictions, and I should be able to make a non-anonymous complaint about that department involving my own personal experience that will be read by a federal employee.
If you have 100 officers and not one of them reported any wrongdoing by their colleagues-- not letting a friend or co-worker or relative off when they should have been arrested, not taking or planting drugs, not racially profiling suspects, nothing at all wrong-- that could be a great sign that your department is perfect. But if you released information that one or two of your employees was in fact guilty of some infraction and was terminated on the basis of a complaint by another officer, I would trust your department more.
If even after all the press coverage and a video dash cam in operation a officer thinks he can yank a person out of a car for no conceivable crime, arrest her, assault her, and leave her a condition that threatens her life over three days, then you have to assume there is no accountability whatsoever within the majority or maybe all the police departments in America. Data would be a good place to start on accountability. While racism and police abuse of authority are clearly related, the abuse of authority and clear rules and reporting may be easier to change than internal feelings and prejudices.