The main difference of course is that murderers occasionally go to prison for killing people. Whereas the percentage of police ever actually convicted for committing a crime while on duty is so low they might as well have immunity.
I would take these figures with a grain of salt because I doubt any accurate statistics of police misconduct are kept.
Of course there is a lot more non-police crime than there are reported incidents of police misconduct. What about simple mistakes? I'm not sure if buffoonery is counted as misconduct. What I'm really referring to though, is what is my own personal exposure to risk of harm? I wouldn't be surprised to find that it is more likely that police might kick my door down and shoot my dogs, either by mistake, misunderstanding, or misuse of power, than say armed burglars.
I think sites like HN and reddit showcase police misconduct and this leads to an exaggerated view of how prevalent it is compared to other dangers. It's similar to the MSM and sexual abuse/stranger danger.
[0]http://www.policemisconduct.net/2010-npmsrp-police-misconduc...
[1]http://www.fbi.gov/stats-services/crimestats